NFL Question 1
Bobby D. Bryant
application of No Free Lunch theorems (NFLs). Some of the posts, like
this one, may be of general interst to t.o.ers, whereas others may be a
bit too esoteric. I'm going to post the esoteric ones anyway, since
there appears to be a couple of regulars here with an advanced
understanding of the subject matter. I'm not going to flag the esoteric
ones as [OT], since we don't really know what's [OT] until we see what
Dembski says next. However, I will put the individual questions (or
groups of closely related questions) in separate theads, so you can
easily ignore the discussions of those that don't interest you.
Here's the first -
I have argued elsewhere that NFL theorems don't have anything
interesting to say about biology, and I haven't seen anything to make me
reconsider that claim.
However, experience tells us that Dembski is going to keep invoking NFLs
even after people show that his claims are out in the weeds. So for the
sake of argument, assume that NFL theorems actually did have something
interesting to say about biology, and ask this question:
Is 'design' an algorithm?
My intuition is "yes", with the possible exception of a hypothetical
oracular design mechanism that can produce a design instantly, without
"thinking about it".
If my intuition is correct, doesn't Dembski shoot himself in the foot?
His only escape that I can see is to claim that any designer must be of
the supernatural variety (to provide immunity from his interpretation of
NFL theorems), but such a stance would conflict with the pretence that
the ID movement is secular science.
As for "oracular design", if a purported oracle actually produced
workable designs then it would seem to fall among the supernatural
variety of designers, per above. OTOH, if it just spat out a random
"design" without regard for whether it actually solved the problem, then
the oracle would appear to be a "blind search" algorithm with a single
iteration, and fall under Dembski's NFL claims anyway.
So: is 'design' an algorithm?
Bobby Bryant
Austin, Texas
Ivar Ylvisaker
>Is 'design' an algorithm?
>
>My intuition is "yes", with the possible exception of a hypothetical
>oracular design mechanism that can produce a design instantly, without
>"thinking about it".
>
>If my intuition is correct, doesn't Dembski shoot himself in the foot?
>His only escape that I can see is to claim that any designer must be of
>the supernatural variety (to provide immunity from his interpretation of
>NFL theorems), but such a stance would conflict with the pretence that
>the ID movement is secular science.
>
>As for "oracular design", if a purported oracle actually produced
>workable designs then it would seem to fall among the supernatural
>variety of designers, per above. OTOH, if it just spat out a random
>"design" without regard for whether it actually solved the problem, then
>the oracle would appear to be a "blind search" algorithm with a single
>iteration, and fall under Dembski's NFL claims anyway.
>
>So: is 'design' an algorithm?
>
An algorithm is a recipe that one obeys mindlessly and mechanically.
I doubt that many human designers will agree that this accurately
describes their professional activities. Of course, we might all be
machines. But no one has proven this yet.
Is scientific progress generated by an algorithm?
Are new algorithms generated solely from older algorithms? If yes,
where did the first algorithm come from?
Ivar
Michael Altarriba
Ivar Ylvisaker wrote:
> Bobby D. Bryant wrote:
>
>> Is 'design' an algorithm?
>>
>> My intuition is "yes", with the possible exception of a hypothetical
>> oracular design mechanism that can produce a design instantly, without
>> "thinking about it".
>>
>> If my intuition is correct, doesn't Dembski shoot himself in the foot?
>> His only escape that I can see is to claim that any designer must be of
>> the supernatural variety (to provide immunity from his interpretation of
>> NFL theorems), but such a stance would conflict with the pretence that
>> the ID movement is secular science.
>>
>> As for "oracular design", if a purported oracle actually produced
>> workable designs then it would seem to fall among the supernatural
>> variety of designers, per above. OTOH, if it just spat out a random
>> "design" without regard for whether it actually solved the problem, then
>> the oracle would appear to be a "blind search" algorithm with a single
>> iteration, and fall under Dembski's NFL claims anyway.
>>
>> So: is 'design' an algorithm?
>>
>
> An algorithm is a recipe that one obeys mindlessly and mechanically. I
> doubt that many human designers will agree that this accurately
> describes their professional activities.
By the same token, a neuron is a cell that acts mindlessly and
mechanically, and yet, we think. I'm sure a designer would not agree
that they design via mindless, mechanical means. It does not follow that
an algorithm can't design, as we have seen several examples of GAs
strarting out with randoom code and ending up with actual algorithms.
For the record, I don't know whether or not human thought is
algorithmic. How would we design an experiment to determine this?
>Of course, we might all be machines. But no one has proven this yet.
We aren't machines made out of atoms? What makes you think so? Perhaps
you should expand on what you mean by "machine" in this context.
>
> Is scientific progress generated by an algorithm?
>
I"ve always thought of the scientific method as a kind of algorithm.
> Are new algorithms generated solely from older algorithms? If yes,
> where did the first algorithm come from?
>
GAs can generate algorihms, so no, new algorithms do not come solely
from old ones.
> Ivar
>
>
Jonathan Stone
Bobby D. Bryant <bdbr...@mail.utexas.edu> wrote:
>Is 'design' an algorithm?
Without taking a stance either way:
If you claim `design' is an algorithm, can you write down the
algorithm? How is `design' any different in this respect than writing
poetry, painting pictures, or sculpting scupltures, or writing music,
or for that matter, doing science?
If you will, for the sake of discussion, grant us the Church-Turing
thesis, what you are asking is whether any of these activities can be
produced or simulated by computers. The scientific answer to that
seems to be "yes", in certain microworlds (aka "toy domains") and "no"
in the general, everyday human-level sense of the world.
Note, though, that I'm not using "toy domain" as a pejorative,
not at all; here it means, more-or-less,
"whatever computers can do easily enough that
we can put it in a first or second course on AI";
and thus it shifts over time.
>My intuition is "yes", with the possible exception of a hypothetical
>oracular design mechanism that can produce a design instantly, without
>"thinking about it".
which I think tells us much more about your intuition than
about the problem.
>So: is 'design' an algorithm?
To which the only sensible answer is: What do you mean by "is an algorithm?"
If it _is_, nobody knows how to state the algorithm (we even have a
hard time teaching it to some humans.) Are you going all Platonic on
us, or what?
[....]
Matt Silberstein
[snip]
> If you claim `design' is an algorithm, can you write down the
> algorithm?
So what if he can't (currently) produce this algorith? So what if it
will forever remain beyond any human's ability to produce this
algorithm? I don't see how that says there is not some algorithmic
process behind design. His abilities and knowledge do not determine
what is or is not an algorithm.
> How is `design' any different in this respect than writing
> poetry, painting pictures, or sculpting scupltures, or writing music,
> or for that matter, doing science?
A good question, but a good question does not an argument make. I
suspect that for the purpose of the NFL and this discussion they are
all sufficiently similar.
> If you will, for the sake of discussion, grant us the Church-Turing
> thesis, what you are asking is whether any of these activities can be
> produced or simulated by computers.
Not quite. It asks whether they can be produced/similated by a
universal turing machine. No actual computer is an actual universal
turing machine. They all have finite limitations that the UTM lacks.
[snip]
Jonathan Stone
Matt Silberstein <mat...@ix.netcom.com> wrote:
>jona...@DSG.Stanford.EDU (Jonathan Stone) wrote in message news:<a1b3e5$7qe$1...@Pescadero.DSG.Stanford.EDU>...
>will forever remain beyond any human's ability to produce this
>algorithm? I don't see how that says there is not some algorithmic
>process behind design. His abilities and knowledge do not determine
>what is or is not an algorithm.
Then it is a question with no import, a question that boots nothing.
[[ See several references to "going Platonic" in another thread . ]]
>> If you will, for the sake of discussion, grant us the Church-Turing
>> thesis, what you are asking is whether any of these activities can be
>> produced or simulated by computers.
>
>Not quite. It asks whether they can be produced/similated by a
>universal turing machine. No actual computer is an actual universal
>turing machine. They all have finite limitations that the UTM lacks.
Oh dear, and I thought this discussion was amongst informed adults,
where everyone took that as that as read.
Matt, is this a rather roundabout way to say everyone in the thread
must expicitly state **every** little detail, in case poseurs like
youself participate? Or are you trying [to coin a phrase] trying to
teach your grandparent to suck eggs, by parroting something you just
learnt here last month?
Or -- like Fred Colon commenting on seaweed inside Leonard's Boat - do
you genuinely think you just Made a Contribution?
John Wilkins
Jon
simply becuase *you* know or relaise something obvious is not to
guarantee that *everybody* does. Matt's point (which is also mine in the
"info from nowhere" thread about Lisp) is that at best we can physically
approximate a UTM or anything formal. But this point is not often
appreciated. For example, I once gave a talk to philosophers (ghod help
me) on why species were not classes because they were not intensionally
defined, and for chrissakes one of them said to me "these are such
*little* classes you are talking about". It took me a while to realise
that he meant that he was used to dealing with what Lewis called Big
Sets - Cantorian transfinite thingies. Now that may be true in terms of
logic, but this is biology we are talking about - there are very few
infinite objects in biology. I would make the same point here - when we
are dealing with realworld cases, such formalisations as the NFL simply
do not apply. UTMs do not exist. Etc.
You might find this unproblematic, or you may not. I get the impression
that you think the lambda calculus actually exists somewhere. I do not.
Lisp is not the lambda calculus, and any installation of Lisp on a
computer is at best an approximation (perhaps a very good one) of it. If
you had a theorem proven that was the result of a Lisp run that went
through three brownouts, two CPU failures and a hard disk crash, would
you trust that it had delivered something radically true from the lambda
calculus? I wouldn't.
But then, poseurs like Matt and I are not worth discussing these things
with, are we?
--
John Wilkins
Occasionally making sense for over 46 years
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
>simply becuase *you* know or relaise something obvious is not to
>guarantee that *everybody* does. Matt's point (which is also mine in the
>"info from nowhere" thread about Lisp) is that at best we can physically
>approximate a UTM or anything formal.
John,
I really meant it: where I'm coming from, this is so well-known it
literally goes without saying. The currently-observable universe is
finite. A more well-known saw is that the number of fundamental
particles in the obserable universe is smaller than the configuration
space of chess, or go.
As such, the availble microstates for any computing device is indeed
finite. Real-world computing devices have only finite state, not the
half-inifite tape of a Turing machine.
>You might find this unproblematic, or you may not. I get the impression
>that you think the lambda calculus actually exists somewhere.
Huh? Where on Earth did *that* leap from? All I said was that pure
Lisp is merely syntactic sugar for lambda-calculus; neither one is
more "real" or more a "natural language"[*] than the other.
[*} sensu Wilkins, a sense which does violence to any usage I've ever
heard from linguists, cognitive scientsts, philosophers, or AI
researchersd).
>I do not. Lisp is not the lambda calculus, and any installation of
>Lisp on a computer is at best an approximation (perhaps a very good
>one) of it.
Not at all. Both are Turing-equivalent copmutational devices. It just
so happens that Pure Lisp so close ot lambda-calculus that it *is*
merely syntactic sugar for lambda-calculus.
Mathematicians, after all, are limited to finite blackboards in which
to express and perform their lambda-reductions. (I think any smiley
here would be just as nonserious as yours was earlier).
>If you had a theorem proven that was the result of a Lisp run that went
>through three brownouts, two CPU failures and a hard disk crash, would
>you trust that it had delivered something radically true from the lambda
>calculus? I wouldn't.
Depends on how good the checkpointing was. No different than a
by-hand proof in first order predicate calculus with identity, when a
page gets lost before publication and and the proof is
reconstructed. Or, more pointedly, just as "true" as a proof from a
theorem-prover, or from a proof by hand.
A shame it didnt' come up a month or two back; it woudl have made an
intereseting question for John McCarthy (and Prof. Pfefferman) at
John's retirement dinner.
I'm trying to recall the details of papers which surveyed bugs in hand
proofs when verified by theorem-prover, and bugs in theorem-prover
proofs when checked by hand. Carolyn Talcott, perhaps?
Seriously, John: all this says to me is that you aren't prepared to
defend your original position.
>But then, poseurs like Matt and I are not worth discussing these things
>with, are we?
Nope. I think you're worth it. Even when we disagree,
you have something interesting to say.
People who parrot back a "correction" which everyone in the field
takes as read and which moreover the parroter only picked up a week or
so back, is in a rather different category.
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
John,
Thinking about this some more: perhaps we have very different ideas
about what computer languages mean, in general?
I am perfectly happy with definitions of purely applicative functional
program langauges (pure lisp, Backus' FP, ML, that sort of thing) in
terms of a mapping onto lambda-calculus-like languages.
Doing this sort of thing is, in fact, fairly standard in advanced
programming-language classes and tutorials. (I'm also been exposed to
both definition and implementation of programming languages in terms
of combinator logic, but that's another story.
I'm also happy with defining programming languages in terms of
mappings onto abstract machines: Landin's SECDM machine (which I
metioned earlier); the Warren abstract machine, beloved of Prolog
implementors; and less abstract, BCPL O-code, or Wirth/UCSD P-system
P-codes, or (the language du jour) Java byte-codes.
My experience in advanced programming-language courses is that one
ends up showing how to reduce one of these to some other; and to work
through, in pencil-and-paper or blackboard or whiteboard fashion,
various computations.
What you seem to be saying is that an evaluation of a Pure Lisp
construct on a blackboard is, in your ontology, a purely formal
language, not a natural one. But on the other hand, evaluation of the
identical Pure Lisp expresssion on a real computer is a natural
language.
Or, to say the same thing in a different domain: Horn clauses written
on a blackboard (in a certain arcane ASCII-expressible syntax) are a
"formal" langauge, the same clauses to a Prolog interpreter are a
"natural" language.
Is that _really_ what you're trying to say? Or have I misinterpreted
it somewhere?
Dare I mention the people at, um, Edinburgh, who came up with a
storable implementation of combinators and did computation by hardware
which did combinator reduction?[*] Is a trace of those circuits on a
blackboard 'mathematics' and thus a pure language, but evaluation of
the real hardware is a `natural language'? Ouch....
[*] Combinator logic, as in a logic of manipulation of combinators.
Wherein, if memory serves, an expression like SKK has the same
semantics as the more familiar Y fixed-point operator.
Matt Silberstein
jona...@DSG.Stanford.EDU (Jonathan Stone):
Jonathan, you once and on and on with more because I used "true"
instead of "valid" and "continuum" instead of "spectrum". If you are
going to make a big deal about terminology usage by others then you
had better get your terminology right. "Computer" was simply wrong.
>Or -- like Fred Colon commenting on seaweed inside Leonard's Boat - do
>you genuinely think you just Made a Contribution?
I am sorry, I thought that I could comment to something Jonathan said
without it becoming a personal issue. I guess Jonathan is simply
unable to refrain from making personal remarks, at least where I am
concerned.
--
Matt Silberstein TBC HRL OMM LotL
There is safety in numbers and people and things
And big wads of money and great big diamond rings
J.O.
Bobby D. Bryant
> I really meant it: where I'm coming from, this is so well-known it
> literally goes without saying.
Yes, but this is talk.origins, and half the readers probably don't even
know what a TM is, let alone the finer points of its definition.
People often have to bring up fine points here, because not all of us
know everything worth knowing about biology, cosmology, mathematics,
geology, nuclear physics, etc.
Bobby Bryant
Austin, Texas
John Wilkins
jokes are just confusing the issue. So I will attempt to put my case a
bit more carefully, with no smileys...
Some basic definitions:
A formal language is, for this purpose, a language without ambiguity,
amphiboly, or contradiction, which is fully coherent and formalised.
A natural language is one which can be ambiguous, amphibolous, or
contradictory, and may be unformalised in its syntax or symbolism, and
may also be incoherent.
I'm sure these are not good definitions, but I'm not completely defining
anyway. Let's say these are facets of natural and formal languages.
On this account, Lisp is a formal not a natural language, as you say. So
I will coin a couple of different defining terms to make it clear what I
mean. But hold on to that word "natural", because I'll get back to it,
if only to explain where I thought the obvious pun was...
A concrete object is one that is bounded by spacetime (this is Zalta's
definition), and an abstract object is one that is not nor can be. A
universal property is therefore abstract, and a particular property is
concrete. Call this nominalism, if you like, so long as it is understood
that this is a nominalism relative to some domain.
So let's call a "concrete language" any language which exists in
spacetime, and an "abstract language" one that cannot. Lambda calculus
is an abstract language. Lisp is a concrete language, where implemented
(and abstract where not). You might call this the "concrete
interpretation" of Lisp, or the instantiation of Lisp, if you like.
So, now we have a distinction between a physical (ie, bounded) object -
Lisp on Chaitin's Solaris, or on John Wilkins's Mac or whatever, which
has all kinds of limitations (in terms of pointers available at runtime,
implemented handles, and the like) - and the abstract *definition* of
Lisp in whatever standard it is defined. This is a metaphysical
definition. It doesn't affect the standing of Lisp WRT lambda calculus
or Gödelian arithmetic, &c. It does, however go to the question of how
any formal language is implemented (your nice blackboard example). Chalk
marks on a blackboard are not Lisp. Nor are ink marks on a manual, even
if the symbols represent a reverse polish notation definition of the
language. Where does Lisp exist if not in concrete form (even the
abstract definition must exist in concrete form)? If you are not a
Platonist, you must answer this way.
That is basically my point. Physical things, concrete things, can
instantiate, represent, or approximate formal things, or abstract
things. So, where's the pun? Well you have to peer into the deep dark
recesses of my addled brain. It isn't worth it (or pleasant) so I'll
give you the Author's Summary.
WVO Quine had an essay entitled "Naturalizing Epistemology", in, I
believe, his 1969 (the paper was a few years earlier, but is not in my
current bibliography). Regulars on t.o will know of G. E. Moore's
"naturalistic fallacy" - the error (a category error) of making
something prescriptive (the Good) a natural property; ie, a physical
property. Traditionally epistemology was held to be prescriptive, and so
it could not be "natural" in that sense. Quine aimed to use a natural
selection account - "Creatures inveterately wrong in their inductions
have a pathetic, but praiseworthy, tendency to die before reproducing
their kind", he said - to establish a "naturalised" epistemology.
In philosophical discourse, attempts to naturalise ethics, epistemology,
mind, and so forth have blossomed since Quine's essay. Some notable
naturalisers include Giere, Fodor, Devitt, Dretske, Kornblith, Laudan,
and Strawson, among others. The use of the term "natural" to mean
"non-abstract" has philosophical history.
What naturalism is trying to do, is to establish an account of something
procedural and prescriptive in terms of being concrete, rather than
abstract. So when I said that Lisp was a "natural" language, I merely
meant that any instantiation of Lisp must be a concrete instantiation.
There: if Harter has the most indirect puns, at least they elicit a
groan. Mine are merely needlessly obscure.
Selected References on Philosophical Naturalism
===============================================
Almeder, R. (1998). Harmless Naturalism: The Limits of Science and the
Nature of Philosophy, Chicago, Open Court.
Audi, R. (2000). "Philosophical Naturalism at the Turn of the Century."
Journal of Philosophical Research 25: 27-45.
Bechtel, W. and W. Callebaut (1993). Taking the naturalistic turn, or,
How real philosophy of science is done: conversations with William
Bechtel ... [et al.], Chicago, University of Chicago Press.
Bermudez, J. L. (1999). "Naturalism and Conceptual Norms." Philosophical
Quarterly 49(194): 77-85.
Dean, W. (1989). "Naturalism and Methodologism." American Journal of
Theology and Philosophy 10(2): 99-114.
Devitt, M. (1998). "Naturalism and the A Priori." Philosophical Studies
92(1-2): 45-65.
Dieterle, J. M. (1999). "Mathematical, Astrological, and Theological
Naturalism." Philosophia Mathematica 7(2): 129-135.
Dretske, F. I. (1995). Naturalizing the mind, Cambridge, Mass., MIT
Press.
Elder, C. L. (1989). "Realism, Naturalism and Culturally Generated
Kinds." Philosophical Quarterly 39: 425-444.
French, P. A., T. E. Uehling, et al. (1994). Philosophical naturalism,
Notre Dame, Ind., University of Notre Dame Press.
Furst, L. R. (1971). Naturalism, London, Methuen.
Gibson, R. F. (1987). "Quine On Naturalism and Epistemology." Erkenntnis
27: 57-78.
Godfrey-Smith, P. (1996). Complexity and the function of mind in nature,
Cambridge; New York, Cambridge University Press.
Gordon, J. (1983). "Is Naturalism Inescapable?" Analysis 43: 153-155.
Haldane, J. J. (1989). "Naturalism and the Problem of Intentionality."
Inquiry 32: 305-322.
Ketchum, R. J. (1991). "The Paradox of Epistemology: A Defense of
Naturalism." Philosophical Studies: 45-66.
Kornblith, H. (1993). Inductive inference and its natural ground: an
essay in naturalistic epistemology, Cambridge, Mass., MIT Press.
Laudan, L. (1990). "Normative Naturalism." Philosophy of Science 57(1):
44-59.
Maffie, J. (1990). "Naturalism and the Normativity of Epistemology."
Philosophical Studies 59(3): 333-349.
Manicas, P. T. and A. Rosenberg (1985). "Naturalism, Epistemological
Individualism and "the Strong Programme" in Sociology of Knowledge."
Journal for the Theory of Social Behaviour 15: 76-101.
Martin, M. (1984). "Does the Evidence Confirm Theism More Than
Naturalism?" International Journal for Philosophy of Religion 16:
257-262.
Matthen, M. (1991). "Naturalism and teleology." Biological Teleology:
656-667.
Moreland, J. P. (1998). "Searle's Biological Naturalism and the Argument
from Consciousness." Faith and Philosophy 15(1): 68-91.
Mounce, H. O. (1999). Hume's Naturalism, New York, Routledge.
O'Gorman, P. (1984). "Quine's Epistemological Naturalism." Philosophical
Studies 30: 205-219.
Peressini, A. (1998). "Naturalism, Evolution, and Self-Defeat."
International Journal for Philosophy of Religion 44(1): 41-51.
Pollard, S. and R. B. Graber (1989). "Mathematical Naturalism: an
Anthropological Perspective." Southern Journal of Philosophy 27:
427-441.
Quine, W. V. (1969). Ontological Relativity and Other Essays, New York,
Columbia University Press.
Railton, P. (1989). "Naturalism and Prescriptivity." Social Philosophy
and Policy 7(1): 151-174.
Rosenberg, A. (1990). "Normative Naturalism and the Role of Philosophy."
Philosophy of Science 57(1): 34-43.
Sabates, M. (1999). "Consciousness, Emergence and Naturalism." Teorema
18(1): 139-153.
Sellars, W. (1979). Naturalism and ontology, Reseda, Calif., Ridgeview.
Siegel, H. (1989). "Philosophy of Science Naturalized: Some Problems
With Giere's Naturalism." Studies in History and Philosophy of Science
20: 365-375.
Sosa, E. (1998). P.F. Strawson's Epistemological Naturalism. In The
Philosophy of P.F. Strawson. L. E. Hahn, Ed. Chicago, Open Court.
Stroud, B. (1996). "The Charm of Naturalism." Proceedings and Addresses
of the American Philosophical Association 70(2): 43-55.
Switankowsky, I. (1996). "Cartesian Naturalism." Eidos 13(2): 67-75.
Throop, W. M. and M. L. Knight (1987). "A Pragmatic Reconstruction of
the Naturalism/Anti-Naturalism Debate." Journal for the Theory of Social
Behaviour 17: 93-112.
Trout, J. D. (1998). Measuring the Intentional World: Realism,
Naturalism, and Quantitative Methods in the Behavioral Sciences, New
York, Oxford University Press.
Wagner, S. J. and R. Wagner (1993). Naturalism: a critical appraisal,
Notre Dame, Ind., University of Notre Dame Press.
Ward, A. (1999). "Naturalism and the Mental Realm." Southwest Philosophy
Review 15(1): 157-167.
Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
and in a followup article, Jonathan Stone <jona...@DSG.Stanford.EDU>
wrote:
> In article <1f5ozuk.1k3bg4qeiwodvN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
>
* Curse you, Ian Musgrave, for giving me this virus over the internet
(although I'm not sure how you sent a virtual, or abstract, virus to
cause physical illness. Perhaps you suffered from a category confusion?)
Okay, I said no jokes, but this is in a footnote.
John Wilkins
> In talk.origins I read <a1db4t$gd6$1...@Pescadero.DSG.Stanford.EDU> from
> jona...@DSG.Stanford.EDU (Jonathan Stone):
>
....
> >Or -- like Fred Colon commenting on seaweed inside Leonard's Boat - do
> >you genuinely think you just Made a Contribution?
>
> I am sorry, I thought that I could comment to something Jonathan said
> without it becoming a personal issue. I guess Jonathan is simply
> unable to refrain from making personal remarks, at least where I am
> concerned.
Yeah, now why is that? ISTM you make good contributions to the flow of
the discussion, and you know a fair bit more than I do about computation
and programming. Yet Jon seems to treat me with a dignity I don't
deserve, and you with vitriol you don't.
USENET is a funny non-place, I guess.
Richard Harter
wrote:
[snip much]
>
>In philosophical discourse, attempts to naturalise ethics, epistemology,
>mind, and so forth have blossomed since Quine's essay. Some notable
>naturalisers include Giere, Fodor, Devitt, Dretske, Kornblith, Laudan,
>and Strawson, among others. The use of the term "natural" to mean
>"non-abstract" has philosophical history.
>
>What naturalism is trying to do, is to establish an account of something
>procedural and prescriptive in terms of being concrete, rather than
>abstract. So when I said that Lisp was a "natural" language, I merely
>meant that any instantiation of Lisp must be a concrete instantiation.
>There: if Harter has the most indirect puns, at least they elicit a
>groan. Mine are merely needlessly obscure.
Quite an illuminating little essay. As a purely minor side remark,
your little gambit wasn't a pun; it was word play, playing on the
ambiguity in, you should excuse the expression, natural languages
where words such as "natural" rather naturally have several meanings.
The trick is to create an apparent paradox or contradiction by using a
word in an unexpected way. If you like, you can call it a meaning
pun.
I wouldn't say that it was needlessly obscure - excessively obscure is
more like it.
Richard Harter, c...@tiac.net,
http://www.tiac.net/users/cri, http://www.varinoma.com
Love, no matter how pure, is the most selfish of gifts.
For that reason it is the one gift that must be given.
Jonathan Stone
>
>> I really meant it: where I'm coming from, this is so well-known it
>> literally goes without saying.
>
>Yes, but this is talk.origins, and half the readers probably don't even
>know what a TM is, let alone the finer points of its definition.
And was wondering whether you knew the difference between UTM-based
definitions of P and NP.
But you really draw no distinction between `bringing up a fine point',
on the one hand, and trying to be an obnoxious smartarse by parroting
a point that came up only a week or so back?
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
hi john,
Sorry to hear about your virus. I hate opiates, myself.
(guess it takes all kinds...)
>So let's call a "concrete language" any language which exists in
>spacetime, and an "abstract language" one that cannot. Lambda calculus
>is an abstract language. Lisp is a concrete language, where implemented
>(and abstract where not). You might call this the "concrete
>interpretation" of Lisp, or the instantiation of Lisp, if you like.
John, no hard feelings, but this is nonsense. I can write Lisp or
lambda-calculus on a blackboard, I can do ``reductions'' on it.
People even implement lambda-calculus reduction systems on computers.
You seem to be asserting that when it's on a blackboard, it's
a concrete language, and when it
>So, now we have a distinction between a physical (ie, bounded) object -
>Lisp on Chaitin's Solaris, or on John Wilkins's Mac or whatever, which
>has all kinds of limitations (in terms of pointers available at runtime,
>implemented handles, and the like) - and the abstract *definition* of
>Lisp in whatever standard it is defined.
No, we don't. Any *acutal* Lisp lisp is *always* concrete, including
lisp on a blackboard. I think you are confusing at least three
separate ideas:
a) the *definition* of a Turing-complete language,
b) some implementation of that language (which in
the real world is always finite)
c) sentences in that language.
I grant you that any definition of a Turing-complete language which
explicitly states an infinite store is not realisable in the real
world. But any uh, ``utterance' in such a language is always finite,
and so is any real-world implementation -- whether it's chalk powder
or solid-state semiconductor makes no difference.
There really is no semantic or philosophical difference between
running out of blackboard, or running out of chalk; and running out of
RAM or of disk space. One can always go and buy more and continue,
right up to the limits of one's economic purchasing power. (Or of
one's civilization's technical ability, if it really came to it).
>This is a metaphysical
>definition. It doesn't affect the standing of Lisp WRT lambda calculus
>or Gödelian arithmetic, &c. It does, however go to the question of how
>any formal language is implemented (your nice blackboard example).
>Chalk marks on a blackboard are not Lisp. [...]
John, here you are wrong on the facts, in that you're at odds with
decades of English usage in CS departments Chalk marks of Lisp on the
blackboard *are* Lisp, in exactly the same way as chalk marks of
English on a blackboard are English. That's just a consequence
of how the English language is used.
I don't see any Platonism here at all. The distinction is between
``the definition of Lisp'' (of which there are several), and
``an instance of well-formed Lisp''.
you can do s/Lisp/English/ in the preceding paragraph. There is no
precise definition of English -- but then, we began this digression
when I objected to as misuse of the phrase "natural language".
[.. huge snip ]
>That is basically my point. Physical things, concrete things, can
>instantiate, represent, or approximate formal things, or abstract
>things.
Yes. But it's the definition of Turing-complete computational
formalism which is, _sometimes_, abstract. Specific finite instances
of sentences in the language, or specific finite implementations of
the formalism are entirely concrete -- precisely as much as an English
sentence written on a blackboard.
In case you didn't know: there are also definitions of
almost-but-not-quite Turing-complete languages with finite storage.
The language definition -- what you call the ``formal'' abstract and
not-concrete version -- defines, as part of the language, how an
executing program is notified that it's run out of storage (or just
about to run out, since the notification is done whilst there's still
enough space to do something about it).
(And in case you ask: yes, I *have* made more memory available to
long-running Lisp applications which incurred such a signal, and yes,
I trusted the results as much as I would any mechanised abacus, and
considerably more than a human calculation of the same scope.)
... and if I'm being philosophically naive, I'd appreciate if you let
me know before my brother gets back :).
PS: remind me to mention (gensym) and how AI researchers sometimes
used to cheat by picking symbol names that resonated with human
natural languages....
Chris Ho-Stuart
> I can see that codeine-addled[*] attempts at puns and philosophical
> jokes are just confusing the issue. So I will attempt to put my case a
> bit more carefully, with no smileys...
> Some basic definitions:
> A formal language is, for this purpose, a language without ambiguity,
> amphiboly, or contradiction, which is fully coherent and formalised.
> A natural language is one which can be ambiguous, amphibolous, or
> contradictory, and may be unformalised in its syntax or symbolism, and
> may also be incoherent.
> I'm sure these are not good definitions, but I'm not completely defining
> anyway. Let's say these are facets of natural and formal languages.
> On this account, Lisp is a formal not a natural language, as you say. So
> I will coin a couple of different defining terms to make it clear what I
> mean. But hold on to that word "natural", because I'll get back to it,
> if only to explain where I thought the obvious pun was...
> A concrete object is one that is bounded by spacetime (this is Zalta's
> definition), and an abstract object is one that is not nor can be. A
> universal property is therefore abstract, and a particular property is
> concrete. Call this nominalism, if you like, so long as it is understood
> that this is a nominalism relative to some domain.
> So let's call a "concrete language" any language which exists in
> spacetime, and an "abstract language" one that cannot. Lambda calculus
> is an abstract language. Lisp is a concrete language, where implemented
> (and abstract where not). You might call this the "concrete
> interpretation" of Lisp, or the instantiation of Lisp, if you like.
This does not make sense to me. A concrete implementation will
fail to handle all Lisp programs. If you take this approach,
there is no one Lisp language, and it may even vary with the time
of day on a machine with other processes competing for available
resources. I don't this it makes sense to call this the language.
When I explain programming "language" to an introductory
programming class, I tell them that that "language" here is
a misnomer. A programming language is actually a mathematical
notation for expressing algorithms; and more closely related to
the notations of arithmetic than to real languages.
Programming notations are used to express algorithms.
Natural languages are used to argue, to make love, to
sway emotions, to convey knowledge, to lecture, ...
Cheers -- Chris
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 02:31:54 -0500, john.w...@bigpond.com (John Wilkins)
>wrote:
>I wouldn't say that it was needlessly obscure - excessively obscure is
>more like it.
Specially given the accepted meaning of "natural language" within
linguistics and cognitive science and in AI.
I suggest that we in t.o adopt "Pnatural" language instead.
Jonathan Stone
Matt Silberstein <mat...@ix.netcom.com> wrote:
>In talk.origins I read <a1db4t$gd6$1...@Pescadero.DSG.Stanford.EDU> from
>jona...@DSG.Stanford.EDU (Jonathan Stone):
Matt,
It is a fact that any real-world computer does not have infinite
storage. It is also a fact that we treat computers as real-world
models of Turing machines. The finite storage of any real-world
computer means that we can also treat the computer as a model of
an *extraordinarly* large finite-state machine.
But, given the the exponential explosion of nuimber of states in a
finite-state machine to compute something as simple as *bit parity*,
it is generally not a helpful or insightful model.
The point you make about finite size might, in a very narrow sense, be
correct. But it's a rather unhelpful insight. To practicionters in
the field, it's impossible to tell those who have just obtained that
insight, from someone deliberately trying to be obnoxious (in this
case, about two week after having gotten the insight.)
I apologise sincerely if I mis-identifed your comment as being the of
second kind, rather than the first. If you truly weren't intending to
take a personal jab, then I apologize for responding as if you had.
But that is how I saw it.
PS: no, I don't criticize you for saying "true" rather than "valid".
The criticism is for denying that it really *was* a mistake; moreover
a mistake that, once you defended it, indicated you didn't have a clue
about the subject you were pontificating on at the time.
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Matt Silberstein <mat...@ix.netcom.com> wrote:
>Yeah, now why is that? ISTM you make good contributions to the flow of
>the discussion, and you know a fair bit more than I do about computation
>and programming. Yet Jon seems to treat me with a dignity I don't
>deserve, and you with vitriol you don't.
You don't pretend to know far more about it than you really do. You
don't squink for years over elementary mistakes.
I've never had friends visit town and ask, over dinner, if I had any
more amusing stories about gaffes you made, then lied about.
Perhaps that has something to do with it.
Jonathan Stone
Matt Silberstein <mat...@ix.netcom.com> wrote:
>jona...@DSG.Stanford.EDU (Jonathan Stone) wrote in message news:<a1b3e5$7qe$1...@Pescadero.DSG.Stanford.EDU>...
>
>[snip]
>
>> If you claim `design' is an algorithm, can you write down the
>> algorithm?
>
>So what if he can't (currently) produce this algorith?
Then for all practical purposers, there *is* no algorithm. And it is
intellecutally dishonest (or, in this instance, if you prefer,
Platonic) to proceed as if there was.
Are you asking because you really don't *understand*?
>So what if it
>will forever remain beyond any human's ability to produce this
>algorithm?
If you cannot state an algorithm, then what reason is there to believe
that the collection of human intelligent process we call `design' is,
in fact, an algorithm?
>I don't see how that says there is not some algorithmic
>process behind design. His abilities and knowledge do not determine
>what is or is not an algorithm.
You really don't have a very good understanding of science. If Bobby
wants to make a scientific argument that `design' is an algorithm,
then the onus is on him to demonstrate that it *is* an algorithm.
The most natural way to do so is to give the algorithm.
This is a step Bobby is conspicuously avoiding.
>> How is `design' any different in this respect than writing
>> poetry, painting pictures, or sculpting scupltures, or writing music,
>> or for that matter, doing science?
>
>A good question, but a good question does not an argument make. I
>suspect that for the purpose of the NFL and this discussion they are
>all sufficiently similar.
I think you missed the point of the question -- which was to show
that, scientifically, we don't have sufficient grounds to assume
either way.
>> If you will, for the sake of discussion, grant us the Church-Turing
>> thesis, what you are asking is whether any of these activities can be
>> produced or simulated by computers.
>
>Not quite. It asks whether they can be produced/similated by a
>universal turing machine. No actual computer is an actual universal
>turing machine. They all have finite limitations that the UTM lacks.
I'm looking at this again, with the assumption that it's ill-informed
rather than hostile.
In that light, a better answer is: no, that's not how the standard
terminology in the field works. Computers *are* models of UTMs,
and we can approach a UTM to any arbitrary but finite size.
The mathematical and ontological consequences of that I leave to John
Wilkins to explain. What we have is a series of larger computers: a
computer with one disk, two disks, ... N disks. No matter how many
we have, in principle we can always add one more. Conventional
mathematics is that the limit of such a sequence is infinite.
As for the practical implications: once your working no longer fits in
the margin, then darn well buy books with bigger margins. If your
computation runs out of storage, *buy more*.
Appeals to infiinte storage aren't helpful: First because no matter
how much you have, we can always add more. The limit diverges, or
loosely speaking, in the limit you *have* infinite storage.
Also recall that it takes infinite *time* to access infinite storage,
and infinite time is something we don't have. Practically, humans
usually run into the dilemma of the halting problem before worries
about storage.
Hence what happens in the real world, where practitioners mention the
limitations of finite storage *once*, in the 100-level class, and
take it as read thereafter.
Unless you're serious about really *huge* numbers, , say, computing
irrationals to oodles of digits, or really humungous factorials.
We can discuss that offline.
Bobby D. Bryant
> You really don't have a very good understanding of science. If Bobby
> wants to make a scientific argument that `design' is an algorithm,
> then the onus is on him to demonstrate that it *is* an algorithm.
>
> The most natural way to do so is to give the algorithm. This is a step
> Bobby is conspicuously avoiding.
The literate readers in the audience will notice that I asked a question
and told what my intuition on it was, and did not make any pretense of
making any scientific arguments. Instead, I asked other people to give
their thoughts on the subject so I could form a more solid and better
supported opinion on it. (BTW, thank you to everyone who has gone along
with that plan.)
The non-literate readers in the audience should notice that the last
line of my post was a question rather than an assertion:
So: is 'design' an algorithm?
Of course, those who were to illiterate -- or too blinded by their own
agenda -- to catch that then, probably won't catch it in this recap
either.
The nice thing about Usenet is, no matter what subject you bring up for
discussion, you can always find some self-appointed authority to make ex
cathedra pronouncements on it and insinuate that everyone else is a
dumbass for having any thoughts to the contrary.
Bobby Bryant
Austin, Texas
p.s. - My EA just found the solution to another problem. Thought you'd
like to know.
Matt Silberstein
[snip]
> What you seem to be saying is that an evaluation of a Pure Lisp
> construct on a blackboard is, in your ontology, a purely formal
> language, not a natural one. But on the other hand, evaluation of the
> identical Pure Lisp expresssion on a real computer is a natural
> language.
Both are physical, natural things, not formal ones.
[snip]
Richard Harter
Stone) wrote:
>>This is a metaphysical
>>definition. It doesn't affect the standing of Lisp WRT lambda calculus
>>or Gödelian arithmetic, &c. It does, however go to the question of how
>>any formal language is implemented (your nice blackboard example).
>>Chalk marks on a blackboard are not Lisp. [...]
>
>John, here you are wrong on the facts, in that you're at odds with
>decades of English usage in CS departments Chalk marks of Lisp on the
>blackboard *are* Lisp, in exactly the same way as chalk marks of
>English on a blackboard are English. That's just a consequence
>of how the English language is used.
You seem to be working very hard at not reading what John has to say.
What you say about English usage is quite true; however English usage
regularly blurs a critical distinction that philosophers fuss about,
the distinction between abstract entities and their instantiations.
The chalk marks on the board are simply chalk marks; we call them lisp
because there is an isomorphism between the chalk marks and Lisp as an
abstract entity - or perhaps not; there are some very large and very
ancient boojums in this territory. In any event, your appeal to
English usage in CS departments (if there be such) is an appeal to
irrelevant authority.
Chris Ho-Stuart
[snip]
> You seem to be working very hard at not reading what John has to say.
> What you say about English usage is quite true; however English usage
> regularly blurs a critical distinction that philosophers fuss about,
> the distinction between abstract entities and their instantiations.
> The chalk marks on the board are simply chalk marks; we call them lisp
> because there is an isomorphism between the chalk marks and Lisp as an
> abstract entity - or perhaps not; there are some very large and very
> ancient boojums in this territory. In any event, your appeal to
> English usage in CS departments (if there be such) is an appeal to
> irrelevant authority.
I'm Jonathan on this one. The chalk marks are not Lisp. The
chalk marks, however, do represent sentences in the language
(notation) Lisp. Similarly: magnetic patterns on a hard disk
are not Lisp: they may, however, represent sentences in the
language (notation) Lisp. Trying to distinguish a formal
notation (Lisp-1) from a concrete notation (Lisp-2) on the
basis that I can only fit so much on a chalk board, or excecute
expressions up to a certain size with my compiler, is not
sensible. What is concrete is the chalk. What is abstract is
the notation used to interpret the chalk marks. And this remains
the same language, even for different chalk boards of different
sizes and hence able to handle different numbers of markings.
Cheers -- Chris
John Wilkins
Well I must ask you then what I asked Jon - where does Lisp reside? Does
it exist other than in its concrete implementations.
>
> When I explain programming "language" to an introductory
> programming class, I tell them that that "language" here is
> a misnomer. A programming language is actually a mathematical
> notation for expressing algorithms; and more closely related to
> the notations of arithmetic than to real languages.
>
> Programming notations are used to express algorithms.
>
> Natural languages are used to argue, to make love, to
> sway emotions, to convey knowledge, to lecture, ...
Oh, absolutely. Nobody ever got laid by writing a COBOL program, I'm
sure. But I am aware that this is a matter of abstract objects like
numbers. The debate began on the standing of numbers.
BTW: DNA is not a language, either ;-)
>
> Cheers -- Chris
John Wilkins
> In article <1f5p54u.1kvkgz14yt3stN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> hi john,
>
> Sorry to hear about your virus. I hate opiates, myself.
> (guess it takes all kinds...)
Ta. I will die momentarily, thus resolving this debate. All that is
needed is that vein that's throbbing on my temple to finally burst.
>
>
>
> >So let's call a "concrete language" any language which exists in
> >spacetime, and an "abstract language" one that cannot. Lambda calculus
> >is an abstract language. Lisp is a concrete language, where implemented
> >(and abstract where not). You might call this the "concrete
> >interpretation" of Lisp, or the instantiation of Lisp, if you like.
>
> John, no hard feelings, but this is nonsense. I can write Lisp or
> lambda-calculus on a blackboard, I can do ``reductions'' on it.
*You* can. But can my 10 year old son? If not (and as smart as he is, he
cannot), then lamda calculus does not exist on that blackboard. The
blackboard is not a UTM. It cannot do the reductions itself. Where does
the LC exist here?
>
> People even implement lambda-calculus reduction systems on computers.
>
> You seem to be asserting that when it's on a blackboard, it's
> a concrete language, and when it
Yes I am (making the obvious extrapolation) - if it exists in a physical
form, then it is not abstract, but then it is not LC simpliciter. For a
start, symbols of LC are not subject to the terrors of the duster, but
chalk marks on blackboards are.
>
>
> >So, now we have a distinction between a physical (ie, bounded) object -
> >Lisp on Chaitin's Solaris, or on John Wilkins's Mac or whatever, which
> >has all kinds of limitations (in terms of pointers available at runtime,
> >implemented handles, and the like) - and the abstract *definition* of
> >Lisp in whatever standard it is defined.
>
> No, we don't. Any *acutal* Lisp lisp is *always* concrete, including
> lisp on a blackboard. I think you are confusing at least three
> separate ideas:
> a) the *definition* of a Turing-complete language,
> b) some implementation of that language (which in
> the real world is always finite)
> c) sentences in that language.
>
> I grant you that any definition of a Turing-complete language which
> explicitly states an infinite store is not realisable in the real
> world. But any uh, ``utterance' in such a language is always finite,
> and so is any real-world implementation -- whether it's chalk powder
> or solid-state semiconductor makes no difference.
>
> There really is no semantic or philosophical difference between
> running out of blackboard, or running out of chalk; and running out of
> RAM or of disk space. One can always go and buy more and continue,
> right up to the limits of one's economic purchasing power. (Or of
> one's civilization's technical ability, if it really came to it).
Oh I agree with this statement. Does this mean, that if you run out of
chalk when doing a derivation in Lisp, that Lisp is unable to do that
derivation? No. Hence *I* am not confusing Lisp with its implementation.
I believe you are. So, face the demon, Jon - where do formal languages
exist? Are they physical? If so, where are their definitions? If all
definitions of Lisp were destroyed tomorrow, would Lisp cease to exist?
In my view, it would. That makes all definitions of Lisp concrete
objects, and hence
>
>
> >This is a metaphysical
> >definition. It doesn't affect the standing of Lisp WRT lambda calculus
> >or Gödelian arithmetic, &c. It does, however go to the question of how
> >any formal language is implemented (your nice blackboard example).
> >Chalk marks on a blackboard are not Lisp. [...]
>
> John, here you are wrong on the facts, in that you're at odds with
> decades of English usage in CS departments Chalk marks of Lisp on the
> blackboard *are* Lisp, in exactly the same way as chalk marks of
> English on a blackboard are English. That's just a consequence
> of how the English language is used.
But English is a historical object, bounded by time and space. The LC is
not (theoretically).
As to being at odds with decades of CS usage, I observe that when I did
my computing, I noted that their usage was at odds with centuries of
philosophical logic usage. I got used to the idea. But their usage is
not authoritative.
>
> I don't see any Platonism here at all. The distinction is between
> ``the definition of Lisp'' (of which there are several), and
> ``an instance of well-formed Lisp''.
>
> you can do s/Lisp/English/ in the preceding paragraph. There is no
> precise definition of English -- but then, we began this digression
> when I objected to as misuse of the phrase "natural language".
>
>
> [.. huge snip ]
>
> >That is basically my point. Physical things, concrete things, can
> >instantiate, represent, or approximate formal things, or abstract
> >things.
>
> Yes. But it's the definition of Turing-complete computational
> formalism which is, _sometimes_, abstract. Specific finite instances
> of sentences in the language, or specific finite implementations of
> the formalism are entirely concrete -- precisely as much as an English
> sentence written on a blackboard.
Fine, so what *is* the abstraction, ontologically speaking? I say that
it is a matter of a relation between symbols and cognitive devices,
which are physical.
>
> In case you didn't know: there are also definitions of
> almost-but-not-quite Turing-complete languages with finite storage.
> The language definition -- what you call the ``formal'' abstract and
> not-concrete version -- defines, as part of the language, how an
> executing program is notified that it's run out of storage (or just
> about to run out, since the notification is done whilst there's still
> enough space to do something about it).
Nice, but not a counterinstance. Let's remain here with Turing-complete
languages.
>
> (And in case you ask: yes, I *have* made more memory available to
> long-running Lisp applications which incurred such a signal, and yes,
> I trusted the results as much as I would any mechanised abacus, and
> considerably more than a human calculation of the same scope.)
>
>
> ... and if I'm being philosophically naive, I'd appreciate if you let
> me know before my brother gets back :).
You are. (Sorry, you asked). This is the medieval universals debate
begun by Roscellin, who argued that a universal (an abstract in our
terminology) is mere breath "flatus vocus". I don't go this far, but it
does have a history (actually, I think that history begins sometime
between Aristotle and Porphyry, but is not preserved).
Then there is the type-token distinction, in place since Peirce. We are
discussing the ontological status of the type, not the tokens.
>
> PS: remind me to mention (gensym) and how AI researchers sometimes
> used to cheat by picking symbol names that resonated with human
> natural languages....
Even that's not new - philosophers have been doing *that* for centuries
also. Take the word "idea" for example: you might think it had a
philosophical meaning that somehow related to its ordinary usage...
Chris Ho-Stuart
> Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
[snip]
>> This does not make sense to me. A concrete implementation will
>> fail to handle all Lisp programs. If you take this approach,
>> there is no one Lisp language, and it may even vary with the time
>> of day on a machine with other processes competing for available
>> resources. I don't this it makes sense to call this the language.
>
> it exist other than in its concrete implementations.
Lisp, being an abstract formal system, does not exist as
a physical thing. The concrete implementations are not
Lisp: they are Lisp interpreters. Lisp resides in the same
place as lambda calculus.
But let me ask you: if my computer has less memory available
than yours, but they run the same software to interpret Lisp
programs, and there are some Lisp programs you can execute
that cause my smaller machine to run out of memory -- can we
should we say that both machines are running Lisp programs?
If my niece has a smaller vocabulary than me, then can we
really say that she speaks English, or is she using some other
language?
Cheers -- Chris
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 15:09:12 -0500, jona...@DSG.Stanford.EDU (Jonathan
>Stone) wrote:
>
>
>You seem to be working very hard at not reading what John has to say.
I dont think so. I think I'm trying to convey that comptuer
scientists put the boundaries in a different place than John's
ontology woudl have them fall.
>What you say about English usage is quite true; however English usage
>regularly blurs a critical distinction that philosophers fuss about,
>the distinction between abstract entities and their instantiations.
>The chalk marks on the board are simply chalk marks; we call them lisp
>because there is an isomorphism between the chalk marks and Lisp as an
>abstract entity
No. The isomorphism is between fragments ofLisp on the blackboard,
;and fragments of Lisp in any of the concrete implementations of Lisp
which John is willing to accept as abstract.
There **is no difference**, given that humans are capable of executing
Lisp framgents on blackboards.
>- or perhaps not; there are some very large and very
>ancient boojums in this territory. In any event, your appeal to
>English usage in CS departments (if there be such) is an appeal to
>irrelevant authority.
If you think that, then I owe you and John an apology. The appeal is
not to authority, but again, to clear communication.
I'm not sure why, but you seem to be working very, very hard to ingore
my original point: where John placed his demarcation point is simply
*wrong*. not as a matter of philosophy, or of English usage, but as a
matter of simple fact. Either you or John can visit a grad-level
programming class, and see both Lisp evaluation and lambda-calculus
reduction being done on whiteboards. The whiteboards are, indeed,
concrete and finite.
I'm damned if I can see how one is "concrete" but the other isn't.
And that was -- if you recall -- what John was claiming early on:
that Lisp was a "Pnatural language" but lambda-calculus wasn't.
Yes, there are ancient boojums here. John is trying to fence them in.
I think the problem is that -- as with the ancient Greeks -- Church
and Turing movedx the fences from where Johns, ahhh, idealistic
position would have them lie.
(Yes, thats a joke. I only hope John gets a laugh; flu in midsummer is
no fun at all.)
Richard Harter
<host...@sky.fit.qut.edu.au> wrote:
>Richard Harter <c...@tiac.net> wrote:
>[snip]
>> You seem to be working very hard at not reading what John has to say.
>> What you say about English usage is quite true; however English usage
>> regularly blurs a critical distinction that philosophers fuss about,
>> the distinction between abstract entities and their instantiations.
>> The chalk marks on the board are simply chalk marks; we call them lisp
>> because there is an isomorphism between the chalk marks and Lisp as an
>> abstract entity - or perhaps not; there are some very large and very
>> ancient boojums in this territory. In any event, your appeal to
>> English usage in CS departments (if there be such) is an appeal to
>> irrelevant authority.
>
>I'm Jonathan on this one. The chalk marks are not Lisp.
I dunno if you are with him or not. His wording suggests that he
thinks that the stuff on the blackboard is lisp, actual lisp. Perhaps
you and he can agree whilst using apparently contradictory words.
>The
>chalk marks, however, do represent sentences in the language
>(notation) Lisp. Similarly: magnetic patterns on a hard disk
>are not Lisp: they may, however, represent sentences in the
>language (notation) Lisp.
Are you thereby agreeing with John that Lisp is an abstract entity not
present in space-time? (Not that John necessarily holds that position
- he is pushing a different position.) As a side note, you shouldn't
confuse "language" and "notation".
>Trying to distinguish a formal
>notation (Lisp-1) from a concrete notation (Lisp-2) on the
>basis that I can only fit so much on a chalk board, or excecute
>expressions up to a certain size with my compiler, is not
>sensible.
I might or might not agree. Do you understand the difficulty inherent
in your claim?
>What is concrete is the chalk. What is abstract is
>the notation used to interpret the chalk marks. And this remains
>the same language, even for different chalk boards of different
>sizes and hence able to handle different numbers of markings.
Exactly where would this notation be? What is the nature of its
existence? How can these finite transient events, chalk marks on a
blackboard and the like, be coupled to those abstractions existing
outside of time and space? Plato wants to hear your answer.
Marty Fouts
c...@tiac.net (Richard Harter) writes:
> On 8 Jan 2002 18:14:07 -0500, Chris Ho-Stuart
> <host...@sky.fit.qut.edu.au> wrote:
>
[snip]
> >What is concrete is the chalk. What is abstract is the notation
> >used to interpret the chalk marks. And this remains the same
> >language, even for different chalk boards of different sizes and
> >hence able to handle different numbers of markings.
>
> Exactly where would this notation be? What is the nature of its
> existence? How can these finite transient events, chalk marks on a
> blackboard and the like, be coupled to those abstractions existing
> outside of time and space? Plato wants to hear your answer.
>
Plato's cave wouldn't be nearly so dark and shadow ridden if he bought
the new iMac and used it as a table lamp.
So, if someone were now to discover Fermat had written down his proof
in a different margin, later, did the proof exist during the
intervening period when Fermat died and it was rediscovered, or not?
If so, then those marks serve as the connection and we can move on to
figuring out how. If not, than what purpose did those marks serve?
Marty
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
>
>*You* can. But can my 10 year old son? If not (and as smart as he is, he
>cannot), then lamda calculus does not exist on that blackboard. The
>blackboard is not a UTM. It cannot do the reductions itself. Where does
>the LC exist here?
John ,
Your claim of yesterday was that there's some numinous(?) difference
between Lisp on a blackboard and lambda-calculus on a blackboard. My
point is that, empirically, there is no such difference.
We can write English on a blackboard. You and I, and your ten-year-old
son, can comprehend it. But the two-year-old who I toss over my head
every time I meet him, cannot.
I hope you are not aruging that the English goes away if we leave the
two-year-old alone in the room? (and no fair appealing to erasers,
either. Let's say the blackboard is above a two-year-old's reach.)
Praps you should go back to bed until you are feeling less Searle-ey.
(PS: I do hope that gets a laugh. I'm no Harter, but I am not quite
as unsophisticated as you think :)
Chris Ho-Stuart
> On 8 Jan 2002 18:14:07 -0500, Chris Ho-Stuart
> <host...@sky.fit.qut.edu.au> wrote:
>
>>[snip]
>>> You seem to be working very hard at not reading what John has to say.
>>> What you say about English usage is quite true; however English usage
>>> regularly blurs a critical distinction that philosophers fuss about,
>>> the distinction between abstract entities and their instantiations.
>>> The chalk marks on the board are simply chalk marks; we call them lisp
>>> because there is an isomorphism between the chalk marks and Lisp as an
>>> abstract entity - or perhaps not; there are some very large and very
>>> ancient boojums in this territory. In any event, your appeal to
>>> English usage in CS departments (if there be such) is an appeal to
>>> irrelevant authority.
>>
>>I'm Jonathan on this one. The chalk marks are not Lisp.
>
> thinks that the stuff on the blackboard is lisp, actual lisp. Perhaps
> you and he can agree whilst using apparently contradictory words.
>
>>chalk marks, however, do represent sentences in the language
>>(notation) Lisp. Similarly: magnetic patterns on a hard disk
>>are not Lisp: they may, however, represent sentences in the
>>language (notation) Lisp.
>
> present in space-time? (Not that John necessarily holds that position
> - he is pushing a different position.) As a side note, you shouldn't
> confuse "language" and "notation".
I agree with your side note, and am of the opinion that
Lisp is commonly called a computer "language" when it is
really a notation. (Same goes for C, and Cobol, and so on.)
Mathematicians use "language" to mean a set of finite strings.
A notation, as I understand it, is a language in the
mathematical sense along with some semantics to relate
it to a formal model. That is, it is a "concrete" syntax
plus some semantics. (But concrete syntax is still an
abstract entity in the sense we are using.)
"Language" in common discourse is something else again.
As I said previously, I tell my introductory programming
students that programming "languages" are better thought
of as "notations". I think this can help them get over
the hurdle of realising that the compiler is not reading
their programs with the kind of sympathetic intuition
they except from people reading an essay: but is doing
calculations more like a calculator taking an equation
and calculating a result. But I digress....
Yes, Lisp is an abstract entity not present in space-time.
>>Trying to distinguish a formal
>>notation (Lisp-1) from a concrete notation (Lisp-2) on the
>>basis that I can only fit so much on a chalk board, or excecute
>>expressions up to a certain size with my compiler, is not
>>sensible.
>
> I might or might not agree. Do you understand the difficulty inherent
> in your claim?
Perhaps not...
>>What is concrete is the chalk. What is abstract is
>>the notation used to interpret the chalk marks. And this remains
>>the same language, even for different chalk boards of different
>>sizes and hence able to handle different numbers of markings.
>
> Exactly where would this notation be? What is the nature of its
> existence? How can these finite transient events, chalk marks on a
> blackboard and the like, be coupled to those abstractions existing
> outside of time and space? Plato wants to hear your answer.
The notation is not in space-time; asking "where" it is does
not seem to me to be a sensible question. I consider that transient
events are coupled to abstractions by our action in agreeing
to some meaning or interpretation for the notation.
I'm no philosopher and would be unable to decribe the problems
Plato was worried about. but I consider that Lisp is in the same
place as lambda-calculus. The difference between lambda-calculus
and Lisp is that Lisp incorporates more concrete syntax; but that
remains an abstraction as well.
Cheers -- Chris
Jonathan Stone
>I'm Jonathan on this one. The chalk marks are not Lisp. The
>chalk marks, however, do represent sentences in the language
>(notation) Lisp. Similarly: magnetic patterns on a hard disk
>are not Lisp: they may, however, represent sentences in the
>language (notation) Lisp. Trying to distinguish a formal
>notation (Lisp-1) from a concrete notation (Lisp-2) on the
>basis that I can only fit so much on a chalk board, or excecute
>expressions up to a certain size with my compiler, is not
>sensible. What is concrete is the chalk. What is abstract is
>the notation used to interpret the chalk marks. And this remains
>the same language, even for different chalk boards of different
>sizes and hence able to handle different numbers of markings.
Thank you, Chris. That exposition is very clear. (Myself, given time,
I would draw a distinction between ``notation'' on the one hand, and
an association of ``semantics'' with the notation. One concrete
example which captures the point: English written in Morse rather
than the conventional Latin alphabet. But that's just recursing on
where the boojums are hiding.)
However, I *think* John is also trying to make another distinction: to
distinguish in-principle infinite computational mechanisms (which, in
John's ontology, are therefore abstract and incapable of realization),
from _practical_, finite implementations of otherwise-Turing-complete
computational ``formalisms'' -- which, as physical, empirical devices,
occupy a different spot in John's ontology).
Where I'm coming from is that _in practice_, there's no discernable
difference between a blackboard full of Lambda-calculus and a
blackboard full of pure, applicative Lisp. The isomorphism between
these two formalisms is so close that, amongst computer scientists
doing theoretical work in programming langauges and AI researchers
(at least the ones I hang out with), the difference is regarded as
syntactic sugar.
Yet in the ontolgy John wants to draw around the boojums, those two
formalisms -- lambda-calculus vs. Lisp -- fall into two boxes, which
John has carefully designed so as to be disjoint.
Me, I dont see much difference, _except_ that historically one came
first, and so mathematicians and logicians and philosophers are more
comfortable with it. (Me? I'm just a heathen, happy to derive Godel's
incompleteness result from Church-Turing and the halting problem. I
know Richard objected to that in the past, but its not uncommon
amongst CS theoreticians. Over drinks, they will say it's a two-line
proof; but they seem to be somewhat stricter if there are
mathematicians in the room. (In some ways, appealing to the
Church-Turing thesis is a bit like appealing to the
Continuum Hypothesis....))
Hm. On reflection, perhaps the second response I gave to Matt's point
about infinite space is actually on-topic to the "infinite space"
issue here. At least as a datapoint on how we scruffies look at it.
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 18:14:07 -0500, Chris Ho-Stuart
><host...@sky.fit.qut.edu.au> wrote:
>
>I dunno if you are with him or not. His wording suggests that he
>thinks that the stuff on the blackboard is lisp, actual lisp. Perhaps
>you and he can agree whilst using apparently contradictory words.
Richard,
Can I ask you to clarify just what do you mean by "acutal", here?
I have been quite careful and consisten to say that the fragments
of Lisp written on a blackboard is Lisp, in *exactly* the same sense
that English written on a blackboard is English.
If you'd prefer we say `fragments _of_ English", or
"senttences _in_ English", then I will try and follow that convention.
But please don't take epistemic positions and attribute them to me,
OK? We've had more than enough accusations of creationist debating
tactics as it is.
[Snip notation vs. language [vs semantics]": Richard and I have
the same questions here]
>I might or might not agree. Do you understand the difficulty inherent
>in your claim?
I dunno if he does or not. I don't buy Chris' use of "notation". You
can wave the shadow of Plato as you like; humans still demonstrate
competence in this -- yes, it *is* occasionally an exam question --
and I see no reason to accept a world of Forms outside the cave.
>>What is concrete is the chalk. What is abstract is
>>the notation used to interpret the chalk marks. And this remains
>>the same language, even for different chalk boards of different
>>sizes and hence able to handle different numbers of markings.
>
>Exactly where would this notation be? What is the nature of its
>existence?
We can start with:
``Recursive Functions of symbolic expression and their Computation
By Machine'', Part 1, Communications of the ACM, April 1960.
See http://www-formal.stanford.edu/jmc/recursive/recursive.html.
If I ask John nicely, I might be able to see a copy with concrete
ink marks on paper. Pure lisp is a different story.
These days, you'd probably want the revised common lisp report, or the
"silver bible" by Guy steele and Jerrry Sussman. My copy is a bootleg
of the first edition, from Digital Press, kindly photocopied by
Digital Equipment Corp, Wellington, New Zealand.
>How can these finite transient events, chalk marks on a
>blackboard and the like, be coupled to those abstractions existing
>outside of time and space? Plato wants to hear your answer.
Depends. John is happy to say that he and I and his ten-year-old can
understand English. John is definitely not saying that Enlish doesn't
exist, unless Plato is sitting outside his cave with the Formal copy of
Big Oxford, and Strunk & White. Are you?
Jonathan Stone
>I agree with your side note, and am of the opinion that
>Lisp is commonly called a computer "language" when it is
>really a notation. (Same goes for C, and Cobol, and so on.)
Chris,
Is lambda-calculus a notation? How about Turing machines, or recursive
function theory?
I'm asking because I think your ontology rides rough-shod over John's
distinction between infinte-storage and finite-storage notations.
(I, on the otherh and, fail to see any difference between finite-size
*instances* of otherwise-equivalent computational "notation"s).
>Mathematicians use "language" to mean a set of finite strings.
No, I dont think they do. Computer scientists, and formal approaches
to langauge since Chomsky (and arguably, perhaps Emil Post), use
"language" to mean some set of strings, drawn over some alphabet,
recognised by some automaton defined in a formal language, where the
automaton has both _terminal_ symbols (i.e., those that can appear in
the string) and _nonterminal_ symnbosl (i.e., symbosl which do not
appear in the input but which are used by the automaton for its
internal purposes). The string is _in_ the language if and only if
the automaton, when started in a designated `start' state, consumes
the entire input and halts in a state marked as `acceptiong'.
The automaton is, of cousre, a set of states, and a function from
(state, input symbol ) to state -- plus or minus an action on an
ancillary push-down state or input tape.
Technically, a formal language is a 4-tuple. These days, textbooks
just denote the starting state as S, and a formal definition of a
language language is effectively a 3-tuple.
Languages can easily be infinite sets. Simple examples ( excluding
kleene-closures) include the set of all well-formed parenthesised
expressions, and the set of strings which are palindromes over some
alphabet.
Do mathematicians have some other formal definition of "language"?
>A notation, as I understand it, is a language in the
>mathematical sense along with some semantics to relate
>it to a formal model. That is, it is a "concrete" syntax
>plus some semantics. (But concrete syntax is still an
>abstract entity in the sense we are using.)
True i guess, but that's no reason to go postal, er, Platonic over it.
Is there anyone willing to say that Plato is sitting outside the cave,
holding an ideal Palindrome?
>but is doing
>calculations more like a calculator taking an equation
>and calculating a result. But I digress....
``Me too''.
>Yes, Lisp is an abstract entity not present in space-time.
The _definition_ of Lisp is abstract. Expressions in Lisp written on
some concrete media are no less and no more concrete than expressions
in English written on the same media.
Is English an abstract entity not present in space-time?
I confess to having known people who *speak* in Lisp, viz:
one: foodp?
another one: nil
<time passes>
another one: foodp?
first one: t
at which point they go off for lunch. I wish I was making this up,
but I'm not.
>The notation is not in space-time; asking "where" it is does
>not seem to me to be a sensible question. I consider that transient
>events are coupled to abstractions by our action in agreeing
>to some meaning or interpretation for the notation.
One approach, perhpas, is to say that the meaning lies inside our
individual heads; and we're just lucky that the world we each see, and
the meanings we draw from it, seem close enough for communication.
There are alternatives to vats-in-brains besides Platonism.
Then again, I hear that ``Australasian pragmatism'' gets a bad name
for more-or-less precisely this sort of thinking....
Chris Ho-Stuart
> In article <3c3b...@news.qut.edu.au>,
> Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
>>Richard Harter <c...@tiac.net> wrote:
>
>>I agree with your side note, and am of the opinion that
>>Lisp is commonly called a computer "language" when it is
>>really a notation. (Same goes for C, and Cobol, and so on.)
>
> Chris,
>
> Is lambda-calculus a notation? How about Turing machines, or recursive
> function theory?
Actually, I originally made the distinction between language
and notation to try and help first year students avoid thinking
of computers as some kind of alien intelligence with their own
special language comparable to a language like English or French.
It sometimes is a help for them to think of executing a program
as more like calculating the result of an equation than reading
and interpreting an essay.
I still talk about computer languages: but still remind students
in their first lecture that this actually means some kind of
notation for writing algorithms.
And then, with regular/context-sensitive/context-free/etc languages
in my mind, I was seduced into making a distinction between formal
languages as a set of strings, and notation as a set of strings with
some well defined meaning.
But anyhow -- no, I would not class lambda-calculus, or turing
machines, or recursive function theory, as notation -- or as
language for that matter. I would consider them mathematical
models, and in each case there is a range of ways to express
terms/machines/functions which is where the language or notation
comes into it. But I still consider that formal languages are
also another kind of mathematical abstraction, having the same
kind of abstract existence as the lambda-calculus, etc.
> I'm asking because I think your ontology rides rough-shod over John's
> distinction between infinte-storage and finite-storage notations.
> (I, on the otherh and, fail to see any difference between finite-size
> *instances* of otherwise-equivalent computational "notation"s).
I guess I missed something, then. The terms of lambda-calculus,
and the corresponding well formed sentences of Lisp, are equally
unbounded in size.
>>Mathematicians use "language" to mean a set of finite strings.
>
> No, I dont think they do. Computer scientists, and formal approaches
> to langauge since Chomsky (and arguably, perhaps Emil Post), use
> "language" to mean some set of strings, drawn over some alphabet,
> recognised by some automaton defined in a formal language, where the
> automaton has both _terminal_ symbols (i.e., those that can appear in
> the string) and _nonterminal_ symnbosl (i.e., symbosl which do not
> appear in the input but which are used by the automaton for its
> internal purposes). The string is _in_ the language if and only if
> the automaton, when started in a designated `start' state, consumes
> the entire input and halts in a state marked as `acceptiong'.
>
> The automaton is, of cousre, a set of states, and a function from
> (state, input symbol ) to state -- plus or minus an action on an
> ancillary push-down state or input tape.
>
> Technically, a formal language is a 4-tuple. These days, textbooks
> just denote the starting state as S, and a formal definition of a
> language language is effectively a 3-tuple.
I disagree. Technically, the automaton is the 4-tuple, and the set
of strings it accepts is the language.
This allows one to speak of equivalence of different automata,
based on whether they accept the same language, amongst other
things.
We can also speak of non-recursive languages, not accepted by
any automaton, or even a Turing machine.
> Languages can easily be infinite sets. Simple examples ( excluding
> kleene-closures) include the set of all well-formed parenthesised
> expressions, and the set of strings which are palindromes over some
> alphabet.
Of course. I said a language was a set of finite strings, not a
finite set of strings. Actually, you can also have languages of
infinite strings, with special kinds of automata and some new
distinctions between the various w-languages; but this is less
common.
There are also automata accepting trees, (both finite trees and
infinite trees) although it is not usual to speak of sets of
trees as a "language".
> Do mathematicians have some other formal definition of "language"?
We are thinking of the same thing. I am simply using "language" to
refer to the set of strings, and you appear to be using "language"
to refer to the automaton; which I find a bit unusual, but
comprehensible.
>>A notation, as I understand it, is a language in the
>>mathematical sense along with some semantics to relate
>>it to a formal model. That is, it is a "concrete" syntax
>>plus some semantics. (But concrete syntax is still an
>>abstract entity in the sense we are using.)
>
> True i guess, but that's no reason to go postal, er, Platonic over it.
> Is there anyone willing to say that Plato is sitting outside the cave,
> holding an ideal Palindrome?
>
>>but is doing
>>calculations more like a calculator taking an equation
>>and calculating a result. But I digress....
>
> ``Me too''.
>
>>Yes, Lisp is an abstract entity not present in space-time.
>
> The _definition_ of Lisp is abstract. Expressions in Lisp written on
> some concrete media are no less and no more concrete than expressions
> in English written on the same media.
Yes.
> Is English an abstract entity not present in space-time?
Yes; though as above, a particular sentence spoken in
English is another thing.
> I confess to having known people who *speak* in Lisp, viz:
>
> one: foodp?
> another one: nil
> <time passes>
> another one: foodp?
> first one: t
>
> at which point they go off for lunch. I wish I was making this up,
> but I'm not.
CS people do tend to be a bit weird.
>>The notation is not in space-time; asking "where" it is does
>>not seem to me to be a sensible question. I consider that transient
>>events are coupled to abstractions by our action in agreeing
>>to some meaning or interpretation for the notation.
>
> One approach, perhpas, is to say that the meaning lies inside our
> individual heads; and we're just lucky that the world we each see, and
> the meanings we draw from it, seem close enough for communication.
>
> There are alternatives to vats-in-brains besides Platonism.
> Then again, I hear that ``Australasian pragmatism'' gets a bad name
> for more-or-less precisely this sort of thinking....
Losing me... like I said: I'm philosophically naive.
Cheers -- Chris
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
[...]
>
>Oh I agree with this statement. Does this mean, that if you run out of
>chalk when doing a derivation in Lisp, that Lisp is unable to do that
>derivation? No. Hence *I* am not confusing Lisp with its implementation.
No; I beleive you are instead confusing concrete instances of Lisp,
with a formal definition of Lisp. As I keep saying: I prefer to deal
with finite utterances.
>I believe you are. So, face the demon, Jon - where do formal languages
>exist?
Where do unicorns exist? we have lots of ink and other utterances
which refer to unicorns.
English is a wonderful language; it leets one write things which have
no intension. John, where do numbers exist? If their definitions were
destroyed, are you saying numbers would cease to exist, too?
Where does that leave us when it come sto counting fossil tree rings,
rings which are reliably dated to long before the human species, let
alone our invention of ideas like "number?
>Are they physical? If so, where are their definitions? If all
>definitions of Lisp were destroyed tomorrow, would Lisp cease to exist?
If all English dictionaries were destroyed tomorrow, would English
cease to exist? Likewise, if all _implementations_ of Lisp were
destroyed tomorrow, would the _definition_ cease to exist?
Hm? How about vice-versa?
>In my view, it would.
Just as a matter of fact, I think your viewpoint is
insupportable. Individual copies of the ink and paper which define the
formal system do exist, as real ink and paper. It is technically
possible that some anti-Lisp Jihad could can destroy all copies; yet
fail to destroy either all the the implementations of Lisp
interpreters, or fail to destroy all human knowledge of Lisp. There
might still be instances of Lisp implementations embedded inside verra
complicated toasters, doing useful work. In that sense, "Lisp" would
exist-- as implementations. But we might no longer have a name for the
abstraction which those surviving implementations share.
Does the abstraction cease to exist if we destroy its definition, yet
leave some instances of the implementation around? I dunno, but I hear
a Rosetta stone waiting in the wings.
Was the Latin language destroyed when all native speakers of it died?
>But English is a historical object, bounded by time and space. The LC is
>not (theoretically).
Wrong. (I think you must be feeling very ill.) All usages of the
lambda calculus *do* exist as historical objects, bounded in time and
space, in exactly the same way that usage of English do.
Why do you keep arguing otherwise? No, I tkae that back. Why do you
argue that lambda-calculus is any different from, say, English?
Are you trying to push for the lambda-calculus having some
platonic Form?
All those concrete examples are no more and no less physically bounded
than the English language. I see no difference between the existence
of the English language as historical objects, and the lambda calculus.
You seem awfully hung-up on the lambda-calculus being different.
I truly cannot see why.
John, are you trying to say that there's something about the
denotation of mathematical abstractions like "lambda calculus",
which does not apply to abstractions like "the english language"?
I don't see that at all.
Are you arguing (with Richard -) that the English language has some
Platonic ideal somewhere? That would be very unlike you.
>As to being at odds with decades of CS usage, I observe that when I did
>my computing, I noted that their usage was at odds with centuries of
>philosophical logic usage. I got used to the idea. But their usage is
>not authoritative.
Oh yes it is, when it comes to CS. Just as English speakers are
authoritative when it comew to English. Again: if you destroy all the
English speakers and readers, does that destroy English?
More to the point; if we destroy all speakers and readers and writers
of Ancient Egyptian, does that destroy the Ancient Egyptian language?
(Think carefully about Rosetta stones before you answer.)
>
>Fine, so what *is* the abstraction, ontologically speaking? I say that
>it is a matter of a relation between symbols and cognitive devices,
>which are physical.
Excuse me, mlud. I __think__ this is where I opt for crude
Australasian pragmatism, and to hell with ontology as `prior' to
knowlege of the real world. But, mlud, I must plead a one-week recess
to check with my brother, after his flight gets in :-)
Provisionally, though, and subject to revision (I am working in
another window, editing this between compiles, so I'm rather distracted):
I am happy with the idea of symbols. I lost track of what "it"
referred to. I have no idea what you mean by "cognitive devices" in
this context.
>> In case you didn't know: there are also definitions of
>> almost-but-not-quite Turing-complete languages with finite storage.
>> The language definition -- what you call the ``formal'' abstract and
>> not-concrete version -- defines, as part of the language, how an
>> executing program is notified that it's run out of storage (or just
>> about to run out, since the notification is done whilst there's still
>> enough space to do something about it).
>
>Nice, but not a counterinstance. Let's remain here with Turing-complete
>languages.
No, lets not. Lets stick with real-world concrete instances, since
that's where I disagreed with you.
>You are. (Sorry, you asked). This is the medieval universals debate
>begun by Roscellin, who argued that a universal (an abstract in our
>terminology) is mere breath "flatus vocus". I don't go this far, but it
>does have a history (actually, I think that history begins sometime
>between Aristotle and Porphyry, but is not preserved).
Scuse me, but having been round this a few times, I'm fairly certain
that *NOT* what I'm trying to say. I think you may be reading too much
into my repeated comparisons with the English language.
I have it on reliable authority (ie., a student of Kim's) that when
what I say when left to my own devies sounds a hell of a lot more like
Australasian pragmatism than the impression you seem to have gathered.
That pragmatism is, if anything, tempered by the work I did years ago
in AI, but that's another story.
Science does seem to be a pretty darn good way of learning about the
world (and of describing it). As for what philosophers
traditionally hold about the primacy of ontology -- well, as Sam Vimes
said: arseholes to the lot of 'em.
(This, I hear, is approximately what occasionally happens to my
brother in post-colloquium sessions. He puts forth his viewpoint, the
other students say, "oh, you're just an Australasian", and continue on
regardless. Then again, he may be egging them on.)
[A good hundred lines more, snipped as hoplessly off-topic, which
I can flesh it out via email if anyone really cares.]
John Wilkins
Okay, so you have a real or imagined problem with Matt. Take it
elsewhere. please? Apart from the fact that Matt is a friend of mine and
others here, the one thing I find most boring in t.o is the continual
posts of Did! Did not! and self-defense that gets run here, which is of
no interest to anyone but the combatants. One might call this the Nyikos
Motif.
When Matt says something that is essentially what I or someone else
says, then it deserves the same kind of treatment you'd give me or
someone else. Personalities are never as interesting in an ASCII medium
as the persons concerned think.
Jonathan Stone
Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
>Jonathan Stone <jona...@dsg.stanford.edu> wrote:
>> In article <3c3b...@news.qut.edu.au>,
>> Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
>>>Richard Harter <c...@tiac.net> wrote:
>>
>>>I agree with your side note, and am of the opinion that
>>>Lisp is commonly called a computer "language" when it is
>>>really a notation. (Same goes for C, and Cobol, and so on.)
>>
>> Chris,
>>
>> Is lambda-calculus a notation? How about Turing machines, or recursive
>> function theory?
>
>But anyhow -- no, I would not class lambda-calculus, or turing
>machines, or recursive function theory, as notation -- or as
>language for that matter. I would consider them mathematical
>models, and in each case there is a range of ways to express
>terms/machines/functions which is where the language or notation
>comes into it. But I still consider that formal languages are
>also another kind of mathematical abstraction, having the same
>kind of abstract existence as the lambda-calculus, etc.
Chris,
There is a point behind my question. One of the the point i keep
failing to get across to John is this:
For concrete, finite-sized instances (``utterances''?),
There is no difference between the lambda-calculus and a
suitably-circumscribed subset of Lisp, other than syntactic sugar.
Yet you and John want to place the finite utterances of Pure Lisp in
one ontological handbasket, and the finite utterances of
lambda-calculus in a different basket. Yet there is no difference in
semantics or manipulation, other than trivial differences in syntax.
[One has to "simplify" Pure lisp quite vigorously, but I'm simplifying
solely for the sake of exposition].
Just where do you see a difference?
>I guess I missed something, then. The terms of lambda-calculus,
>and the corresponding well formed sentences of Lisp, are equally
>unbounded in size.
Yes, of course. My reading is that John was (not to say "is") confused
between the finite limitations of any _implementation_ of Lisp (or
equally any utterance in Lisp), versus the definitional assumption
(explicit or implicit) of infinite storage, in formal notations like
UTMs or recursive functions or lambda-calculus.
>I disagree. Technically, the automaton is the 4-tuple, and the set
>of strings it accepts is the language.
Drat. Now I will have to check a textbook. But for me, that was a minor
detail.
The real issue was that you said langauges were *finite* sets... now,
I think we agree on both points I cared about: that languages can have
infinitely sentences as members, and that the sentences can be
arbitrarily long.
I hope that didnt' get lost in the formalisms.
>We can also speak of non-recursive languages, not accepted by
>any automaton, or even a Turing machine.
Uh... can we? How are such languages defined? Is this
semi-computability, or something completely different?
>Of course. I said a language was a set of finite strings, not a
>finite set of strings.
Oh. Sorry. I would have sworn you wrote the second.
(I'm not feeling so good myself)
>There are also automata accepting trees, (both finite trees and
>infinite trees) although it is not usual to speak of sets of
>trees as a "language".
Two-dimensional grammars? A good course in AI will toucnh on them
There was that MIT AI lab thesis on modelling how humans acquire
multi-column arithmetic, as a two-dimensional grammar.
A lovely piece of science. It went over real problem sheets done by
real primary-school children. It found the errors that the human
teachers were aware of -- adding carry back into the same column, that
kind of thing -- *and* it indetified new sets of systematic errors,
errors of which the teachers were previously unaware. L-somebody.
White cover with blue stripes. Must get a proper reference sometime.
[...]
>> ``Me too''.
>>
>>>Yes, Lisp is an abstract entity not present in space-time.
>>
>> The _definition_ of Lisp is abstract. Expressions in Lisp written on
>> some concrete media are no less and no more concrete than expressions
>> in English written on the same media.
>
>Yes.
>
>> Is English an abstract entity not present in space-time?
>
>Yes; though as above, a particular sentence spoken in
>English is another thing.
So far, we seem to be in agreement. (look slike Richard was right).
I am afraid John is reading more into my words than I am willing
to put there.
>> I confess to having known people who *speak* in Lisp, viz:
>>
>> one: foodp?
>> another one: nil
>> <time passes>
>> another one: foodp?
>> first one: t
>>
>> at which point they go off for lunch. I wish I was making this up,
>> but I'm not.
>
>CS people do tend to be a bit weird.
Um, acutally... these were non-CS people: professional programmers
with primarily non-CS backgrouns, who had drifted into programming in
a company doing AI. `Real' CS people don't talk like that. Well,
Apart from at Tech Square, I guess, where people ask who wants tea,
who wants coffee, and then count in binary on each hand...
[...]
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
>Well I must ask you then what I asked Jon - where does Lisp reside? Does
>it exist other than in its concrete implementations.
John,
What then is English? where does it reside other than in concrete
>> When I explain programming "language" to an introductory
>> programming class, I tell them that that "language" here is
>> a misnomer. A programming language is actually a mathematical
>> notation for expressing algorithms; and more closely related to
>> the notations of arithmetic than to real languages.
>>
>> Programming notations are used to express algorithms.
>>
>> Natural languages are used to argue, to make love, to
>> sway emotions, to convey knowledge, to lecture, ...
>
>Oh, absolutely. Nobody ever got laid by writing a COBOL program, I'm
>sure.
Oh dear. I'm afraid there really *are* more things, Wilkins,
than dreamt of in your philosophy.
>But I am aware that this is a matter of abstract objects like
>numbers. The debate began on the standing of numbers.
Ah. Now we're getting somerwhere.
What is it that two asteroids and two COBOL programmers, sorry,
angels, have in common? Did they have it in common before we
had a *name* for it?
How about if all things which have that property in common cease to
exist-- does the property cease to exist, too?
I *think* John may be confusing me with someone who confuses the
English word "two" and the Roman numeral 2, with the number two.
But that's not what I'm saying at all.
Matt Silberstein
jona...@DSG.Stanford.EDU (Jonathan Stone):
>In article <3spk3uod8n3sr84qt...@4ax.com>,
>Matt Silberstein <mat...@ix.netcom.com> wrote:
>>In talk.origins I read <a1db4t$gd6$1...@Pescadero.DSG.Stanford.EDU> from
>>jona...@DSG.Stanford.EDU (Jonathan Stone):
>
>Matt,
>
>
>It is a fact that any real-world computer does not have infinite
>storage. It is also a fact that we treat computers as real-world
>models of Turing machines. The finite storage of any real-world
>computer means that we can also treat the computer as a model of
>an *extraordinarly* large finite-state machine.
>
>But, given the the exponential explosion of nuimber of states in a
>finite-state machine to compute something as simple as *bit parity*,
>it is generally not a helpful or insightful model.
>
>
>The point you make about finite size might, in a very narrow sense, be
>correct. But it's a rather unhelpful insight.
If it did not help you then it did not help you. You do not define the
universe of potential audience members.
>To practicionters in
>the field, it's impossible to tell those who have just obtained that
>insight, from someone deliberately trying to be obnoxious (in this
>case, about two week after having gotten the insight.)
>
>I apologise sincerely if I mis-identifed your comment as being the of
>second kind, rather than the first. If you truly weren't intending to
>take a personal jab, then I apologize for responding as if you had.
>But that is how I saw it.
You thought that I had learned this two weeks ago? And was showing off
my knowledge? Jonathan, you astound me, you really do.
>PS: no, I don't criticize you for saying "true" rather than "valid".
>The criticism is for denying that it really *was* a mistake; moreover
>a mistake that, once you defended it, indicated you didn't have a clue
>about the subject you were pontificating on at the time.
Actually, if we are going to bring up old wars I would rather discuss
the system of my right hand and left thumb. Or the scientific
definition of life you claimed to have.
--
Matt Silberstein TBC HRL OMM LotL
There is safety in numbers and people and things
And big wads of money and great big diamond rings
J.O.
Matt Silberstein
jona...@DSG.Stanford.EDU (Jonathan Stone):
>In article <76998029.02010...@posting.google.com>,
>Matt Silberstein <mat...@ix.netcom.com> wrote:
>>jona...@DSG.Stanford.EDU (Jonathan Stone) wrote in message news:<a1b3e5$7qe$1...@Pescadero.DSG.Stanford.EDU>...
>>
>>[snip]
>>
>>> If you claim `design' is an algorithm, can you write down the
>>> algorithm?
>>
>>So what if he can't (currently) produce this algorith?
>
>Then for all practical purposers, there *is* no algorithm. And it is
>intellecutally dishonest (or, in this instance, if you prefer,
>Platonic) to proceed as if there was.
Not at all. It is entirely possible that someone could show that a
finite algorithm exists but that we don't currently know it. Or, one
could do what Bobby did, try to show that if it were an algorithm then
some other argument is wrong. His current ability is pretty much
irrelevant. He was not proposing to use the algorithm in any way.
Rather he was commenting on Dembski's work.
>Are you asking because you really don't *understand*?
No, I am asking because I disagree. I am asking you to make an
argument for why his current ability to produce the algorithm is
relevant. Asserting it is not such an argument. Nor is insulting
people who happen to disagree.
>>So what if it
>>will forever remain beyond any human's ability to produce this
>>algorithm?
>
>If you cannot state an algorithm, then what reason is there to believe
>that the collection of human intelligent process we call `design' is,
>in fact, an algorithm?
This is a more relevant question and different than the argument you
made above. The only argument he produced is intuition, not terribly
convincing. My own feeling, again not terribly authoritative, is that
the human brain is a machine and, as such, can only do what machines
can do.
>>I don't see how that says there is not some algorithmic
>>process behind design. His abilities and knowledge do not determine
>>what is or is not an algorithm.
>
>You really don't have a very good understanding of science. If Bobby
>wants to make a scientific argument that `design' is an algorithm,
>then the onus is on him to demonstrate that it *is* an algorithm.
He asked a question. And then proposed that if the answer were one
thing then we would have a particular result. What this has to do with
my knowledge of science escapes me. If you wanted to you could have
asked him to show it is an algorithm. But there are other ways to do
that in place of producing the algorithm.
>The most natural way to do so is to give the algorithm.
>This is a step Bobby is conspicuously avoiding.
That is one way, but not the only way.
>
>>> How is `design' any different in this respect than writing
>>> poetry, painting pictures, or sculpting scupltures, or writing music,
>>> or for that matter, doing science?
>>
>>A good question, but a good question does not an argument make. I
>>suspect that for the purpose of the NFL and this discussion they are
>>all sufficiently similar.
>
>I think you missed the point of the question -- which was to show
>that, scientifically, we don't have sufficient grounds to assume
>either way.
>>> If you will, for the sake of discussion, grant us the Church-Turing
>>> thesis, what you are asking is whether any of these activities can be
>>> produced or simulated by computers.
>>
>>Not quite. It asks whether they can be produced/similated by a
>>universal turing machine. No actual computer is an actual universal
>>turing machine. They all have finite limitations that the UTM lacks.
>
>I'm looking at this again, with the assumption that it's ill-informed
>rather than hostile.
>
>In that light, a better answer is: no, that's not how the standard
>terminology in the field works. Computers *are* models of UTMs,
>and we can approach a UTM to any arbitrary but finite size.
Yep, models of. Agreed. They are not UTMs, I hope we can agree on
that. I took your point to mean that since current computers don't
design, design was not an algorithm. You seemed to argue that real
computers people really have can only do these things in toy domains.
And that since they could not do them in real domains, that said
something about whether it was an algorithm. Appealing to the current
state of the art in computing does not tell us whether or not design
(or art or whatever) is an algorithm. That was why I made the point
about UTM vs computers. In the context you clearly meant real
computers that exist today running existent software.
[snip]
Chris Ho-Stuart
Right. I agree. Lisp = lambda-calculus + syntactic sugar.
I tend to use "language" to mean the syntactic syntactic
sugar involved in representing terms as strings, and do not
mind that "language" sometimes includes semantics as well, as
in a programming language like Lisp. And you can take terms in
lambda-calculus as being a kind of language as well, by extending
the meaning of "language" a bit beyond my normal usage. If you ask
about my normal usage, I would say that I normally take language
as sets of strings (and sometimes sets of strings plus semantics).
But I can fairly easily adapt and speak comfortably with colleagues
speaking of languages as sets of terms.
They are all mathematical abstractions, as stated in the
final sentence of my paragraph quoted above, and syntactic
sugar is no more to be found in space-time than the calculus.
We can write or speak utterances in space-time, and we can do
calculations in space-time; but the models (Lisp, and
lambda-calculus) are abstractions.
> Yet you and John want to place the finite utterances of Pure Lisp in
> one ontological handbasket, and the finite utterances of
> lambda-calculus in a different basket. Yet there is no difference in
> semantics or manipulation, other than trivial differences in syntax.
I am not sure, but I think John and I are using different baskets.
As stated above, I put both Lisp and lambda-calculus in
the same basket of mathematical abstraction not existing
in space-time. So I think I am rather closer to your
view than to John's; but await the definitive word on that
from Harter. :-)
> [One has to "simplify" Pure lisp quite vigorously, but I'm simplifying
> solely for the sake of exposition].
>
> Just where do you see a difference?
In the syntactic sugar. Not in the matter of concrete existence.
>>I guess I missed something, then. The terms of lambda-calculus,
>>and the corresponding well formed sentences of Lisp, are equally
>>unbounded in size.
>
> Yes, of course. My reading is that John was (not to say "is") confused
> between the finite limitations of any _implementation_ of Lisp (or
> equally any utterance in Lisp), versus the definitional assumption
> (explicit or implicit) of infinite storage, in formal notations like
> UTMs or recursive functions or lambda-calculus.
>
>>I disagree. Technically, the automaton is the 4-tuple, and the set
>>of strings it accepts is the language.
>
> Drat. Now I will have to check a textbook. But for me, that was a minor
> detail.
>
> The real issue was that you said langauges were *finite* sets... now,
> I think we agree on both points I cared about: that languages can have
> infinitely sentences as members, and that the sentences can be
> arbitrarily long.
No; I said they were sets of finite strings; but you do
recognize this below in portions of your post I have snipped.
> I hope that didnt' get lost in the formalisms.
>
>>We can also speak of non-recursive languages, not accepted by
>>any automaton, or even a Turing machine.
>
> Uh... can we? How are such languages defined? Is this
> semi-computability, or something completely different?
We could define a language to be the set of all strings
NOT accepted by a given Turing machine. Some languages in
this class are not recursively enumerable: it is the class co-re.
In fact, this is just the bottom of the "Arithmetic Hierachy",
with successively larger classes of languages getting more and
more uncomputable. This hierachy is known not to collapse.
Let S0 be the class of re-languages.
Let P0 be the class of co-re-languages.
Let D0 be their intersection (which is the class of primitive
recursive language).
We get to the next level of the hierarchy by using turing machines
augmented with an oracle that solves problems in the previous
levels of the hierarchy. Such machines are not physically
realisable, of course.
More usually, this hierachy is defined in terms of acceptance
by a formula which is a sequence of quantifiers followed by a
fully deciadable predicate. The height up the hierarchy
corresponds to the number of alternations in quantifiers
required. This works using Goedel numbers for strings.
For example: let COF (co-finite) be the set of goedel numbers
for turing machines that accept all but a finite number of
inputs.
COF = { M | exists n @ forall x @ exists t @
"len(x) <= n OR machine M accepts x within t steps" }
This language is in S3 of the arithmetic hierarchy (3 quantifiers,
ending in exists).
Reference (from which I cribbed this definition)
<http://www.cs.cornell.edu/Info/Courses/Spring-97/CS682/l34-ah.ps>
[snip remaining lots of agreement]
Cheers -- Chris
Jonathan Stone
Matt Silberstein <mat...@ix.netcom.com> wrote:
>In talk.origins I read <a1fmnu$sad$1...@Pescadero.DSG.Stanford.EDU> from
>jona...@DSG.Stanford.EDU (Jonathan Stone):
>>>So what if he can't (currently) produce this algorith?
>>
>>Then for all practical purposers, there *is* no algorithm. And it is
>>intellecutally dishonest (or, in this instance, if you prefer,
>>Platonic) to proceed as if there was.
>
>Not at all. It is entirely possible that someone could show that a
>finite algorithm exists but that we don't currently know it.
Wrong. First they have to show that the collection of processes
lumped under hte rather vague and general English word `design'
*is* an algorithm, rather than (as don Knuth once put it)
an Art, rather than a science.
[...]
>No, I am asking because I disagree. I am asking you to make an
>argument for why his current ability to produce the algorithm is
>relevant. Asserting it is not such an argument.
Unless Bobby can show an algorithm, and show that the algorithm *is*
what we call `design', he(?) cannot establish that the collection of
human procesases *is* in fact that algorithm, or indeed that it is
any `algorithm' at all.
That is precisely what Alan Turing's original Turing test is all
about: a experimental protocol to decide whether computers (i.e.,
algorithms, algorithms in execution) can pass as human to other
humans, to the point where a human conversing iwth a compute and
with a human cannot tell which is which.
[...]
>>If you cannot state an algorithm, then what reason is there to believe
>>that the collection of human intelligent process we call `design' is,
>>in fact, an algorithm?
>
>This is a more relevant question and different than the argument you
>made above.
Excuse me. I thought it was the *same* argument, only made explicit
rather than likening `desing' to other humn intellectual and creative
endeavours, which are generally regarded as being non-algorithmic.
>He asked a question. And then proposed that if the answer were one
>thing then we would have a particular result. What this has to do with
>my knowledge of science escapes me.
I read Bobby's text as more of a rhetorical device device. I read it
as appealing to intution that `design' really *is* an algorithm.
Perhaps I was mistaken. But that is certainly how the argument
seemed to continue.
If Bobby wants to be scientific, then yes, I think the onus is on
Bobby to show that `design' really *is* an algoirthm; and to abandon
any rhetorical devices which make the if-then, well, rhetorical rather
than substantive. If you disagree with that, then yes, I think that
does reflect on your ounderstanding of science.
> If you wanted to you could have
>asked him to show it is an algorithm. But there are other ways to do
>that in place of producing the algorithm.
Really? Like what?
>>The most natural way to do so is to give the algorithm.
>>This is a step Bobby is conspicuously avoiding.
>
>That is one way, but not the only way.
Again, like what? We have a fairly large set of human activity lumped
under the very general term `design'. We have the hypothesis that
this activity is an algorithm. The only way *I* know to show that is
to give some algorithm, then demonstrate that the process of
excecuting the algorithm, and the human behaviours we call `design',
really are the same.
What other ways did you have in mind?
[...]
>>I think you missed the point of the question -- which was to show
>>that, scientifically, we don't have sufficient grounds to assume
>>either way.
>>>> How is `design' any different in this respect than writing
>>>> poetry, painting pictures, or sculpting scupltures, or writing music,
>>>> or for that matter, doing science?
>>>
>>>A good question, but a good question does not an argument make. I
>>>suspect that for the purpose of the NFL and this discussion they are
>>>all sufficiently similar.
Me too, but neither your suspicions nor mine really cut it.
As far as I know, attempts to reduce painting and storytelling to
algorithms haven't gone much beyond the ancient 'spew' program, which
generated fake national-enquirer headlines and ... What was the other
version? Candidates for `Dukes of Hazzard' plotlines? ... by plugging
phrases from a dictionary together. I once got the output
``Evil aliens are the parents of my love child, says Joan Rivers--
exclusive pictures show all!
which doesn't bode terribly well for Bobby's hypothesis. :)
>>I'm looking at this again, with the assumption that it's ill-informed
>>rather than hostile.
I hope John w. has caught up with this one.
>>In that light, a better answer is: no, that's not how the standard
>>terminology in the field works. Computers *are* models of UTMs,
>>and we can approach a UTM to any arbitrary but finite size.
>
>Yep, models of. Agreed. They are not UTMs, I hope we can agree on
>that. I took your point to mean that since current computers don't
>design, design was not an algorithm. [continue below]
No. I'm saying that there is no *demonstration* that `design' is an
algorithm; that the current research is not even *beginning* to
automate many of the sorts of things humans do that we call design;
and that therefore, there is no reason to put much stock in the
hypothesis that what we call design *is* expressible as an algorithm.
[Matt contiues on the same line]
>You seemed to argue that real
>computers people really have can only do these things in toy domains.
The reference to "toy domains" was __completely_ separate. I've
probably said here, several times, what i think about AI: that AI is
what get gets done in AI labs, and once we figure out how to make
computers do something, its no longer part of the province of AI. (I
invented that independently, when asked in an AI class to define what
AI was. I used to think that was a wry joke, until I heard other AI
professors saying it. I can give names if you really want).
Hey, Matt, look. The way I read this, it does seem that you mixed up
several rather distinct points, from separate posts, into one
mish-mash; and then attributed the mish-mash to me.
I'm *trying* to be nice. But you seem to have a real talent for taking
things I've said,mashing them, and repeating the mash being what i
*acutally* said. I will take it on faith that you really *think*
that's what I said. Well, I didn't, and I find the misrepresntation
damn' offensive.
>And that since they could not do them in real domains, that said
>something about whether it was an algorithm.
No, I never said any such thing. And again, I find your paraphrase and
its attribution to me, to be really, really, *really* offensive.
My point was quite explicitly about `toy domain', and how that
referred more to AI domains which were well-enough understood that
they become problems in textbooks; and more, that once a problem
becomes a `toy domain', it stops being AI (_sens Jonathan_).
I don't know *why* you misread me so grossly, but I really *do* find
it offensive. Also, I don't see John Wilkins making misunderstandings
like that and attributing them to me. I really don't.
> Appealing to the current
>state of the art in computing does not tell us whether or not design
>(or art or whatever) is an algorithm.
No, it does not. but it does reflect the current scientific state of
the art on such subjects. In the absence of an algorithm-- and
Bobby's understandable reluctance to produce one -- I don't see what
else we can or should consider.
I was not speaking ot the _truth_ of the proposition; as I said, the
only way I know to establish that is to invent an experimental
protocol -- c.f. the Turing test-- and then proffer an algorithm.
Instead, I was speaking to what we should consider as the default
working hypothesis. That's why I offered painting fine art, and
poetry, and those other things: things which we don't usually think of
as algorithms. Or at least, *I* don't.
I assumed the audience would read it that way. Guess I made a mistake.
>That was why I made the point
>about UTM vs computers. In the context you clearly meant real
>computers that exist today running existent software.
Yes, I did mean real computers. I dont know of any other kind of
computers -- _mutatis mutandis_, that is, as technology changes
for different points in time as `today'.
I'm with Chris and John on that one. Until about the late 1950s,
`computer' was a job title for humans using calculating machines.
Since then, it means computing machines which (modulo finite
storage space) are Turing-complete computing devices.
I'm not in the habit of referring to as-yet-uninvented computers, or
imaginary computers like Douglas Adam's `Deep Thought'.
Marty Fouts
john.w...@bigpond.com (John Wilkins) writes:
> I can see that codeine-addled[*] attempts at puns and philosophical
> jokes are just confusing the issue. So I will attempt to put my case a
> bit more carefully, with no smileys...
>
> Some basic definitions:
>
> A formal language is, for this purpose, a language without ambiguity,
> amphiboly, or contradiction, which is fully coherent and formalised.
I came late to this party, so forgive me for asking what may be obtuse
questions: in what sense do you use 'coherent'?
> A natural language is one which can be ambiguous, amphibolous, or
> contradictory, and may be unformalised in its syntax or symbolism, and
> may also be incoherent.
Is amphiboly here meant to construe something other than merely
'ambiguous sentence'? (Does it matter or am I just being pedantic?)
Anyway, here is an alternative set of definitions:
language: "form of speech used by a group of people with a unique set
of vocabulary combined in uniform patterns (syntax) to convey meaning
(semantics)."
"natural language": A human language not specifically designed for
communicating instructions to a computer.
"formal language": a set of strings from an alphabet (alphabet =
finite set of symbols, symbol = an abstract entity having no meaning)
note that 'natural language' is a subset of 'language' but that
'formal language' is a definition from a different domain of
discourse. (Language and natural language are terms from linguistics,
formal language is a term from mathematics.) Although, one can pretty
much bend the rules and make language, natural language, and computer
language all subsets of formal language.
> I'm sure these are not good definitions, but I'm not completely
> defining anyway. Let's say these are facets of natural and formal
> languages.
>
> On this account, Lisp is a formal not a natural language, as you
> say.
Being pedantic, I will note that all computer programming languages
fail your definition of formal, as even the most precisely defined
contain ambiguity.
> So I will coin a couple of different defining terms to make it clear
> what I mean. But hold on to that word "natural", because I'll get
> back to it, if only to explain where I thought the obvious pun
> was...
>
> A concrete object is one that is bounded by spacetime (this is
> Zalta's definition), and an abstract object is one that is not nor
> can be. A universal property is therefore abstract, and a particular
> property is concrete.
Ih this model are subjective properties, such as color, concrete,
because existing in the mind of the observer they exist in spacetime,
or abstract, because differing from observer to observer they are not
fixed?
> Call this nominalism, if you like, so long as it is understood that
> this is a nominalism relative to some domain.
>
> So let's call a "concrete language" any language which exists in
> spacetime, and an "abstract language" one that cannot. Lambda
> calculus is an abstract language. Lisp is a concrete language, where
> implemented (and abstract where not). You might call this the
> "concrete interpretation" of Lisp, or the instantiation of Lisp, if
> you like.
Someone get me a pipe. [reference to Magritte or Freud, your choice.]
Um, you're eliding what it means for 'language' to exist in
spacetime. Is English concrete, because of the OED, or abstract,
because there is no 'implementation'. If its not abstract then how
does it differ from Lambda calculus, since mathematics is, after all,
merely natural language with more precise symbolism?
> So, now we have a distinction between a physical (ie, bounded)
> object - Lisp on Chaitin's Solaris, or on John Wilkins's Mac or
> whatever, which has all kinds of limitations (in terms of pointers
> available at runtime, implemented handles, and the like) - and the
> abstract *definition* of Lisp in whatever standard it is
> defined.
X3.226-1994. I could look up the ISO equivalent if it would help.
But I think there's a Chinese room around the corner. If
implementation on a computer counts as concrete, then does a computer
program that does voice recognition and text-to-speach generation make
English concrete? If so, and I teach the program to 'speak' lambda
calculus notation, how does that differ?
> This is a metaphysical definition. It doesn't affect the
> standing of Lisp WRT lambda calculus or Gödelian arithmetic, &c. It
> does, however go to the question of how any formal language is
> implemented (your nice blackboard example). Chalk marks on a
> blackboard are not Lisp. Nor are ink marks on a manual, even if the
> symbols represent a reverse polish notation definition of the
> language.
If the symbols don't make the language concrete when they exist on a
blackboard, why do they make it concrete when they exists as ASCII
text in a computer? In particular, is the lisp program file on my
laptop, which has no lisp interpreter concrete or not?
> Where does Lisp exist if not in concrete form (even the abstract
> definition must exist in concrete form)? If you are not a Platonist,
> you must answer this way.
My head hurts now. I think i'm getting the talk.origins virus. So far,
all of the comments have really been about utterances in Lisp,
(programs) rather than Lisp itself. Does the computer running a lisp
interpreter "understand" lisp, or is it just a Lisp room? What does
'understand' mean, anyway?
[snip]
> What naturalism is trying to do, is to establish an account of
> something procedural and prescriptive in terms of being concrete,
> rather than abstract. So when I said that Lisp was a "natural"
> language, I merely meant that any instantiation of Lisp must be a
> concrete instantiation. There: if Harter has the most indirect
> puns, at least they elicit a groan. Mine are merely needlessly
> obscure.
>
[snip]
well at least your's are merely obscure and you understand them. mine
are obtuse and even I don't get them.
Marty
Chris Ho-Stuart
[snip]
> No. I'm saying that there is no *demonstration* that `design' is an
> algorithm; that the current research is not even *beginning* to
> automate many of the sorts of things humans do that we call design;
> and that therefore, there is no reason to put much stock in the
> hypothesis that what we call design *is* expressible as an algorithm.
I am going to regret stepping into this....
You express the above as a syllogism.
The form is
Since "current research is not even beginning to automate design"
------------------------------------------------------------------
Thus "there is no reason to put much stock on the hypothesis that
design is expressible as an algorithm"
For this syllogism to be valid (apart from any question about
whether or not the premise or conclusion is true) you need to
show that the only kinds of "reason to put stock in something"
are being able to actually make a plausible start on automating
design. Otherwise your conclusion does not follow.
However, I certainly think there are reasons (which I find
persuasive, though they are not conclusive, and they are not in the
form of a demonstration) for putting some considerable stock on the
hypothesis that design is expressible as an algorithm; or better,
that there are algorithms which would exhibit behaviour that is, to
all intentions and purposes, design. (This could be seen as a subset
of the Turing test.) My reasons have nothing to do with current
state of progress in actually implementing design as an algorithm.
I can spell those reasons out if you like, though that is not my
main interest in commenting here. I don't want to set the precedent
for insisting that someone must address every possible criticism
of a line of argument in advance, on pain of being subject to the
kinds of rather aggressive rhetoric seen in the thread.
Also rather intriguing is a contrast with our other discussion, in
which we have considered that the space of algorithms representable
in formalisms like lambda-calculus is unbounded. One would have to
concede that that proportion of the space of all algorithms which
we have begun to explore with current research is rather small...
and hence not necessarily a useful indicator of what algorithms can
and cannot do. I prefer considerations that have a more general scope,
far beyond that tiny part of the space of all algorithms which we have
so for managed to implement.
Also interesting is that our current implementations of algorithms
are of a certain limited form: mostly variations on deterministic
Von Neuman machines. We can write algorithms for quite different
formalisms; but the actual implementations tend to involve in
practice a mapping to Von Neuman machines of some kind. By putting
too much stock on this particular concrete form, you seem (as far
as I can see) to be straying into John Wilkin's camp... :-)
Cheers -- Chris
Richard Harter
Stone) wrote:
>In article <3c3b7294...@news.SullyButtes.net>,
>Richard Harter <c...@tiac.net> wrote:
>>On 8 Jan 2002 15:09:12 -0500, jona...@DSG.Stanford.EDU (Jonathan
>>Stone) wrote:
>>
>>
>>You seem to be working very hard at not reading what John has to say.
>
>I dont think so.
But then, you wouldn't think so, don't you know.
>I think I'm trying to convey that comptuer
>scientists put the boundaries in a different place than John's
>ontology woudl have them fall.
Well, perhaps, but computer scientists tend to be dreadfully naive
about things like ontology.
[snip sundry]
>>- or perhaps not; there are some very large and very
>>ancient boojums in this territory. In any event, your appeal to
>>English usage in CS departments (if there be such) is an appeal to
>>irrelevant authority.
>
>If you think that, then I owe you and John an apology. The appeal is
>not to authority, but again, to clear communication.
Apologize first without qualification; it's good for the soul. Don't
argue, just do it.
There is a real problem about this notion of clear communication which
is directly relevant here. Language as it is used necessarily embeds
all sorts of assumptions; quite naturally these presumptions can
become unquestionable because they are invisible.
One of the areas where this comes into play is in discussions of
ontology. President Clinton caught a good deal of flak for his
"depends on what the meaning of 'is' is". The word "is", particularly
as it is (shudder) used in English is used in a surprisingly wide
variety of ways. There are similar problems with words like "real".
Many people in the CS world (not you, of course, or anyone present)
tend to be very dogmatic. It's a natural reaction. In CS one is an
intellectual environment which is sharply delimited with hard clear
lines. However the sharp intellectual clarity of CS in some ways is
an illusion.
In the early years of the past century there was rather a lot of
discussion of the foundations of mathematics. There were several
distinct schools of thought as to the nature of mathematical truth -
platonism, intuitionism, formalism, and so on. [Bear with me, I'm
sure that this is old stuff to you but it is not to others.] One of
the viewpoints was the formalist approach which said, roughly, that
mathematics is a game played by making uninterpreted marks on pieces
of paper according to specific rules. A great merit of the formalist
approach is that it provided an agenda for a broad research program
that has led by various routes to computer science, UTM's, ATM's etc.
In the course of implementing this research program the original
pieces of paper with the pencil marks have vanished. That is, they
haven't physically vanished. They and their equivalents, the
blackboards and whiteboards, and mag tapes, and CD's, and rams and
roms, and flaishing lights are all physically present. In many
mindsets they are mentally invisible; the mind's eye looks at the
chalk marks and sees a lisp program or statement in the lambda
calculus. In short, people look at physical things and "see" the
abstractions they have been trained to see.
One can take the view that these abstractions are no more than
convenient fictions, that the chalk on the board, et cetera is what is
real. (I will leave it to John to say for himself if the current
remarks are in any way related to his thoughts.) Quite often
abstractions are eliminable, i.e., we can replace references to atoms
and such like by chains of circumlocutions referencing physical
observations. Similarly we can explore the view that language
referencing abstract machines and such like can be systematically
replaced by physical references. As a practical matter we shall go
happily on talking about electrons, atoms, the law of universal
gravitation, and universal Turing machines just as though they exist.
It is ever so much more convenient to do so. Still, we can explore
the view that the existence referred to are so many empty words.
There even is some profit in doing so. For one thing it is very easy
for empty abstractions, i.e., abstract terminology that is not
eliminable, to creep into our discussions. For another thing these
abstract engines of thought imply claims about the universe that are
not true. For example, one of the Peano axioms asserts that every
number has a successor, i.e., for every number x there is a number y
such that y=x+1. In other words given a number x we can add 1 to it.
There is a definite sense in which this is not true. The integers
which we and our artifacts have referenced or will ever reference is
finite; inevitably there has been or will be integer which are
referenced but whose successor will never be referenced. Paradoxically
we will never know what these integers are. In short, the statement
that we *can* add 1 to every integer is false. A good platonist (as
the term is used in the philosophy of mathematics) insists that the
successors (and the whole infinite suite of integers) exists
willy-nilly. This view is supportable if the word "exists" is given
different meanings according to context.
>
>I'm not sure why, but you seem to be working very, very hard to ingore
>my original point: where John placed his demarcation point is simply
>*wrong*. not as a matter of philosophy, or of English usage, but as a
>matter of simple fact. Either you or John can visit a grad-level
>programming class, and see both Lisp evaluation and lambda-calculus
>reduction being done on whiteboards. The whiteboards are, indeed,
>concrete and finite.
>
>I'm damned if I can see how one is "concrete" but the other isn't.
>And that was -- if you recall -- what John was claiming early on:
>that Lisp was a "Pnatural language" but lambda-calculus wasn't.
But that wasn't what John was claiming. That is the problem; he
simply wasn't saying what you say he was saying. I'm not going to
argue with about he said or meant though - John can speak for himself.
>
>
>Yes, there are ancient boojums here. John is trying to fence them in.
>I think the problem is that -- as with the ancient Greeks -- Church
>and Turing movedx the fences from where Johns, ahhh, idealistic
>position would have them lie.
>
>(Yes, thats a joke. I only hope John gets a laugh; flu in midsummer is
>no fun at all.)
>
>
Richard Harter, c...@tiac.net,
Richard Harter
<host...@sky.fit.qut.edu.au> wrote:
[snip a lot of stuff that Jonathan answered very well]
>> Exactly where would this notation be? What is the nature of its
>> existence? How can these finite transient events, chalk marks on a
>> blackboard and the like, be coupled to those abstractions existing
>> outside of time and space? Plato wants to hear your answer.
>
>The notation is not in space-time; asking "where" it is does
>not seem to me to be a sensible question. I consider that transient
>events are coupled to abstractions by our action in agreeing
>to some meaning or interpretation for the notation.
>
>I'm no philosopher and would be unable to decribe the problems
>Plato was worried about. but I consider that Lisp is in the same
>place as lambda-calculus. The difference between lambda-calculus
>and Lisp is that Lisp incorporates more concrete syntax; but that
>remains an abstraction as well.
I babbled about this at greater length elsewhere for which see. You
see, though, I hope, that you are using words such as "is" and
"exists" in ways that are very different. Paris exists. (I have my
doubts about South Dakota.) To say that "Lisp exists" says something
very different from "Paris exists". What is more it is very hard to
pin down what you (or I or anyone) means by saying "Lisp exists".
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 19:09:30 -0500, jona...@DSG.Stanford.EDU (Jonathan
>Stone) wrote:
>Well, perhaps, but computer scientists tend to be dreadfully naive
>about things like ontology.
I think I have to give you aclean miss on that one. Have I failed to
mention that my brother is a PhD candidate in philosophy, that my
mother has a masters in philosophy, and that I spent Friday evenings
for some years drinking wiht a philosopher of John's acquaintance?
I've been red-eyed for about 3 weeks. I have an _exceptionally_ low
tolerance for GAs -- which is not Bobby's fault at all.
I am sure no-one *he* works with claims that GAs give comparable TSP
results, with comparable compute time, to algorithmic methods,
forgetting to highlight gthat the GAs ran on machines 20x faster.
bnut this Does Happen, as they say; and my hostility is not nearly so
far from the norm as some here seem to think. And I havent even
mentioned the stories of Koza's students, who tried ot do GP in
the same machine-learning domain where I did my AI.)
>Apologize first without qualification; it's good for the soul.
> Don't argue, just do it.
But what is it I should be apologizing to you and John *for*?
>In the early years of the past century there was rather a lot of
>discussion of the foundations of mathematics.
[[[ snip lots -- i am a-Frege you are going to Russel through places where I
can only Cantor.]
>In short, people look at physical things and "see" the
>abstractions they have been trained to see.
Ah, now I see I shouldn't have snipped my digressions after all.
No smiley.
>One can take the view that these abstractions are no more than
>convenient fictions, that the chalk on the board, et cetera is what is
>real.
I think you mean `fictions, that _only_ the chalk on the board is real'
[...]
> As a practical matter we shall go
>happily on talking about electrons, atoms, the law of universal
>gravitation, and universal Turing machines just as though they exist.
>It is ever so much more convenient to do so. Still, we can explore
>the view that the existence referred to are so many empty words.
Richard, this is surreal. I've been teh one wishing to talk about
the concerete writings, and the interpretations humans give them.
I have studiously avoideding saying that the abstractions are Forms,
or platonic Ideals, or any such thing. have *also* studiously avoided
saying that they have no existence other than as the utterances.
<Granny>see, I dont hve no truck with *either* view.<\Granny>
If apologies are going around as good for the soul, perhaps you and
John owe me one for casting me into the other camp, simply 'cause I'm
not in the one (I won't even ask to take it in kind against some of
the ones I indisputably owe Matt and Bobby Bryant.)
[...]
>There is a definite sense in which this is not true. The integers
>which we and our artifacts have referenced or will ever reference is
>finite; inevitably there has been or will be integer which are
>referenced but whose successor will never be referenced. Paradoxically
>we will never know what these integers are. In short, the statement
>that we *can* add 1 to every integer is false. A good platonist (as
>the term is used in the philosophy of mathematics) insists that the
>successors (and the whole infinite suite of integers) exists
>willy-nilly. This view is supportable if the word "exists" is given
>different meanings according to context.
I thought even philosophers of mathematics called that neo-Platonism.
I guess I learnt something; thank you.
>>I'm damned if I can see how one is "concrete" but the other isn't.
>>And that was -- if you recall -- what John was claiming early on:
>>that Lisp was a "Pnatural language" but lambda-calculus wasn't.
>
>But that wasn't what John was claiming. That is the problem; he
>simply wasn't saying what you say he was saying. I'm not going to
>argue with about he said or meant though - John can speak for himself.
On this, sir, I sincerely beleive you are mistaken. This (the t.o)
discussion didn't start about numbers. it started about John's mention
of "natural" languages. If you will allow me to skip over the
distinction between a `language' and utterances in that language,
since that's what John's first two or three posts did.
John also said, later that Lisp was a natural language
[_sensu_Wilkins], but that lambda-calculus was not.
I will check, as soon as I can.
Marty Fouts
c...@tiac.net (Richard Harter) writes:
[snip]
> I babbled about this at greater length elsewhere for which see. You
> see, though, I hope, that you are using words such as "is" and
> "exists" in ways that are very different. Paris exists. (I have my
> doubts about South Dakota.)
Both Paris and South Dakota exist in the only way that matters: I have
eaten in them. New York city fails to exist, although New York state
does not.
> To say that "Lisp exists" says something very different from "Paris
> exists". What is more it is very hard to pin down what you (or I or
> anyone) means by saying "Lisp exists".
Lisp exists in the same way that the movie _PI_ exists. It's yet
another thing that the laughing ghod will have to remeber to include
when it gets around to inventing the dreamer to dream us.
Marty
Chris Ho-Stuart
> On 8 Jan 2002 19:44:52 -0500, Chris Ho-Stuart
> <host...@sky.fit.qut.edu.au> wrote:
>
> [snip a lot of stuff that Jonathan answered very well]
>
>>> Exactly where would this notation be? What is the nature of its
>>> existence? How can these finite transient events, chalk marks on a
>>> blackboard and the like, be coupled to those abstractions existing
>>> outside of time and space? Plato wants to hear your answer.
>>
>>The notation is not in space-time; asking "where" it is does
>>not seem to me to be a sensible question. I consider that transient
>>events are coupled to abstractions by our action in agreeing
>>to some meaning or interpretation for the notation.
>>
>>I'm no philosopher and would be unable to decribe the problems
>>Plato was worried about. but I consider that Lisp is in the same
>>place as lambda-calculus. The difference between lambda-calculus
>>and Lisp is that Lisp incorporates more concrete syntax; but that
>>remains an abstraction as well.
>
> see, though, I hope, that you are using words such as "is" and
> "exists" in ways that are very different. Paris exists. (I have my
> very different from "Paris exists". What is more it is very hard to
> pin down what you (or I or anyone) means by saying "Lisp exists".
And similarly for English, I presume?
But in fact, I have never said "Lisp exists". In fact, the only
two times so far this year I have used the word "exists" in posts
was where I wrote out a first order logical formula including two
existential quantifiers. I accept that this is *very* different
from how I use "is".
Oh wait: I said "Lisp exists" just above. Oh drat, I did it again.
I honestly don't see your problem. I did not say Lisp exists; I
simply attempted to answer your question about where Lisp was, which
struck me as a silly question. Lisp is not a something in space-time
at all: all we have in space-time are sentences of Lisp. I and some
of my friends can recognize a scribble of chalk marks as a Lisp
sentence: and so you can take "Lisp" as an adjective if you like.
We can do some abstact reasoning about all possible Lispy sentences
but this abstract collection is not a thing in spacetime either.
And furthermore...
When my friends and I see a Lispy sentence, we have reached an
agreement on something we can with it to it, resulting in a new
Lispy sentence which we can go ahead and bring into real existence by
writing it on a chalk board as well. (We call this game "evaluation"
or "reduction" -- and the game can be found in the same place as
you might find chess.)
So Lisp is also a verb. (And since "Lisp" is a verb, it is also a
word, which makes it a noun. But I think you are not playing that
little game, which Jonathan called you on as well, in slightly
different terms with respect to "two").
When I have referred to Lisp as syntax plus semantics in other
posts, you can take this as a first step for a dictionary entry
for the word Lisp: as an adjective plus a verb.
Cheers -- "Chris"
Marty Fouts
c...@tiac.net (Richard Harter) writes:
[snip]
> In the early years of the past century there was rather a lot of
> discussion of the foundations of mathematics.
Early years? Kline, rest his sould, was still writing well into the
'80s. ;) Mathematics: The Loss of Certainty was published in '82,
IIRC, and is still in print.
More realistically, there are still world conferences on the
philosophy of mathematics.
> There were several distinct schools of thought as to the nature of
> mathematical truth - platonism, intuitionism, formalism, and so on.
Are. It's been less than three days since the last argument I had with
a strict constructionist.
> [Bear with me, I'm sure that this is old stuff to you but it is not
> to others.] One of the viewpoints was the formalist approach which
> said, roughly, that mathematics is a game played by making
> uninterpreted marks on pieces of paper according to specific rules.
> A great merit of the formalist approach is that it provided an
> agenda for a broad research program that has led by various routes
> to computer science, UTM's, ATM's etc.
Indeed. This is pure mathematics. All else is 'mere application' ;)
> In the course of implementing this research program the original
> pieces of paper with the pencil marks have vanished. That is, they
> haven't physically vanished. They and their equivalents, the
> blackboards and whiteboards, and mag tapes, and CD's, and rams and
> roms, and flaishing lights are all physically present. In many
> mindsets they are mentally invisible; the mind's eye looks at the
> chalk marks and sees a lisp program or statement in the lambda
> calculus. In short, people look at physical things and "see" the
> abstractions they have been trained to see.
Some of them do it without the physical representations. (No, I'm not
jealous. Not even a little.)
> One can take the view that these abstractions are no more than
> convenient fictions, that the chalk on the board, et cetera is what
> is real. (I will leave it to John to say for himself if the current
> remarks are in any way related to his thoughts.)
So if Fermat really did have the proof that wouldn't fit into the
margin (and thus never wrote it down anywhere) is it a convenient
fiction or a real abstraction? And does the fact that we don't know
what it was change its reality and existence?
> Quite often abstractions are eliminable, i.e., we can replace
> references to atoms and such like by chains of circumlocutions
> referencing physical observations. Similarly we can explore the
> view that language referencing abstract machines and such like can
> be systematically replaced by physical references. As a practical
> matter we shall go happily on talking about electrons, atoms, the
> law of universal gravitation, and universal Turing machines just as
> though they exist. It is ever so much more convenient to do so.
> Still, we can explore the view that the existence referred to are so
> many empty words.
> There even is some profit in doing so. For one thing it is very
> easy for empty abstractions, i.e., abstract terminology that is not
> eliminable, to creep into our discussions. For another thing these
> abstract engines of thought imply claims about the universe that are
> not true. For example, one of the Peano axioms asserts that every
> number has a successor, i.e., for every number x there is a number y
> such that y=x+1. In other words given a number x we can add 1 to
> it. There is a definite sense in which this is not true. The
> integers which we and our artifacts have referenced or will ever
> reference is finite; inevitably there has been or will be integer
> which are referenced but whose successor will never be
> referenced.
You're assuming facts not in evidence, that is, that time has an upper
bound, at least so far as intelligence is concerned. Were that
statement not true then your observation does not hold.
> Paradoxically we will never know what these integers are. In short,
> the statement that we *can* add 1 to every integer is false.
I think you mean 'will' rather than 'can'. There is no integer you can
name that I can't add one to.
> A good platonist (as the term is used in the philosophy of
> mathematics) insists that the successors (and the whole infinite
> suite of integers) exists willy-nilly. This view is supportable if
> the word "exists" is given different meanings according to context.
Somehow I feel like bringing up Quine's indispensibility argument at
this point, but I think I'll dispense with that idea.
(http://plato.stanford.edu/entries/mathphil-indis/ for those who care)
[snip]
> >Yes, there are ancient boojums here. John is trying to fence them
> >in. I think the problem is that -- as with the ancient Greeks --
> >Church and Turing moved the fences from where Johns, ahhh,
> >idealistic position would have them lie.
> >
> >(Yes, thats a joke. I only hope John gets a laugh; flu in midsummer is
> >no fun at all.)
You have no shame sir. ;)
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 19:09:30 -0500, jona...@DSG.Stanford.EDU (Jonathan
>Stone) wrote:
>>In article <3c3b7294...@news.SullyButtes.net>,
>>Richard Harter <c...@tiac.net> wrote:
>>>On 8 Jan 2002 15:09:12 -0500, jona...@DSG.Stanford.EDU (Jonathan
>>>Stone) wrote:
>>>
>>>
>>>You seem to be working very hard at not reading what John has to say.
Let's skip a huge amount. Richard says that
>But that wasn't what John was claiming. That is the problem; he
>simply wasn't saying what you say he was saying. I'm not going to
>argue with about he said or meant though - John can speak for himself.
I said I'd check, and I did, and I think Richard is mistaken.
>> [...]
ethis started when John Wilkins posted, in answer to one of my posts to
Tim Tyler's very spendid and worthwhile interpretation of irreversibility[*]
JW> This is a matter of philosophical debate. However, on this matter I
JW> agree with you. For me, information is a semantic construct in a natural
JW> language,
Jw>
[*] See: _the hitch-hikers' guide to the Universe, Douglas Adams,
BBC radio play, in the speech given as Arthur Dent's village
is being demolished to make way for a bypass.
I repsonded:
JS> <splork> _Natural_ language?
John responded right back:
JW>Yep (I'm glad you got this). The *existence* of information is
JW>anecessarily in a natural language. This is because no formal languages
JW>actually exist. That is to say, you cannot (in my ontology) assert
JW>
JW>hereexists(x)(Lx)
JW>
JW>Where Lx is "x is a formal language", because, as I never tire of
JW>saying, abstract objects do not exist (in the physical universe). So if
JW>information exists, it is in a natural language, while if it is a
JW>property of a formal language, it does not exist. Such information as
JW>exists is in a natural language :-)
Which is the smiley I ignored, because as the later context of the
post (and later the thread) makes it clear John is fairly serious in
point, if not about the term.
John offers some definitions to help the clarification, and continues:
JW> On this account, Lisp is a formal not a natural language, as you say. So
JW> I will coin a couple of different defining terms to make it clear what I
JW> mean. But hold on to that word "natural", because I'll get back to it,
JW> if only to explain where I thought the obvious pun was...
The way I read it, John conceded our initial point of disagreement in
the first sentence of that paragraph.
And in the same article, John continues:
JW>So let's call a "concrete language" any language which exists in
JW>spacetime, and an "abstract language" one that cannot. Lambda calculus
JW>ais an abstract language. Lisp is a concrete language, where implemented
JW>(and abstract where not). You might call this the "concrete
JW>interpretation" of Lisp, or the instantiation of Lisp, if you like.
Richard, can you help me out here? I'm reading this again, as
suggested, and I still think you are mistaken. John is confusing what
I call `utterances' of lambda-calculus or Lisp (which you and I and
Chris agree are finite) with what i see as a mere implemntation
detail: whether the formal langauge definion does, or does not,
specify behaviour in the event that storage is exhausted.
Now, I have no idea what it means for _languages_ to exist in
space-time. I dont see how English `exist in spacetime' as a
_language_. Senttences in that langauge, yes. Utterances in the
language, yes. Written records, yes.
Individuals with compentecy, of one kind or another (The two-year old
I mentioned loves Maurice Sendak's books)
But all of that also hold equally true for Lisp *and* for
lambda-calculus. Yet John is saying, clearly and distinctly, that
"Lisp is concrete where implemnted" (in the same says he says English
is concrete); but lambda-calculus is not. I see that very clearly.
So just what is it you think I'm making an effort to misunderstand, again?
Now, my server NNTP may be runnin behind, but John has been silent so
far on whether Lisp-competent humans doing reductions on a blackboard
is an `implementation': i.e., is it `concrete' or `abstract'.
Whichever way John calls it, I cannot see *any* interally-consistent
way for John to say that humans doing beta-reduction on
lambda-calculus on a blackboard is anything but the same.
I trust you will recall that the alleged _ontological_ distinction
between lambda-calculus and Lisp is *precisely* one of the points on
which I challenged John. Do you still think I am misreading him?
Is this a good time to repeat that there really *are*
programming-languge implementations which do explicit lambda-calculus
reduction? Are you going to side with John, that those are `abstract'
and therefore have no meaning? Or will you say that these
programming- language implementation sof lambda-caclulus are `concrete'?
If the latter, I think you then have to also concede that John's
claim, that some languages are `concrete' whereas others are
`abstract' doesn't hold up. At least not in the way John has
defined it so far.
John Wilkins
> John Wilkins <john.w...@bigpond.com> wrote:
> > Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
> [snip]
> >> This does not make sense to me. A concrete implementation will
> >> fail to handle all Lisp programs. If you take this approach,
> >> there is no one Lisp language, and it may even vary with the time
> >> of day on a machine with other processes competing for available
> >> resources. I don't this it makes sense to call this the language.
> >
> > Well I must ask you then what I asked Jon - where does Lisp reside? Does
> > it exist other than in its concrete implementations.
>
> Lisp, being an abstract formal system, does not exist as
> a physical thing. The concrete implementations are not
> Lisp: they are Lisp interpreters. Lisp resides in the same
> place as lambda calculus.
Granted. What I'm trying to get Jon to admit, and which, IIANM he did,
is that forms exist outside time and space. This is Platonism to the
letter. Is this also what you accept?
Not that there's anything *wrong* with Platonism. Some of my best
friends are Platonists, although I wouldn't want my daughter to marry
one.
>
> But let me ask you: if my computer has less memory available
> than yours, but they run the same software to interpret Lisp
> programs, and there are some Lisp programs you can execute
> that cause my smaller machine to run out of memory -- can we
> should we say that both machines are running Lisp programs?
Yes, if by "running" you mean "implementing".
>
> If my niece has a smaller vocabulary than me, then can we
> really say that she speaks English, or is she using some other
> language?
Everyone uses an idiolect, which is a kind of local instantiation of a
dialect. A language is a dialect with a dictionary (used to be an army,
but the press is mightier than the sword).
English is not an abstract thing except in the minds of certain teachers
and indignant letter writers to newspapers. It is a historical, changing
thing, with no essence. It has no "definition" or intension, other than
the "historical intension" of "descended historically from the initial
English language community", and even that is not entirely true - like a
species, once it is no longer monophyletic, a language is no longer
possessed of an essence.
>
> Cheers -- Chris
John Wilkins
> On 8 Jan 2002 18:14:07 -0500, Chris Ho-Stuart
> <host...@sky.fit.qut.edu.au> wrote:
>
> >Richard Harter <c...@tiac.net> wrote:
> >[snip]
> >> You seem to be working very hard at not reading what John has to say.
> >> What you say about English usage is quite true; however English usage
> >> regularly blurs a critical distinction that philosophers fuss about,
> >> the distinction between abstract entities and their instantiations.
> >> The chalk marks on the board are simply chalk marks; we call them lisp
> >> because there is an isomorphism between the chalk marks and Lisp as an
> >> abstract entity - or perhaps not; there are some very large and very
> >> ancient boojums in this territory. In any event, your appeal to
> >> English usage in CS departments (if there be such) is an appeal to
> >> irrelevant authority.
> >
> >I'm Jonathan on this one. The chalk marks are not Lisp.
>
> I dunno if you are with him or not. His wording suggests that he
> thinks that the stuff on the blackboard is lisp, actual lisp. Perhaps
> you and he can agree whilst using apparently contradictory words.
>
> >The
> >chalk marks, however, do represent sentences in the language
> >(notation) Lisp. Similarly: magnetic patterns on a hard disk
> >are not Lisp: they may, however, represent sentences in the
> >language (notation) Lisp.
>
> Are you thereby agreeing with John that Lisp is an abstract entity not
> present in space-time? (Not that John necessarily holds that position
> - he is pushing a different position.) As a side note, you shouldn't
> confuse "language" and "notation".
Richard herewith demonstrates that my inability to state things clearly
is not a necessary bar to understanding me. This comes as a definite
relief to my fevered and analgesed brain.
>
> >Trying to distinguish a formal
> >notation (Lisp-1) from a concrete notation (Lisp-2) on the
> >basis that I can only fit so much on a chalk board, or excecute
> >expressions up to a certain size with my compiler, is not
> >sensible.
>
> I might or might not agree. Do you understand the difficulty inherent
> in your claim?
>
> >What is concrete is the chalk. What is abstract is
> >the notation used to interpret the chalk marks. And this remains
> >the same language, even for different chalk boards of different
> >sizes and hence able to handle different numbers of markings.
>
> Exactly where would this notation be? What is the nature of its
> existence? How can these finite transient events, chalk marks on a
> blackboard and the like, be coupled to those abstractions existing
> outside of time and space? Plato wants to hear your answer.
>
Bugger Plato - *I* want to hear the answer. But I warn you all - one
mention of the Third World, or World 3, or Popper in any context
connected here will result in immediate inhumation.
John Wilkins
> In article <1f5qle2.2urdyc263p6zN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> >Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
>
> >
> >*You* can. But can my 10 year old son? If not (and as smart as he is, he
> >cannot), then lamda calculus does not exist on that blackboard. The
> >blackboard is not a UTM. It cannot do the reductions itself. Where does
> >the LC exist here?
>
> John ,
>
> Your claim of yesterday was that there's some numinous(?) difference
> between Lisp on a blackboard and lambda-calculus on a blackboard. My
> point is that, empirically, there is no such difference.
Nope. At all times I say that any formal language that is expressed in
physical terms is concrete. I do not have a place for abstract entities
in my poverty-stricken ontology. They do not "exist". My argument is
that the properties ascribed to abstract symbolisms and procedural
languages are in fact not causal. They do nothing. We abstract the
(usually syntactic) properties of systems in terms of formalisations
(ie, what we find similar in the way we describe them), but these are
(useful) fictions.
>
> We can write English on a blackboard. You and I, and your ten-year-old
> son, can comprehend it. But the two-year-old who I toss over my head
> every time I meet him, cannot.
>
> I hope you are not aruging that the English goes away if we leave the
> two-year-old alone in the room? (and no fair appealing to erasers,
> either. Let's say the blackboard is above a two-year-old's reach.)
English is, as I said, a historical object. If we leave the room to -
say - Hindi speakers, then it is bounded elsewhere (where we, and the
rest of the English language community exist as physical objects in
relation to each other). But do to it what was done to Estruscan, and it
may very well vanish.
>
> Praps you should go back to bed until you are feeling less Searle-ey.
I've been accused of many sins, but being in agreement with Searle ain't
one of them. I think you are misreading me through some spectacles that
filter out my egregious and extreme nominalism.
>
> (PS: I do hope that gets a laugh. I'm no Harter, but I am not quite
> as unsophisticated as you think :)
It rated 1.2 on the groan scale.
John Wilkins
> In article <1f5qle2.2urdyc263p6zN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> >Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
> [...]
> >
> >Oh I agree with this statement. Does this mean, that if you run out of
> >chalk when doing a derivation in Lisp, that Lisp is unable to do that
> >derivation? No. Hence *I* am not confusing Lisp with its implementation.
>
> No; I beleive you are instead confusing concrete instances of Lisp,
> with a formal definition of Lisp. As I keep saying: I prefer to deal
> with finite utterances.
>
>
> >I believe you are. So, face the demon, Jon - where do formal languages
> >exist?
>
> Where do unicorns exist? we have lots of ink and other utterances
> which refer to unicorns.
I know very well where unicorns exist - as uninstantiated complex
semantic descriptions, and I know where semantic descriptions occur, too
(in heads and societies of heads). But unicorns and ink about unicorns
are not the same. Ink and utterances exist, as physical cases.
>
> English is a wonderful language; it leets one write things which have
> no intension. John, where do numbers exist? If their definitions were
> destroyed, are you saying numbers would cease to exist, too?
Yes. I am saying exactly that. Remove all counting entities from the
universe - that is, all systems capable of instantiating a definition of
number - and there is no number in the universe.
>
> Where does that leave us when it come sto counting fossil tree rings,
> rings which are reliably dated to long before the human species, let
> alone our invention of ideas like "number?
Tree rings exist. Ideas exist. But ideas of tree rings exist only when
there is a head capable of having that idea.
>
>
> >Are they physical? If so, where are their definitions? If all
> >definitions of Lisp were destroyed tomorrow, would Lisp cease to exist?
>
> If all English dictionaries were destroyed tomorrow, would English
> cease to exist? Likewise, if all _implementations_ of Lisp were
> destroyed tomorrow, would the _definition_ cease to exist?
> Hm? How about vice-versa?
Answered above.
>
>
>
>
> >In my view, it would.
>
> Just as a matter of fact, I think your viewpoint is
> insupportable. Individual copies of the ink and paper which define the
> formal system do exist, as real ink and paper. It is technically
> possible that some anti-Lisp Jihad could can destroy all copies; yet
> fail to destroy either all the the implementations of Lisp
> interpreters, or fail to destroy all human knowledge of Lisp. There
> might still be instances of Lisp implementations embedded inside verra
> complicated toasters, doing useful work. In that sense, "Lisp" would
> exist-- as implementations. But we might no longer have a name for the
> abstraction which those surviving implementations share.
>
> Does the abstraction cease to exist if we destroy its definition, yet
> leave some instances of the implementation around? I dunno, but I hear
> a Rosetta stone waiting in the wings.
>
> Was the Latin language destroyed when all native speakers of it died?
No, because non-native speakers continued to exist. But if all that
remained of the Latin language was the name, would the Latin words and
their meanings "exist"? Does that mean if I define arbitrarily and
intensionally some abstract language Google (hey, they do such wonderful
work for the internet community, I say pay them back :-), does that mean
that there now exists a whole syntax and vocabulary of Googlese? Nope.
It doesn't exist. Just a name.
>
>
>
> >But English is a historical object, bounded by time and space. The LC is
> >not (theoretically).
>
> Wrong. (I think you must be feeling very ill.) All usages of the
> lambda calculus *do* exist as historical objects, bounded in time and
> space, in exactly the same way that usage of English do.
See the slide? I do not deny that the *uses* of LC are bounded. Of
course they are - so is your blackboard with LC symbols. But LC itself
isnot.
>
> Why do you keep arguing otherwise? No, I tkae that back. Why do you
> argue that lambda-calculus is any different from, say, English?
>
> Are you trying to push for the lambda-calculus having some
> platonic Form?
Not exactly. I am trying to argue that an abstract object like LC does
not, in point of fact, exist.
>
> All those concrete examples are no more and no less physically bounded
> than the English language. I see no difference between the existence
> of the English language as historical objects, and the lambda calculus.
> You seem awfully hung-up on the lambda-calculus being different.
> I truly cannot see why.
>
>
> John, are you trying to say that there's something about the
> denotation of mathematical abstractions like "lambda calculus",
> which does not apply to abstractions like "the english language"?
> I don't see that at all.
No, "the English language" is an abstraction. But English is not - it is
a bounded object. All classes that are intensionally defined rather than
ostensively defined are abstract.
>
>
>
>
>
>
>
>
> Are you arguing (with Richard -) that the English language has some
> Platonic ideal somewhere? That would be very unlike you.
>
> >As to being at odds with decades of CS usage, I observe that when I did
> >my computing, I noted that their usage was at odds with centuries of
> >philosophical logic usage. I got used to the idea. But their usage is
> >not authoritative.
>
> Oh yes it is, when it comes to CS. Just as English speakers are
> authoritative when it comew to English. Again: if you destroy all the
> English speakers and readers, does that destroy English?
>
> More to the point; if we destroy all speakers and readers and writers
> of Ancient Egyptian, does that destroy the Ancient Egyptian language?
> (Think carefully about Rosetta stones before you answer.)
Okay, let's take the Rosetta stone (incidentally, it is a decryption key
for Linear B Minoan, not Ancient Egyptian). Suppose all evidence of
Linear B had been destroyed, except for the Rosetta Stone. We would have
enough to decrypt only as much of Linear B as it preserved. I would say
that only that part of Linear B that was preserved and translatable into
known languages "exists". If, ten minutes before it was found by modern
scholars, an earthquake or volcanic eruption destroyed the Rosetta
stone, then Linear B would no longer exist anywhere. Its boundaries
would have been finalised, as it were.
>
> >
> >Fine, so what *is* the abstraction, ontologically speaking? I say that
> >it is a matter of a relation between symbols and cognitive devices,
> >which are physical.
>
> Excuse me, mlud. I __think__ this is where I opt for crude
> Australasian pragmatism, and to hell with ontology as `prior' to
> knowlege of the real world. But, mlud, I must plead a one-week recess
> to check with my brother, after his flight gets in :-)
Well, I do look forward to what he has to say. I'm sure, being a
philosopher, he'll be able to find things to attack. But I am most
interested in what *you* have to say. Clearly and distinctly now - do
abstract objects exist independently of their concrete instantiations or
not, the way, say, your fossil tree rings did? Answer that, and we'll
move on from there.
>
> Provisionally, though, and subject to revision (I am working in
> another window, editing this between compiles, so I'm rather distracted):
>
> I am happy with the idea of symbols. I lost track of what "it"
the abstraction
> referred to. I have no idea what you mean by "cognitive devices" in
> this context.
Brains, texts, computers, abaci, message sticks, marks in the sand,
community rules, etc.
>
>
>
> >> In case you didn't know: there are also definitions of
> >> almost-but-not-quite Turing-complete languages with finite storage.
> >> The language definition -- what you call the ``formal'' abstract and
> >> not-concrete version -- defines, as part of the language, how an
> >> executing program is notified that it's run out of storage (or just
> >> about to run out, since the notification is done whilst there's still
> >> enough space to do something about it).
> >
> >Nice, but not a counterinstance. Let's remain here with Turing-complete
> >languages.
>
> No, lets not. Lets stick with real-world concrete instances, since
> that's where I disagreed with you.
No. You disagreed that formal languages were concrete, when we'd cleared
away my bad attempt at a pun.
>
>
>
> >You are. (Sorry, you asked). This is the medieval universals debate
> >begun by Roscellin, who argued that a universal (an abstract in our
> >terminology) is mere breath "flatus vocus". I don't go this far, but it
> >does have a history (actually, I think that history begins sometime
> >between Aristotle and Porphyry, but is not preserved).
>
> Scuse me, but having been round this a few times, I'm fairly certain
> that *NOT* what I'm trying to say. I think you may be reading too much
> into my repeated comparisons with the English language.
>
> I have it on reliable authority (ie., a student of Kim's) that when
> what I say when left to my own devies sounds a hell of a lot more like
> Australasian pragmatism than the impression you seem to have gathered.
> That pragmatism is, if anything, tempered by the work I did years ago
> in AI, but that's another story.
As I recall we Antipodeans were supposed (by Feyerabend) to be rampant
physicalists, but pragmatism is an American disease (I am afflicted with
both, of ocurse).
>
> Science does seem to be a pretty darn good way of learning about the
> world (and of describing it). As for what philosophers
> traditionally hold about the primacy of ontology -- well, as Sam Vimes
> said: arseholes to the lot of 'em.
Quine wrote (an American pragmatist, BTW) that "to be is to be the value
of a variable" in a scientific theory, and I happen to agree, in a
limited way, with him [at least one way things exist is as values of
variables, but we have to be careful not to put the epistemic cart
before the ontological horse]. But that does not mean that we need to
have a theory for things to exist in; nor do we need a proper
metaphysics before we can do science. Hell, you can do some science if
you believe in creationism, to a degree. Why not if you are merely
saddled with the wrong ontology? It's no accident that the Metaphysics
comes *after* the Physics in Aristotle's corpus.
But wrong ontologies do have epistemic, heuristic and practical
consequences. One of these is to think that things which do not exist
have some causal role. For instance, under neo-Platonic assumptions,
medievals and later morphological idealists thought that Form had a
causal role in biology. That *is* a category error. Another is that
Information has a causal role in biology (effectively the same mistake,
actually). Information, being an abstraction, affects nothing. It
resides in the (concrete) descriptions of biological processes we make
after the event.
>
> (This, I hear, is approximately what occasionally happens to my
> brother in post-colloquium sessions. He puts forth his viewpoint, the
> other students say, "oh, you're just an Australasian", and continue on
> regardless. Then again, he may be egging them on.)
Australians never do this, but Kiwis may be less discrete. :-)
>
> [A good hundred lines more, snipped as hoplessly off-topic, which
> I can flesh it out via email if anyone really cares.]
Jonathan Stone
Richard Harter <c...@tiac.net> wrote:
>On 8 Jan 2002 19:44:52 -0500, Chris Ho-Stuart
><host...@sky.fit.qut.edu.au> wrote:
>
>
>[snip a lot of stuff that Jonathan answered very well]
Wow, you really mean that? I am putting this right beside the
thank-you letter I got from Richard stallman, for explaining hwy he
was wrong about something; and the thank-you from Andy Tanenbaum for
fixing his bug in the Minix 1.2 driver a day before it went to press.
>I babbled about this at greater length elsewhere for which see. You
>see, though, I hope, that you are using words such as "is" and
>"exists" in ways that are very different. Paris exists. (I have my
>doubts about South Dakota.)
Oh. I heard that South Daklota was real, but North Dakota (and
its missile fields) was a CIA plot to scare the Russians out of
their jodphurs. Now you say SoDak may not be real either.
Should I file that one with the Ma-and-Pa-Kettle explanation?
>To say that "Lisp exists" says something
>very different from "Paris exists". What is more it is very hard to
>pin down what you (or I or anyone) means by saying "Lisp exists".
Richard,
I'm on the verge of needing codein myself, here. But all I've been
asking is, what on Urath does John mean when he says that
and just how does
"English exists"
entail different ontologocal baggage from
"Lisp exists"
and
"lambda calculus exists"
because I have read John most carefully, and he is saying that one of
those three is ontologically not like the others.
(Apart from John's caveats about implementation. Those are red herrings;
or rather, a lack of knowlege of really arcane coding, on John's part.)
Chris Ho-Stuart
>> Lisp, being an abstract formal system, does not exist as
>> a physical thing. The concrete implementations are not
>> Lisp: they are Lisp interpreters. Lisp resides in the same
>> place as lambda calculus.
>
> Granted. What I'm trying to get Jon to admit, and which, IIANM he did,
> is that forms exist outside time and space. This is Platonism to the
> letter. Is this also what you accept?
Lisp does not exist as a physical thing, and saying it exists
as something else is using "exists" in a quite different way,
I think.
Lisp could be thought of as a form, or as an adjective used to
describe certain utterences that exist in the real world. As an
abstract form, or as an adjective for utterances: it is still
in the same category as English, as far as I can see; except
that the utterances are used for different purposes, as we
agreed earlier.
I really really really really [.. repeat some arbitrary number of
times ..] really suspect that the distinction you are drawing
between English and Lisp is invalid.
English is an abstract form (albeit a vague one!). There is, in
the abstract, a notion of English utterances, and in the concrete,
we can not hope that "they" will all be uttered in space time. But
we can recognize them. Lisp is an abstract form: and there "are"
many Lispy utterances that will never exist in the concrete; but
we can recogize Lispy utterances.
Do you think this means there exists an abstract English form
as a Platonic essense (whatever that means)?
And (whatever answer you give to that questions) why not give
the same answer with respect to Lisp?
I think, from reading the rest of your post, that you might
say that Lisp is formal, meaning unamiguous, which means that
recognition of Lispy utterences is not as problematic as
recognition of English utterances.
But *you* understand that difference as implying the existence
of a Platonic essence? Why would that difference mean that
English "exists" in some sense that Lisp does not?
And, in point of fact, Lisp is not completely unambiguous, in
that there are various dialects, and changes to standards over
time.
I have no idea, frankly, whether I am a Platonist or not. But I
certainly think that Lisp is not a thing that exists in the same
sense that a given Lispy program exists, or even in the same sense
that a formal grammar for Lisp can exist. I accept that the
identification of Lispy sentences is unambiguous; or at least less
ambiguous. Identificaiton of English sentences has shades of grey,
both in variation over time and various local idiosyncracies.
But what does that have to do with "existence"?
> Not that there's anything *wrong* with Platonism. Some of my best
> friends are Platonists, although I wouldn't want my daughter to marry
> one.
>
>> But let me ask you: if my computer has less memory available
>> than yours, but they run the same software to interpret Lisp
>> programs, and there are some Lisp programs you can execute
>> that cause my smaller machine to run out of memory -- can we
>> should we say that both machines are running Lisp programs?
>
> Yes, if by "running" you mean "implementing".
>
>> If my niece has a smaller vocabulary than me, then can we
>> really say that she speaks English, or is she using some other
>> language?
>
> Everyone uses an idiolect, which is a kind of local instantiation of a
> dialect. A language is a dialect with a dictionary (used to be an army,
> but the press is mightier than the sword).
>
> English is not an abstract thing except in the minds of certain teachers
> and indignant letter writers to newspapers. It is a historical, changing
> thing, with no essence. It has no "definition" or intension, other than
> the "historical intension" of "descended historically from the initial
> English language community", and even that is not entirely true - like a
> species, once it is no longer monophyletic, a language is no longer
> possessed of an essence.
OK. There is a basis for distinction with Lisp here: in that Lisp
is less ambiguous, and that we can point to a concrete utterance
in another language (a grammar, for example) that describes Lisp.
Actually, there are a couple of such definitions; and the
definitions do change somewhat over time. But why does that mean
English exists in some sense that Lisp does not? What sense is that?
Cheers -- Chris
Jonathan Stone
>I know very well where unicorns exist - as uninstantiated complex
>semantic descriptions, and I know where semantic descriptions occur, too
>(in heads and societies of heads). But unicorns and ink about unicorns
>are not the same. Ink and utterances exist, as physical cases.
Wow. At last I think we've gotten beyond talking past each other.
Does this mean you're feeling better?
>> English is a wonderful language; it leets one write things which have
>> no intension. John, where do numbers exist? If their definitions were
>> destroyed, are you saying numbers would cease to exist, too?
>
>Yes. I am saying exactly that. Remove all counting entities from the
>universe - that is, all systems capable of instantiating a definition of
>number - and there is no number in the universe.
I really meant _just_ the defintion, not the competence too; you
answered a question I asked a bit later.
Okay, next question: suppose for the sake of argument that we, or some
other species, rediscover an isomorphic concept. Is that the _same_
concept, or a different one? How does that extend to independent
discoveries of what we decide is the same idea?
How 'bout Marty's question about Fermat's last theorem?
>Tree rings exist. Ideas exist. But ideas of tree rings exist only when
>there is a head capable of having that idea.
So far so good. Are numbers _just_ ideas? Is it meaningful to ask
about the number of tree rings in some point in time before the
idea `numbers of tree rings' is formed?
[snip stuff answered above]
>> Was the Latin language destroyed when all native speakers of it died?
>
>No, because non-native speakers continued to exist. But if all that
>remained of the Latin language was the name, would the Latin words and
>their meanings "exist"?
Somewhere about where, I asked you about Rosetta stones. I'm sure I
did. But none of this tells me why Lisp exists as `concrete',
but lambda-calculus does not.
>Does that mean if I define arbitrarily and
>intensionally some abstract language Google (hey, they do such wonderful
>work for the internet community, I say pay them back :-),
I can pass that on ot Larry and Sergey next time I see them,
if you'd like.
>
>See the slide? I do not deny that the *uses* of LC are bounded. Of
>course they are - so is your blackboard with LC symbols. But LC itself
>isnot.
I tell you, three times now, that there are computer programming
langauges which are explicilty implementeed via lambda-calculusn
reductions. Will you now abandon your original claim that Lisp is a
`natural' language but lambda-calculus is not? Will you concede that
your original claim, that LIsp was concrete but Lambda-calcus is not,
was incorrect?
A yes or no will suffice.
>Not exactly. I am trying to argue that an abstract object like LC does
>not, in point of fact, exist.
I tell you, four time now, that LC is *not* just an abstract object;
it is implemented as conceretely as Lisp. Even if it weren't, I submit
that I and other humans *can* demonstrate competence in both
lambda-calculus and Lisp. Soetimes on the same blackboard, indeed
sometimes during the same class period.
I sumbit that your claim, that LC and Lisp somehow carry different
ontological baggage, is *ddemonstrably* incorrect,and for for exactly
the same reasons you caved in, and said that Lisp was concrete only
where implemented. Am I going wrong somewhere?
>> All those concrete examples are no more and no less physically bounded
>> than the English language. I see no difference between the existence
>> of the English language as historical objects, and the lambda calculus.
>> You seem awfully hung-up on the lambda-calculus being different.
>> I truly cannot see why.
>>
>>
>> John, are you trying to say that there's something about the
>> denotation of mathematical abstractions like "lambda calculus",
>> which does not apply to abstractions like "the english language"?
>> I don't see that at all.
>
>No, "the English language" is an abstraction. But English is not - it is
>a bounded object. All classes that are intensionally defined rather than
>ostensively defined are abstract.
I find not a thing comprehensible in this. the token English without
qoutes, seems no different that Lambda-calculus without quotes. Your
claim is that English exists in the real world in some way that lambda
calculus (or Lisp, sans implementation) does not. That's bullshit.
Between us, Chris and I (and marty, it seems) know communities
competent in both, in just the same sense that they are competent in
English. You can say that it exists in their heads or collections o f
heads, fine (thats' where I was going in the stuff I snipped.)
But as a question of empirical fact, the difference you wish to draw
between Lambda-Calculus and Lisp *DOES NOT EXIST*.
Will you at least admit you were wrong about that much?
>Okay, let's take the Rosetta stone (incidentally, it is a decryption key
>for Linear B Minoan, not Ancient Egyptian).
Nope. Demotic plus something Gk (true) but definitely hieroglyphics
and cartouches is what I recall. From a biography of wosshame, the
French guy, Champoullion (sp!!). Linear B is the _hard_ one,
deciphered by... [cheats, uses Google] Michael Ventris in the 50s,
using data from the grave-robbing, sorry, excavation at Knossos from
the first half of the century. Linear A is still a mystery, AFAIK.
>Linear B would no longer exist anywhere. Its boundaries
>would have been finalised, as it were.
I am *almost* willing to buy this, along with the obvious
generalizations. The caveat is that modern cryptanalysis has gotten
to the point where significan progress can be mad even in the complete
absence of speakers, and of cribs like the Rosetta stone.
It may e possible for a langauge to die, to be left with nothing
more than a buncho f artifracts -- a little more than the Phaistos disk,
or Orongorongo, to be sure -- along with context.
Suppose, for the sake of argument, that sup-dup computers somehow
manage to correate the bits and come up with putative translations
that make sense in context; and that also make sense in
newly-discovered, heretofore unknown artifcats.
What would your philosophy say about that?
>Well, I do look forward to what he has to say. I'm sure, being a
>philosopher, he'll be able to find things to attack. But I am most
>interested in what *you* have to say. Clearly and distinctly now - do
>abstract objects exist independently of their concrete instantiations or
>not, the way, say, your fossil tree rings did? Answer that, and we'll
>move on from there.
You first. Concede your initial assertion that Lisp was concrete, and
Lambda-calculs was abstract, was not in fact correct, and I'll email
you what *I* wrote.
>> referred to. I have no idea what you mean by "cognitive devices" in
>> this context.
>
>Brains, texts, computers, abaci, message sticks, marks in the sand,
>community rules, etc.
Ah, thank you.
>>
>> No, lets not. Lets stick with real-world concrete instances, since
>> that's where I disagreed with you.
>
>No. You disagreed that formal languages were concrete, when we'd cleared
>away my bad attempt at a pun.
Here I must return Mr. Harter's compliments about misreading. I
beleive I said, quite consistently that specific fragments--- that is,
*utterances* -- were concrete, irrespectivce of whether the utterances
written on the blackboard were in Lisp or in lambda-calculus. You, on
the other hand said that `languages' may be subject to ontological
excess-postage, according to whether the `language' was Lisp or
nlambda-calculus.
I *did* check, and I quoted you, and that *is* what you said.
Also, I think you will also find that it was me (not you) who started
drawing distinctions between utterances, and langauge implementation;
and matters of definition.
I will still cheerfully assert that I see no no ontological difference
between a collection of heads which agree on a semantics for
lambda-calculus; and a collection of heads which agree on a
semantics for Pure (applicative-only) Lisp, or indeed on a
semantics for English.
>As I recall we Antipodeans were supposed (by Feyerabend) to be rampant
>physicalists, but pragmatism is an American disease (I am afflicted with
>both, of ocurse).
Well then, I guess that gets me coming and going.
>> [A good hundred lines more, snipped as hoplessly off-topic, which
>> I can flesh it out via email if anyone really cares.]
Argh, it sems you really *did* want that. By the weekend -- its
going 1am here.
Marty Fouts
jona...@DSG.Stanford.EDU (Jonathan Stone) writes:
[snip]
> How 'bout Marty's question about Fermat's last theorem?
It's a trick question. The laughing ghod in the corner knows the
answer, though.
[snip]
Richard Harter
>
>c...@tiac.net (Richard Harter) writes:
>
>[snip]
>
>> In the early years of the past century there was rather a lot of
>> discussion of the foundations of mathematics.
>
>Early years? Kline, rest his sould, was still writing well into the
>'80s. ;) Mathematics: The Loss of Certainty was published in '82,
>IIRC, and is still in print.
>
>More realistically, there are still world conferences on the
>philosophy of mathematics.
Indeed. However mathematicians don't care anymore. :-)
>
>> There were several distinct schools of thought as to the nature of
>> mathematical truth - platonism, intuitionism, formalism, and so on.
>
>Are. It's been less than three days since the last argument I had with
>a strict constructionist.
Also are, yes. Why did you argue with the strict constructionist?
Strict constructionism is the one true faith.
>> [Bear with me, I'm sure that this is old stuff to you but it is not
>> to others.] One of the viewpoints was the formalist approach which
>> said, roughly, that mathematics is a game played by making
>> uninterpreted marks on pieces of paper according to specific rules.
>> A great merit of the formalist approach is that it provided an
>> agenda for a broad research program that has led by various routes
>> to computer science, UTM's, ATM's etc.
>
>Indeed. This is pure mathematics. All else is 'mere application' ;)
>
>> In the course of implementing this research program the original
>> pieces of paper with the pencil marks have vanished. That is, they
>> haven't physically vanished. They and their equivalents, the
>> blackboards and whiteboards, and mag tapes, and CD's, and rams and
>> roms, and flaishing lights are all physically present. In many
>> mindsets they are mentally invisible; the mind's eye looks at the
>> chalk marks and sees a lisp program or statement in the lambda
>> calculus. In short, people look at physical things and "see" the
>> abstractions they have been trained to see.
>
>Some of them do it without the physical representations. (No, I'm not
>jealous. Not even a little.)
Ha! They look at the physical representations too; it just happens
that they (the physical representations) aren't there.
>> One can take the view that these abstractions are no more than
>> convenient fictions, that the chalk on the board, et cetera is what
>> is real. (I will leave it to John to say for himself if the current
>> remarks are in any way related to his thoughts.)
>
>So if Fermat really did have the proof that wouldn't fit into the
>margin (and thus never wrote it down anywhere) is it a convenient
>fiction or a real abstraction? And does the fact that we don't know
>what it was change its reality and existence?
I dunno. I suppose on said view the proof would have been his thought
processes (whatever they might be in terms of physical events.) I
would take it, though, that the proof existed while he had it in mind
and does not exist now. The question of the reality of past events is
another matter entirely.
>> There even is some profit in doing so. For one thing it is very
>> easy for empty abstractions, i.e., abstract terminology that is not
>> eliminable, to creep into our discussions. For another thing these
>> abstract engines of thought imply claims about the universe that are
>> not true. For example, one of the Peano axioms asserts that every
>> number has a successor, i.e., for every number x there is a number y
>> such that y=x+1. In other words given a number x we can add 1 to
>> it. There is a definite sense in which this is not true. The
>> integers which we and our artifacts have referenced or will ever
>> reference is finite; inevitably there has been or will be integer
>> which are referenced but whose successor will never be
>> referenced.
>
>You're assuming facts not in evidence, that is, that time has an upper
>bound, at least so far as intelligence is concerned. Were that
>statement not true then your observation does not hold.
Good point, although the assumption is that the set of integers
constructed by humanity over its entire history is bounded. Even if
time is unbounded the duration of the human species is likely (but not
certainly) bounded. Even if the duration of the human species is
unbounded the maximum of the integers referenced may be bounded.
>> Paradoxically we will never know what these integers are. In short,
>> the statement that we *can* add 1 to every integer is false.
>
>I think you mean 'will' rather than 'can'. There is no integer you can
>name that I can't add one to.
No, I mean 'can' as well as 'will'. You can add one to every integer
that I name; you can't add one to every integer. For that matter you
can't name almost all of the integers. There is a paradox which may
be original with me as follows:
Consider the set S of all integers that humans will ever name during
the course of its existence. [We may take it that the duration of the
human species as species is bounded even if time is unbounded.] It is
readily seen that S is a finite set. Let N be the largest integer in
S. Now consider the number N+1.
>
>> A good platonist (as the term is used in the philosophy of
>> mathematics) insists that the successors (and the whole infinite
>> suite of integers) exists willy-nilly. This view is supportable if
>> the word "exists" is given different meanings according to context.
>
>Somehow I feel like bringing up Quine's indispensibility argument at
>this point, but I think I'll dispense with that idea.
Speaking of having no shame.
>
>(http://plato.stanford.edu/entries/mathphil-indis/ for those who care)
I'm not much enthused about the indispensibility argument but then I'm
not an admirer of Quine's ontology (although he is fun to read) - I
take the view that fuzzing together the different senses of "exists"
and "is" is a serious error.
[snip]
Richard Harter
Stone) wrote:
>In article <3c3beb76...@news.SullyButtes.net>,
>Richard Harter <c...@tiac.net> wrote:
>>On 8 Jan 2002 19:44:52 -0500, Chris Ho-Stuart
>><host...@sky.fit.qut.edu.au> wrote:
>>
>>
>>[snip a lot of stuff that Jonathan answered very well]
>
>Wow, you really mean that? I am putting this right beside the
>thank-you letter I got from Richard stallman, for explaining hwy he
>was wrong about something; and the thank-you from Andy Tanenbaum for
>fixing his bug in the Minix 1.2 driver a day before it went to press.
You're welcome.
[snip speculation about the existence of South Dakota]
>Richard,
>
>I'm on the verge of needing codein myself, here. But all I've been
>asking is, what on Urath does John mean when he says that
I opine that his latest posting is clear. He is professing (whether
he is serious or not is another matter) a fairly radical nominalism in
which abstract entities do not exist apart from the physical baggage
associated with them.
>
>
>and just how does
>
> "English exists"
English "exists" because there are speakers of English, works written
in English, dictionaries, et cetera. The word, English, is a label
for things going on in the "real world".
>entail different ontologocal baggage from
>
> "Lisp exists"
I think what he is saying is that Lisp exists in the sense of being a
computer language extant in the real world with manuals,
implementations, et cetera, whereas he is denying that it exists in
the sense of there being an abstract Lisp.
>and
>
> "lambda calculus exists"
Here he is denying that the lambda calculus, considered as an
abstraction, exists. It is arguable that he is not being consistent
because, after all, there are books and such like about the lambda
calculus. However his point is that it is conceived of as being an
abstraction and as such by his lights doesn't exist. Mind you, he is
willing to concede the existence of the idea of the lambda calculus,
just not the thing itself.
[These interpretations subject to correction by John - it's his baby.]
>because I have read John most carefully, and he is saying that one of
>those three is ontologically not like the others.
As noted elsewhere in other words, the word "exists" is somewhat of a
weasel word. Quine begins one of his essays with the observation that
the question of ontology can be summarized in three words - what is
there - and answered in one - everything. He notes, however, that
there is some room for arguing about cases. Somewhere, perhaps in the
same essay, he expresses the view that it is quite all right to apply
to "exists" to abstract entities; lacking physicality does not disbar
them from existence. One can defend this view on the grounds that in
formal logic it does not matter if the entities being referred to are
physical or not. I opine that this argument is correct but
misleading. None of the entities in formal logic are real.
John Wilkins
> john.w...@bigpond.com (John Wilkins) writes:
>
> > I can see that codeine-addled[*] attempts at puns and philosophical
> > jokes are just confusing the issue. So I will attempt to put my case a
> > bit more carefully, with no smileys...
> >
> > Some basic definitions:
> >
> > A formal language is, for this purpose, a language without ambiguity,
> > amphiboly, or contradiction, which is fully coherent and formalised.
>
> I came late to this party, so forgive me for asking what may be obtuse
> questions: in what sense do you use 'coherent'?
Logically. No contradictions can be derived from the elements of a
coherent language. But this was off the top of my head, to assuage Jon's
concerns. It's neither sufficient, nor, I suspect (as you note below)
necessary.
>
> > A natural language is one which can be ambiguous, amphibolous, or
> > contradictory, and may be unformalised in its syntax or symbolism, and
> > may also be incoherent.
>
> Is amphiboly here meant to construe something other than merely
> 'ambiguous sentence'? (Does it matter or am I just being pedantic?)
I was being pleonastic :-) [Gad, a fellow can't have *any* fun among
this crowd. Pedants, all of them.]
>
> Anyway, here is an alternative set of definitions:
>
> language: "form of speech used by a group of people with a unique set
> of vocabulary combined in uniform patterns (syntax) to convey meaning
> (semantics)."
>
> "natural language": A human language not specifically designed for
> communicating instructions to a computer.
>
> "formal language": a set of strings from an alphabet (alphabet =
> finite set of symbols, symbol = an abstract entity having no meaning)
>
> note that 'natural language' is a subset of 'language' but that
> 'formal language' is a definition from a different domain of
> discourse. (Language and natural language are terms from linguistics,
> formal language is a term from mathematics.) Although, one can pretty
> much bend the rules and make language, natural language, and computer
> language all subsets of formal language.
Hmm. I was taught that natural languages were things that could not be
explicitly cast into formal language terms (ie, 2nd order logic, etc)
without ambiguity. But undergrad logic is 20 years ago, and I have made
a practice of avoiding logic where possible ever since. I was never
going to contribute to that field.
>
> > I'm sure these are not good definitions, but I'm not completely
> > defining anyway. Let's say these are facets of natural and formal
> > languages.
> >
> > On this account, Lisp is a formal not a natural language, as you
> > say.
>
> Being pedantic, I will note that all computer programming languages
> fail your definition of formal, as even the most precisely defined
> contain ambiguity.
Truly? I didn't know that. In that case can we say that no computer
language is reducible to the lambda calculus that Jon is so fond of, or
does that also have ambiguity in it?
>
> > So I will coin a couple of different defining terms to make it clear
> > what I mean. But hold on to that word "natural", because I'll get
> > back to it, if only to explain where I thought the obvious pun
> > was...
> >
> > A concrete object is one that is bounded by spacetime (this is
> > Zalta's definition), and an abstract object is one that is not nor
> > can be. A universal property is therefore abstract, and a particular
> > property is concrete.
>
> Ih this model are subjective properties, such as color, concrete,
> because existing in the mind of the observer they exist in spacetime,
> or abstract, because differing from observer to observer they are not
> fixed?
Insofar as they are properties (please, can we *not* get into philosophy
of mind here? I have the flu. I refuse to consider philosophy of mind
while I have the flu), they are physical. I suspect that subjective
"properties" are not properties, strictly speaking.
>
> > Call this nominalism, if you like, so long as it is understood that
> > this is a nominalism relative to some domain.
> >
> > So let's call a "concrete language" any language which exists in
> > spacetime, and an "abstract language" one that cannot. Lambda
> > calculus is an abstract language. Lisp is a concrete language, where
> > implemented (and abstract where not). You might call this the
> > "concrete interpretation" of Lisp, or the instantiation of Lisp, if
> > you like.
>
> Someone get me a pipe. [reference to Magritte or Freud, your choice.]
>
> Um, you're eliding what it means for 'language' to exist in
> spacetime. Is English concrete, because of the OED, or abstract,
> because there is no 'implementation'. If its not abstract then how
> does it differ from Lambda calculus, since mathematics is, after all,
> merely natural language with more precise symbolism?
Perhaps I am. I did not want to assume my general conclusion here. I do
think that mathematics is a natural language with a more precise
symbolism. Truly abstract languages are a kind of asymptote, like truth
(ooh, now *that* aside should upset one or two). But I was willing to
admit that LC was an abstract "entity" for the purposes of the argument.
>
> > So, now we have a distinction between a physical (ie, bounded)
> > object - Lisp on Chaitin's Solaris, or on John Wilkins's Mac or
> > whatever, which has all kinds of limitations (in terms of pointers
> > available at runtime, implemented handles, and the like) - and the
> > abstract *definition* of Lisp in whatever standard it is
> > defined.
>
> X3.226-1994. I could look up the ISO equivalent if it would help.
Ta.
>
> But I think there's a Chinese room around the corner. If
> implementation on a computer counts as concrete, then does a computer
> program that does voice recognition and text-to-speach generation make
> English concrete? If so, and I teach the program to 'speak' lambda
> calculus notation, how does that differ?
My Q&D solution would be that the Chinese room understands some subset
of English, and that if that were all the English in the world, then
that *would* be "understanding English" - there's be nothing else to
contrast it to. It appears I am a formalist after all.
>
> > This is a metaphysical definition. It doesn't affect the
> > standing of Lisp WRT lambda calculus or Gödelian arithmetic, &c. It
> > does, however go to the question of how any formal language is
> > implemented (your nice blackboard example). Chalk marks on a
> > blackboard are not Lisp. Nor are ink marks on a manual, even if the
> > symbols represent a reverse polish notation definition of the
> > language.
>
> If the symbols don't make the language concrete when they exist on a
> blackboard, why do they make it concrete when they exists as ASCII
> text in a computer? In particular, is the lisp program file on my
> laptop, which has no lisp interpreter concrete or not?
>
> > Where does Lisp exist if not in concrete form (even the abstract
> > definition must exist in concrete form)? If you are not a Platonist,
> > you must answer this way.
>
> My head hurts now. I think i'm getting the talk.origins virus. So far,
> all of the comments have really been about utterances in Lisp,
> (programs) rather than Lisp itself. Does the computer running a lisp
> interpreter "understand" lisp, or is it just a Lisp room? What does
> 'understand' mean, anyway?
Depends on what "does" means :-)
No - you are quite right; Lisp is a natural language after all. I'll now
stick with that. It is natural because the abstract entity "Lisp" is a
fiction. Abstraction is a moment in representation, not a state of being
:-)
>
> [snip]
>
> > What naturalism is trying to do, is to establish an account of
> > something procedural and prescriptive in terms of being concrete,
> > rather than abstract. So when I said that Lisp was a "natural"
> > language, I merely meant that any instantiation of Lisp must be a
> > concrete instantiation. There: if Harter has the most indirect
> > puns, at least they elicit a groan. Mine are merely needlessly
> > obscure.
> >
>
> [snip]
>
> well at least your's are merely obscure and you understand them. mine
> are obtuse and even I don't get them.
>
> Marty
No... I begin to think that mine are equally obtuse ;-)
John Wilkins
the formalist approach you discuss below - have you got a ref that
outlines that history?
I do agree with your interpretation of me.
Richard Harter <c...@tiac.net> wrote:
Late note - I should have checked my own shelves earlier...
John Wilkins
> In article <1f5qn55.w6ddgh1udwjdbN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> >Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
>
> >Well I must ask you then what I asked Jon - where does Lisp reside? Does
> >it exist other than in its concrete implementations.
>
> John,
>
> What then is English? where does it reside other than in concrete
>
>
> >> When I explain programming "language" to an introductory
> >> programming class, I tell them that that "language" here is
> >> a misnomer. A programming language is actually a mathematical
> >> notation for expressing algorithms; and more closely related to
> >> the notations of arithmetic than to real languages.
> >>
> >> Programming notations are used to express algorithms.
> >>
> >> Natural languages are used to argue, to make love, to
> >> sway emotions, to convey knowledge, to lecture, ...
> >
> >Oh, absolutely. Nobody ever got laid by writing a COBOL program, I'm
> >sure.
>
> Oh dear. I'm afraid there really *are* more things, Wilkins,
> than dreamt of in your philosophy.
I'm not *listening*... La la la la I can't hear you.
>
> >But I am aware that this is a matter of abstract objects like
> >numbers. The debate began on the standing of numbers.
>
> Ah. Now we're getting somerwhere.
>
> What is it that two asteroids and two COBOL programmers, sorry,
> angels, have in common? Did they have it in common before we
> had a *name* for it?
>
> How about if all things which have that property in common cease to
> exist-- does the property cease to exist, too?
We now are bound to have a discussion on what it is to have a property.
Next thing, we'll be discussing book zeta inthe Metaphysics.
>
> I *think* John may be confusing me with someone who confuses the
> English word "two" and the Roman numeral 2, with the number two.
> But that's not what I'm saying at all.
No, *I'm* the one who does that...
John Wilkins
> In article <3c3bc230...@news.SullyButtes.net>,
> Richard Harter <c...@tiac.net> wrote:
> >On 8 Jan 2002 19:09:30 -0500, jona...@DSG.Stanford.EDU (Jonathan
> >Stone) wrote:
>
> >Well, perhaps, but computer scientists tend to be dreadfully naive
> >about things like ontology.
>
> I think I have to give you aclean miss on that one. Have I failed to
> mention that my brother is a PhD candidate in philosophy, that my
> mother has a masters in philosophy, and that I spent Friday evenings
> for some years drinking wiht a philosopher of John's acquaintance?
Did I mention tat not all philosophers are always right about
everything? Even Moms with Masters? And sometimes, even Kim is wrong
(tell him I said that and I'll be after you with a 2 by 4. He edits the
journal in which I am most likely to publish...), although not usually
about ontology.
>
> I've been red-eyed for about 3 weeks. I have an _exceptionally_ low
> tolerance for GAs -- which is not Bobby's fault at all.
>
> I am sure no-one *he* works with claims that GAs give comparable TSP
> results, with comparable compute time, to algorithmic methods,
> forgetting to highlight gthat the GAs ran on machines 20x faster.
> bnut this Does Happen, as they say; and my hostility is not nearly so
> far from the norm as some here seem to think. And I havent even
> mentioned the stories of Koza's students, who tried ot do GP in
> the same machine-learning domain where I did my AI.)
>
> >Apologize first without qualification; it's good for the soul.
> > Don't argue, just do it.
>
> But what is it I should be apologizing to you and John *for*?
>
>
> >In the early years of the past century there was rather a lot of
> >discussion of the foundations of mathematics.
>
> [[[ snip lots -- i am a-Frege you are going to Russel through places where I
> can only Cantor.]
Fools Rescher in where angels fear to tread.
>
>
> >In short, people look at physical things and "see" the
> >abstractions they have been trained to see.
>
> Ah, now I see I shouldn't have snipped my digressions after all.
> No smiley.
>
>
> >One can take the view that these abstractions are no more than
> >convenient fictions, that the chalk on the board, et cetera is what is
> >real.
>
> I think you mean `fictions, that _only_ the chalk on the board is real'
>
>
> [...]
>
> > As a practical matter we shall go
> >happily on talking about electrons, atoms, the law of universal
> >gravitation, and universal Turing machines just as though they exist.
> >It is ever so much more convenient to do so. Still, we can explore
> >the view that the existence referred to are so many empty words.
>
> Richard, this is surreal. I've been teh one wishing to talk about
> the concerete writings, and the interpretations humans give them.
>
> I have studiously avoideding saying that the abstractions are Forms,
> or platonic Ideals, or any such thing. have *also* studiously avoided
> saying that they have no existence other than as the utterances.
>
> <Granny>see, I dont hve no truck with *either* view.<\Granny>
>
> If apologies are going around as good for the soul, perhaps you and
> John owe me one for casting me into the other camp, simply 'cause I'm
> not in the one (I won't even ask to take it in kind against some of
> the ones I indisputably owe Matt and Bobby Bryant.)
Jon, if you are not a Platonist in this respect, then what *are* you? I
can think of a few traditional views that one can take here, but what
you say surely sounds like Platonism.
Utterances are physical things (events, as it happens*). What makes one
physical thing (a Lisp symbol on a blackboard) the same as another (a
Lisp symbol in the registers of a CPU, or in your head)?
>
> [...]
>
> >There is a definite sense in which this is not true. The integers
> >which we and our artifacts have referenced or will ever reference is
> >finite; inevitably there has been or will be integer which are
> >referenced but whose successor will never be referenced. Paradoxically
> >we will never know what these integers are. In short, the statement
> >that we *can* add 1 to every integer is false. A good platonist (as
> >the term is used in the philosophy of mathematics) insists that the
> >successors (and the whole infinite suite of integers) exists
> >willy-nilly. This view is supportable if the word "exists" is given
> >different meanings according to context.
>
> I thought even philosophers of mathematics called that neo-Platonism.
> I guess I learnt something; thank you.
No; neo-Platonism has a rather more specific sense than that, and not in
philosophy of maths. Richard is right - in the context of the philosophy
of maths, this is Platonism. Here's Paul Bernays' definition: "...the
tendency of which we are speaking consists in viewing the [mathematical]
objects as cut off from all links with the reflecting subject." Quine is
rather more direct: he says that there are three (and only three)
options: "realism" (Platonic realism, or in then-current terms, logicism
- the view that universals or abstract entities have being independently
of the mind); conceptualism (or intuitionism, sensu Brouwer, and
Poincaré - that there are universals, but they are mind-made), and
nominalism (or formalism - that the syntactical rules of mathematics are
useful notations we happen to use by agreement). Refs below.
Now you have rejected the nominalist/formalist account. You say you are
not a Platonist/realist. Are you then an intuitionist? It has some
consequences. Quine says it "...countenances the use of bound variables
to refer to abstract entities only when those entities are capable of
being cooked up individually from ingredients specified in advance."
This means, among other things, that it cannot deliver full Cantorian
set theory.
>
>
> >>I'm damned if I can see how one is "concrete" but the other isn't.
> >>And that was -- if you recall -- what John was claiming early on:
> >>that Lisp was a "Pnatural language" but lambda-calculus wasn't.
> >
> >But that wasn't what John was claiming. That is the problem; he
> >simply wasn't saying what you say he was saying. I'm not going to
> >argue with about he said or meant though - John can speak for himself.
>
> On this, sir, I sincerely beleive you are mistaken. This (the t.o)
> discussion didn't start about numbers. it started about John's mention
> of "natural" languages. If you will allow me to skip over the
> distinction between a `language' and utterances in that language,
> since that's what John's first two or three posts did.
> John also said, later that Lisp was a natural language
> [_sensu_Wilkins], but that lambda-calculus was not.
Because LC cannot be identical with symbols in physical form (while Lisp
is, I believe). I already explained the play on "natural" and it got a
Harter seal of approval, so I am satisfied.
>
> I will check, as soon as I can.
WV Quine, 1953. On What There Is. _From A Logical Point of View_, all
cites from p 192 in _Philosophy of Mathematics: selected readings_ eds P
Benacerraf and H Putnam, Oxford, Blackwell: 183-196
P Bernays, 1935. On Platonism in Mathematics. also in Benacerraf and
Putnam, 274-286.
* Unintentional but appropriate pun.
John Wilkins
> In article <1f5r27r.5bxqx574fed4N%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> >Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
>
> >I know very well where unicorns exist - as uninstantiated complex
> >semantic descriptions, and I know where semantic descriptions occur, too
> >(in heads and societies of heads). But unicorns and ink about unicorns
> >are not the same. Ink and utterances exist, as physical cases.
>
> Wow. At last I think we've gotten beyond talking past each other.
> Does this mean you're feeling better?
Somewhat. I can go without coughing up my alveolae for as much as 30
minutes at a time. I may get some sleep tonight.
>
>
>
> >> English is a wonderful language; it leets one write things which have
> >> no intension. John, where do numbers exist? If their definitions were
> >> destroyed, are you saying numbers would cease to exist, too?
> >
> >Yes. I am saying exactly that. Remove all counting entities from the
> >universe - that is, all systems capable of instantiating a definition of
> >number - and there is no number in the universe.
>
> I really meant _just_ the defintion, not the competence too; you
> answered a question I asked a bit later.
>
> Okay, next question: suppose for the sake of argument that we, or some
> other species, rediscover an isomorphic concept. Is that the _same_
> concept, or a different one? How does that extend to independent
> discoveries of what we decide is the same idea?
Ask yourself how we could determine if they were the same? What language
could we use to ask that question in?
In general, Good Tricks (one of the few Dennettisms I like) will be
rediscovered in similar enough circumstances.
>
> How 'bout Marty's question about Fermat's last theorem?
Are you sure that we *did* rediscover Fermat's last theorem? From what I
heard, it was a fair bit more than just a bit too much to include in a
marginal note.
>
>
> >Tree rings exist. Ideas exist. But ideas of tree rings exist only when
> >there is a head capable of having that idea.
>
> So far so good. Are numbers _just_ ideas? Is it meaningful to ask
> about the number of tree rings in some point in time before the
> idea `numbers of tree rings' is formed?
There are physical structures there, mind-independently. There is a
"number" of tree rings only when they are counted. Yes, numbers are just
ideas? What do you say they are?
>
> [snip stuff answered above]
>
>
> >> Was the Latin language destroyed when all native speakers of it died?
> >
> >No, because non-native speakers continued to exist. But if all that
> >remained of the Latin language was the name, would the Latin words and
> >their meanings "exist"?
>
> Somewhere about where, I asked you about Rosetta stones. I'm sure I
> did. But none of this tells me why Lisp exists as `concrete',
> but lambda-calculus does not.
>
>
> >Does that mean if I define arbitrarily and
> >intensionally some abstract language Google (hey, they do such wonderful
> >work for the internet community, I say pay them back :-),
>
> I can pass that on ot Larry and Sergey next time I see them,
> if you'd like.
Sure, but I'm not paying no royalites nohow.
> >
> >See the slide? I do not deny that the *uses* of LC are bounded. Of
> >course they are - so is your blackboard with LC symbols. But LC itself
> >isnot.
>
> I tell you, three times now, that there are computer programming
> langauges which are explicilty implementeed via lambda-calculusn
> reductions. Will you now abandon your original claim that Lisp is a
> `natural' language but lambda-calculus is not? Will you concede that
> your original claim, that LIsp was concrete but Lambda-calcus is not,
> was incorrect?
>
> A yes or no will suffice.
No *and* yes. See my response to Marty elsewhere. I should have said
from the beginning that abstract objects do not *exist* - they have no
ontological standing. But it seemed to me I could not assert that at the
beginning of the argument. Perhaps I should have.
My misremembering (I didn't cheat...)
>
>
> >Linear B would no longer exist anywhere. Its boundaries
> >would have been finalised, as it were.
>
>
> I am *almost* willing to buy this, along with the obvious
> generalizations. The caveat is that modern cryptanalysis has gotten
> to the point where significan progress can be mad even in the complete
> absence of speakers, and of cribs like the Rosetta stone.
Not if there's nothing preserved. Information *can* be destroyed, you
know (oops, sorry, wrong thread).
>
> It may e possible for a langauge to die, to be left with nothing
> more than a buncho f artifracts -- a little more than the Phaistos disk,
> or Orongorongo, to be sure -- along with context.
>
> Suppose, for the sake of argument, that sup-dup computers somehow
> manage to correate the bits and come up with putative translations
> that make sense in context; and that also make sense in
> newly-discovered, heretofore unknown artifcats.
>
> What would your philosophy say about that?
Sure. [Caveats about background assumptions about language and human
needs, etc.] But that wasn't my point - I was saying that if all record
of Linear B (m.m.) were lost, then Linear B would no longer exist.
>
> >Well, I do look forward to what he has to say. I'm sure, being a
> >philosopher, he'll be able to find things to attack. But I am most
> >interested in what *you* have to say. Clearly and distinctly now - do
> >abstract objects exist independently of their concrete instantiations or
> >not, the way, say, your fossil tree rings did? Answer that, and we'll
> >move on from there.
>
> You first. Concede your initial assertion that Lisp was concrete, and
> Lambda-calculs was abstract, was not in fact correct, and I'll email
> you what *I* wrote.
Don't bother - I assume you won't tell porkies.
I said it already in my reply to Marty: LC and Lisp are concrete only;
abstractions do not exist except as concrete acts; there is (ie,
existential quantifier) no such physical or real thing as a formal
language in the ontological sense.
Ha! Used to *dream* of finishing USENET by 1am <peters off into four
Yorkshiremen sketch, and other analgesiacisms>
John Wilkins
> On 9 Jan 2002 03:15:39 -0500, jona...@DSG.Stanford.EDU (Jonathan
> Stone) wrote:
>
....
> >
> >and just how does
> >
> > "English exists"
>
> English "exists" because there are speakers of English, works written
> in English, dictionaries, et cetera. The word, English, is a label
> for things going on in the "real world".
>
> >entail different ontologocal baggage from
> >
> > "Lisp exists"
>
> I think what he is saying is that Lisp exists in the sense of being a
> computer language extant in the real world with manuals,
> implementations, et cetera, whereas he is denying that it exists in
> the sense of there being an abstract Lisp.
Correctamundo.
>
> >and
> >
> > "lambda calculus exists"
>
> Here he is denying that the lambda calculus, considered as an
> abstraction, exists. It is arguable that he is not being consistent
> because, after all, there are books and such like about the lambda
> calculus. However his point is that it is conceived of as being an
> abstraction and as such by his lights doesn't exist. Mind you, he is
> willing to concede the existence of the idea of the lambda calculus,
> just not the thing itself.
>
> [These interpretations subject to correction by John - it's his baby.]
You are its godfather then. I wish I could express myself so clearly.
Damn.
Agreed: *as an abstraction* LC does not exist.
>
> >because I have read John most carefully, and he is saying that one of
> >those three is ontologically not like the others.
>
> As noted elsewhere in other words, the word "exists" is somewhat of a
> weasel word. Quine begins one of his essays with the observation that
> the question of ontology can be summarized in three words - what is
> there - and answered in one - everything. He notes, however, that
> there is some room for arguing about cases. Somewhere, perhaps in the
> same essay, he expresses the view that it is quite all right to apply
> to "exists" to abstract entities; lacking physicality does not disbar
> them from existence. One can defend this view on the grounds that in
> formal logic it does not matter if the entities being referred to are
> physical or not. I opine that this argument is correct but
> misleading. None of the entities in formal logic are real.
Yes. I think we are clear now.
I like the way Quine refers to "those of us who have a tase for desert
landscapes". Someone, I forget who, accused me of having a desert island
ontology; I replied that I like to lie under the shade of the
ontological palm trees, but I know they grow from the sand.
Aatu Koskensilta
Jonathan Stone <jona...@DSG.Stanford.EDU> wrote
[big snip of Stone defining language as a set of finite strings
recognised/produced by an automaton]
> Do mathematicians have some other formal definition of "language"?
[snip again]
Well, if you want to be abstract, you can use any set you wish. Of course,
to be useful, certain formation operations are usually assumed to be given
(negating a sentence, constructing (not necessarily finite) conjunctions,
&c.).
In situations occuring in computer science and related area one does
not really need anything more complex than a (r.e. or recursive) subset
of the set of finite strings over some alphabet (usually finite or
countable).
This comes down to essentially Stone's definition.
--
Aatu Koskensilta
"Wovon man nicht sprechen kann, daruber muss man schweigen"
- Ludwig Wittgenstein, Tractatus Logico-Philosophicus
Aatu Koskensilta
Jonathan Stone <jona...@DSG.Stanford.EDU> kirjoitti
viestissä:a1gglr$vg5$1...@Pescadero.DSG.Stanford.EDU...
> In article <3c3b...@news.qut.edu.au>,
> Chris Ho-Stuart <host...@sky.fit.qut.edu.au> wrote:
[snip]
> >We can also speak of non-recursive languages, not accepted by
> >any automaton, or even a Turing machine.
>
> Uh... can we? How are such languages defined? Is this
> semi-computability, or something completely different?
Consider a very simple uncountable language defined as follows.
1. The ordinary formation rules of 1st order logic apply
2. In addition, if Phi is a set (*not* necessarily finite) of sentences,
so are /\Phi and \/Phi.
We can define semantics for this language (getting a logic) as follows:
1. M |= phi iff M |= phi according to the usual semantics of 1st order
logic
2. M |= /\Phi iff M |= phi for every phi \in Phi
3. M |= \/Phi iff M |= phi for some phi \in Phi
We could now take, say, first order PA and augment it with the
sentence "for all x, \/{S^n(x) | n \in N}". (Notice that this is not
actually a "sentence", but a description of a sentence obtained
by the formation operations. Just as you can consider 56+3^3
a description of an element of a language for representing
numbers.). This would give us a categorical theory in our newly
constructed logic.
And no, the language is not "computable" in any sense, since it's
uncountable. This sort of languages occur frequently in mathematical
logic (though not at the very elementary level).
Marty Fouts
john.w...@bigpond.com (John Wilkins) writes:
> Marty Fouts <usene...@usa.net> wrote:
>
[snip]
> > Is amphiboly here meant to construe something other than merely
> > 'ambiguous sentence'? (Does it matter or am I just being
> > pedantic?)
>
> I was being pleonastic :-) [Gad, a fellow can't have *any* fun among
> this crowd. Pedants, all of them.]
OK. So long as you weren't just being redundant, it's OK. ;)
[snip]
> Hmm. I was taught that natural languages were things that could not
> be explicitly cast into formal language terms (ie, 2nd order logic,
> etc) without ambiguity. But undergrad logic is 20 years ago, and I
> have made a practice of avoiding logic where possible ever since. I
> was never going to contribute to that field.
OK. That works in that DoD.
[snip]
> > Being pedantic, I will note that all computer programming
> > languages fail your definition of formal, as even the most
> > precisely defined contain ambiguity.
>
> Truly? I didn't know that.
Yes. Pedantically speaking, the formal definitions of computer
languages are always incomplete, and so implementations always have
room to be different and so to produce different result for the same
source code. This is codified in language standards descriptions by
the way. There's a whole jargon about 'undefined', 'implementation
defined' and 'conforming' and 'strictly conforming' that highlight
it. (Guess who spent too much of his life on programming language
standards committes?)
> In that case can we say that no computer language is reducible to
> the lambda calculus that Jon is so fond of, or does that also have
> ambiguity in it?
That would depend on what we mean by reducible. For every formally
defined programming language there is a subset that is not
ambiguous. In the case of lisp, the subset is reducible to the lambda
calculus. (McCarthy's original description of lisp, IIRC, is reducible
to the lambda calculus, for instance.)
[snip]
> > Ih this model are subjective properties, such as color, concrete,
> > because existing in the mind of the observer they exist in
> > spacetime, or abstract, because differing from observer to
> > observer they are not fixed?
>
> Insofar as they are properties (please, can we *not* get into philosophy
> of mind here? I have the flu. I refuse to consider philosophy of mind
> while I have the flu), they are physical. I suspect that subjective
> "properties" are not properties, strictly speaking.
OK. We'll skip philosophy of the mind until you're better; but I think
you're eliding a fairly important point. If 'red' is not a property of
my red sweather, what is it?
[snip]
> > But I think there's a Chinese room around the corner. If
> > implementation on a computer counts as concrete, then does a
> > computer program that does voice recognition and text-to-speach
> > generation make English concrete? If so, and I teach the program
> > to 'speak' lambda calculus notation, how does that differ?
>
> My Q&D solution would be that the Chinese room understands some subset
> of English, and that if that were all the English in the world, then
> that *would* be "understanding English" - there's be nothing else to
> contrast it to. It appears I am a formalist after all.
Fair enough. Since we're not doing 'mind', I'll skip the comment
that all speakers of natural languages only know a subset of the
language.
[snip]
> No - you are quite right; Lisp is a natural language after all. I'll now
> stick with that. It is natural because the abstract entity "Lisp" is a
> fiction. Abstraction is a moment in representation, not a state of being
> :-)
Drat. The laughing ghod in the corner notes that you've found the
escape clause in the question about Fermat's theorem.
[snip]
> > well at least your's are merely obscure and you understand
> > them. mine are obtuse and even I don't get them.
> >
> > Marty
>
> No... I begin to think that mine are equally obtuse ;-)
Well, you (and dirt) are only a few days older than me, given your
signature, but I was precocious, so I guess I'll accept that.
> --
> John Wilkins
> Occasionally making sense for over 46 years
Marty
(Occasionally making lunch for over 46 years)
Marty Fouts
john.w...@bigpond.com (John Wilkins) writes:
[snip]
> There are physical structures there, mind-independently. There is a
> "number" of tree rings only when they are counted. Yes, numbers are
> just ideas? What do you say they are?
[snip]
The math community, as Richard Harter has outlined, has toyed with
that question for some time. The consensus is that they can't
agree. In simplest terms there are two schools with different answers
to that question (I bet I'm missing a third:)
1) 'number' is a property of real things, like color, and mass, that
we model with absractions like the Peano axioms.
2) 'number' is an abstraction. Wanna see my Peano axioms?
The basic difference hinges on the answer to the question from
philosophy of mathematics as to whether mathematicians are discovering
properties of the universe or inventing abstractions. As far as I know
the debate on that still continues.
Marty
Jonathan Stone
Aatu Koskensilta <anna.t...@kotiposti.net> wrote:
>
>Jonathan Stone <jona...@DSG.Stanford.EDU> wrote
>[big snip of Stone defining language as a set of finite strings
>recognised/produced by an automaton]
>> Do mathematicians have some other formal definition of "language"?
>[snip again]
>
>Well, if you want to be abstract, you can use any set you wish. Of course,
>to be useful, certain formation operations are usually assumed to be given
>(negating a sentence, constructing (not necessarily finite) conjunctions,
>&c.).
Huh? Negating a sentence? Conjunctions? You are thinking of a
logic (or other semantics for a language), not the notion of a
`formal language'.
>In situations occuring in computer science and related area one does
>not really need anything more complex than a (r.e. or recursive) subset
>of the set of finite strings over some alphabet (usually finite or
>countable).
It's not computer science. The usage goes back to Chomksy. This is
the first time I've ever run into other uses of `formal language',
even when talking with cognitive-scienctists or mathematicians or
logicians. Live and learn, I guess.
>This comes down to essentially Stone's definition.
It's not my definition; the only thing I should get attribution for is
any errors I introduced into the definition.
Jonathan Stone
John Wilkins <john.w...@bigpond.com> wrote:
>Jonathan Stone <jona...@DSG.Stanford.EDU> wrote:
[...]
>Because LC cannot be identical with symbols in physical form (while Lisp
>is, I believe).
No. I offer you a point of fact: you are *WRONG* about this. As a
simple empirical fact, concrete Lambda-calculus can be and *is* used,
in exactly the same way that concrete Lisp is used.
Empirically, the difference which you wish to impose between the
two langauges just doesn't exist. Period. endit. Lambda-calculus
can, and is, used as a fully concrete language in *EXACTLY* the
same way that Pure Lisp is used as a fully concrete language.
Contrariwise, Pure Lisp is sometimes used for the formal purposes
(e.g., defining another langauge), for which Lambda-calculus is
traditionally preferred.
Your ontological partitioning of one from the other is empirically
refuted. It's just *wrong*. At the very least, you have to look at
the usages and decide which usages fall into one ontological bucket,
and which fall in t'other. You *CANNOT* do the partitioning based
solely on the syntax or the semantics of the language.
That is an empirical fact.
Now, can we acknowledge that, and move on?
Matt Silberstein
[snip]
>If 'red' is not a property of my red sweather, what is it?
It is the context, culture, and psychological state dependent name
given to an particular subset of aspects of the sweater. It is "my"
name for "my" perception, not a property of the sweater itself.
Or so it seems to me. I am, AFAICT, a rather severe nominalist myself.
However, unlike some/many/most/all of the people posting to this
thread I do not expect strict consistency of position. Our
thoughts/ontologies/science/epistemologies will not ever map perfectly
to "reality". Neither the ones in your brain nor the ones you utter.
OTOH that may well be obvious and/or irrelevant.
[snip]
Aatu Koskensilta
Jonathan Stone <jona...@DSG.Stanford.EDU> wrote
> In article <a1i358$hqv$1...@news.koti.tpo.fi>,
> Aatu Koskensilta <anna.t...@kotiposti.net> wrote:
> >
> >Jonathan Stone <jona...@DSG.Stanford.EDU> wrote
> >[big snip of Stone defining language as a set of finite strings
> >recognised/produced by an automaton]
> >> Do mathematicians have some other formal definition of "language"?
> >[snip again]
> >
> >Well, if you want to be abstract, you can use any set you wish. Of
course,
> >to be useful, certain formation operations are usually assumed to be
given
> >(negating a sentence, constructing (not necessarily finite)
conjunctions,
> >&c.).
>
> Huh? Negating a sentence? Conjunctions? You are thinking of a
> logic (or other semantics for a language), not the notion of a
> `formal language'.
I'm thinking of one of the notions of 'language' that occur in mathematics.
Of course, such languages are not necessarily formal languages in the
sense you mean.
However, the situation is not quite what you make it to be. Calling the
formation operations 'negation', 'conjunction', &c. I were obviously
thinking of a language to be used in a logic, but this need not be the
case.
For example, we may deal with the language of the 1st order
logic as a formal language, in which case the unary operation
called 'negation' comes down to prefixing a given string with some
symbol, say ~. Every formal language can be construed as the
closure of some set under some operations, if I'm not gravely
mistaken.
> >In situations occuring in computer science and related area one does
> >not really need anything more complex than a (r.e. or recursive) subset
> >of the set of finite strings over some alphabet (usually finite or
> >countable).
>
> It's not computer science. The usage goes back to Chomksy. This is
> the first time I've ever run into other uses of `formal language',
> even when talking with cognitive-scienctists or mathematicians or
> logicians. Live and learn, I guess.
I did not mean to imply that the only area in which such languages
frequently occur is computer science. Although I would be inclined
to consider automata theory and the related formal languages a
part of computer science. This naturally reflects nothing but my
taste.
> >This comes down to essentially Stone's definition.
>
> It's not my definition; the only thing I should get attribution for is
> any errors I introduced into the definition.
Again, I apologise for being inaccurate.
Aatu Koskensilta
I wrote
In order to clarify this a bit, I'll note a few things. In mathematical
logic the language together with the semantics and a system of
deduction would be considered a full logic. However, although
the language was used as a part in the definition of a logic, the
language itself does not have any semantics associated to it.
I gave the semantics merely to motivate the construction of
such an uncountable language. This is a case where a structure
clearly is a language, although not a formal language in the
automata theory or formal language theory sense of the
word, and does occur in actual mathematical practice.
Marty Fouts
john.w...@bigpond.com (John Wilkins) writes:
> I do not have a place for abstract entities in my poverty-stricken
> ontology. They do not "exist". My argument is that the properties
> ascribed to abstract symbolisms and procedural languages are in fact
> not causal. They do nothing.
Could you slow this one down and run it by me again? Why, in your
ontology, do properties have to do something rather than simply be
something? To a physicist it would appear that you just said that
mass does not exist because it does not do something...
> We abstract the (usually syntactic) properties of systems in terms
> of formalisations (ie, what we find similar in the way we describe
> them), but these are (useful) fictions.
OK. I sort of follow you. 'mass' is an abstraction in physics, yet the
abstraction exists because the universe contains a 'real' property
that the abstraction is well suited to describing.
In this sense, then, are you saying that abstractions don't 'exist'
independently of the abstractor? IE, there might be ETs who's physics
doesn't have 'mass' as a proprety because they've sliced the problem
up differently, and so 'mass' doesn't really exist, even though the
'real property it describes does?
[snip]
> English is, as I said, a historical object. If we leave the room to
> - say - Hindi speakers, then it is bounded elsewhere (where we, and
> the rest of the English language community exist as physical objects
> in relation to each other). But do to it what was done to Estruscan,
> and it may very well vanish.
This is approaching the edge of the problem that interests me in this
area. English is an historical accident. It need not have existed. But
is the same thing true about pi? Is it really possible for the ETs in
my example to have generated a physics without 'mass'?
It is not clear to me that some properties that we think of as
abstractions and non causal are not in that sense 'real'. The sense,
to be redundent being that any observer that came along with the right
sensory equipment could discover them. ('color', for example is
subjective, but 'reflecting photons of a certain frequency' is not)
> >
> > Praps you should go back to bed until you are feeling less Searle-ey.
>
> I've been accused of many sins, but being in agreement with Searle ain't
> one of them. I think you are misreading me through some spectacles that
> filter out my egregious and extreme nominalism.
> >
> > (PS: I do hope that gets a laugh. I'm no Harter, but I am not quite
> > as unsophisticated as you think :)
>
> It rated 1.2 on the groan scale.
>
Well, He could, nominally, have written 'Searl-ey' you must be joking.
Marty
Matt Silberstein
> In article <1f5phcy.blupj7iuvl7aN%john.w...@bigpond.com>,
> John Wilkins <john.w...@bigpond.com> wrote:
> >Matt Silberstein <mat...@ix.netcom.com> wrote:
>
>
> >Yeah, now why is that? ISTM you make good contributions to the flow of
> >the discussion, and you know a fair bit more than I do about computation
> >and programming. Yet Jon seems to treat me with a dignity I don't
> >deserve, and you with vitriol you don't.
>
> You don't pretend to know far more about it than you really do. You
> don't squink for years over elementary mistakes.
You mean like your claim that there was a scientific definition of
life? Or that there is something wrong with my defining as system as
my left hand and right thumb? And if I had made the unforgivable
errors in the past what does that have to do with the post in
question?
> I've never had friends visit town and ask, over dinner, if I had any
> more amusing stories about gaffes you made, then lied about.
I've never seen anyone in this group suggest that *John* needs to
check his medication levels.
> Perhaps that has something to do with it.
Perhaps. Perhaps it has to do with your lack of understanding of the
difference between dealing with an argument and dealing with a person.
And perhaps it has to do with your misunderstanding some things years
ago. And rather than admit you might have made an error you decided
that I am worthless and contemptable.
Matt Silberstein
> In article <4oin3u8u6j17hcdtn...@4ax.com>,
> Matt Silberstein <mat...@ix.netcom.com> wrote:
> >In talk.origins I read <a1fmnu$sad$1...@Pescadero.DSG.Stanford.EDU> from
> >jona...@DSG.Stanford.EDU (Jonathan Stone):
>
>
> >>>So what if he can't (currently) produce this algorith?
> >>
> >>Then for all practical purposers, there *is* no algorithm. And it is
> >>intellecutally dishonest (or, in this instance, if you prefer,
> >>Platonic) to proceed as if there was.
> >
> >Not at all. It is entirely possible that someone could show that a
> >finite algorithm exists but that we don't currently know it.
> Wrong. First they have to show that the collection of processes
> lumped under hte rather vague and general English word `design'
> *is* an algorithm, rather than (as don Knuth once put it)
> an Art, rather than a science.
If he/I were making a claim on the basis that such an algorithm
existed and was known, then showing the algorithm is important. He
made no such claim. He wonder if design were an algorithm and wonder
what the implications would be if it were.
I wonder why it disturbs you so to consider design as an algorithm. Do
you think that you engage in some magical process not subject to
finite systems?
> [...]
>
>
> >No, I am asking because I disagree. I am asking you to make an
> >argument for why his current ability to produce the algorithm is
> >relevant. Asserting it is not such an argument.
>
> Unless Bobby can show an algorithm, and show that the algorithm *is*
> what we call `design', he(?) cannot establish that the collection of
> human procesases *is* in fact that algorithm, or indeed that it is
> any `algorithm' at all.
What alternative do we have?
> That is precisely what Alan Turing's original Turing test is all
> about: a experimental protocol to decide whether computers (i.e.,
> algorithms, algorithms in execution) can pass as human to other
> humans, to the point where a human conversing iwth a compute and
> with a human cannot tell which is which.
The tests is about defining what it means to be human, IMHO. (And
likely about what it means to be male and female, but that is a
different discussion.)
> [...]
There was an important point snipped here. In terms of the larger
discussion here it does not matter if design is an algorithm as
strictly defined nor is it important whether the brain is (almost) a
UTM. It matters if design is some finite mechanical process. I assert
that the brain is a machine and that what the brain does other
machines can do. If this is beyond a UTM, so be it. I don't much care
whether or not this fits some category established for CS or for math.
The default assumption in science has to be that the brain is simply a
mechanical device. And other mechanical devices can likely do
something sufficiently similar. If someone wishes to claim otherwise
the burden is on them to show that the brain is somehow special.
This brings up another problem with Dembski's use of the NFL.
Evolution like search algorithms exist. But it is not at all clear
that evolution, as done by biological populations, is an algorithm. I
suspect that we can more easily show that human design is an algorithm
that we can actually show that biological evolution is algorithm. We
can model aspects of evolution with algorithms, we can build
evolutionary algorithms to do (more or less efficient) searches. But
that does not mean that the NFL theorem comes into play.
In addition, I just want to repeat that the standard of efficiency for
math/CS algorithms is not the same as that for biology. In math/CS we
want the solution that provides the best answer in the fewest steps.
In the real world we want the solution that provides the best answer
in the time available. It is not clear that the answers are the same
for both.
> >>If you cannot state an algorithm, then what reason is there to believe
> >>that the collection of human intelligent process we call `design' is,
> >>in fact, an algorithm?
> >
> >This is a more relevant question and different than the argument you
> >made above.
>
> Excuse me. I thought it was the *same* argument, only made explicit
> rather than likening `desing' to other humn intellectual and creative
> endeavours, which are generally regarded as being non-algorithmic.
It is not the same argument. Your first argument was that since Bobby
could not produce the algorithm, he had no right discussing it. Here
you ask for reasons why he might consider it an algorithm. Those are
very different approaches.
> >He asked a question. And then proposed that if the answer were one
> >thing then we would have a particular result. What this has to do with
> >my knowledge of science escapes me.
>
> I read Bobby's text as more of a rhetorical device device.
Well I read it differently and he has explicitly said he meant it
differently. You are free to choose your original reading over the
words of the author.
> I read it
> as appealing to intution that `design' really *is* an algorithm.
> Perhaps I was mistaken. But that is certainly how the argument
> seemed to continue.
How it seemed to you is quite distinct from how it seemed to me. And
the author says he agrees with me.
> If Bobby wants to be scientific, then yes, I think the onus is on
> Bobby to show that `design' really *is* an algoirthm; and to abandon
> any rhetorical devices which make the if-then, well, rhetorical rather
> than substantive. If you disagree with that, then yes, I think that
> does reflect on your ounderstanding of science.
Here you are almost dead wrong. It is acceptable in science to explore
an issue before you look for evidence. It is wrong to see that
exporation as evidence or conclusive. But a key point of science is
looking at an "if A then B" and then trying to find if the As and Bs
exist. We explore the implications of whether design is an algorithm
or not first. If there are no implications, we don't bother to figure
out whether it is. Only if there are implications do we bother to
explore.
>
> > If you wanted to you could have
> >asked him to show it is an algorithm. But there are other ways to do
> >that in place of producing the algorithm.
>
> Really? Like what?
I could show that design is a member of some class of things that are
algorithms.
> >>The most natural way to do so is to give the algorithm.
> >>This is a step Bobby is conspicuously avoiding.
> >
> >That is one way, but not the only way.
>
> Again, like what? We have a fairly large set of human activity lumped
> under the very general term `design'. We have the hypothesis that
> this activity is an algorithm. The only way *I* know to show that is
> to give some algorithm, then demonstrate that the process of
> excecuting the algorithm, and the human behaviours we call `design',
> really are the same.
>
> What other ways did you have in mind?
I could show that the mind is the equivilant of a computer (i.e. close
to a UTM). Then if people design, design is an algorithm.
>
> [...]
>
> >>I think you missed the point of the question -- which was to show
> >>that, scientifically, we don't have sufficient grounds to assume
> >>either way.
>
> >>>> How is `design' any different in this respect than writing
> >>>> poetry, painting pictures, or sculpting scupltures, or writing music,
> >>>> or for that matter, doing science?
> >>>
> >>>A good question, but a good question does not an argument make. I
> >>>suspect that for the purpose of the NFL and this discussion they are
> >>>all sufficiently similar.
>
>
> Me too, but neither your suspicions nor mine really cut it.
Which is why all I offered is my suspicions. Absent an answer the
default is we don't know. If we don't know we can try to see if the
question is solvable or solved or if it is important.
> As far as I know, attempts to reduce painting and storytelling to
> algorithms haven't gone much beyond the ancient 'spew' program, which
> generated fake national-enquirer headlines and ... What was the other
> version? Candidates for `Dukes of Hazzard' plotlines? ... by plugging
> phrases from a dictionary together. I once got the output
>
> ``Evil aliens are the parents of my love child, says Joan Rivers--
> exclusive pictures show all!
>
> which doesn't bode terribly well for Bobby's hypothesis. :)
Why? How does the current state of the art say a thing about his
hypothesis? What we can do now does not tell us what can't be done. We
know that design can be done by some machines so it is likely it can
be done by others.
> >>I'm looking at this again, with the assumption that it's ill-informed
> >>rather than hostile.
>
> I hope John w. has caught up with this one.
>
> >>In that light, a better answer is: no, that's not how the standard
> >>terminology in the field works. Computers *are* models of UTMs,
> >>and we can approach a UTM to any arbitrary but finite size.
> >
> >Yep, models of. Agreed. They are not UTMs, I hope we can agree on
> >that. I took your point to mean that since current computers don't
> >design, design was not an algorithm. [continue below]
>
> No. I'm saying that there is no *demonstration* that `design' is an
> algorithm; that the current research is not even *beginning* to
> automate many of the sorts of things humans do that we call design;
> and that therefore, there is no reason to put much stock in the
> hypothesis that what we call design *is* expressible as an algorithm.
Why do you say "No" when you seem to agree with me. You say our
current abilities inform us about whether design is an algorithm. I
say our current abilities do no such thing.
> [Matt contiues on the same line]
>
> >You seemed to argue that real
> >computers people really have can only do these things in toy domains.
>
> The reference to "toy domains" was __completely_ separate.
It followed in the next sentence in the same paragraph. Here is that
paragraph:
"If you will, for the sake of discussion, grant us the Church-Turing
thesis, what you are asking is whether any of these activities can be
produced or simulated by computers. The scientific answer to that
seems to be "yes", in certain microworlds (aka "toy domains") and "no"
in the general, everyday human-level sense of the world."
If there is a separation I don't see it.
> I've
> probably said here, several times, what i think about AI: that AI is
> what get gets done in AI labs, and once we figure out how to make
> computers do something, its no longer part of the province of AI. (I
> invented that independently, when asked in an AI class to define what
> AI was. I used to think that was a wry joke, until I heard other AI
> professors saying it. I can give names if you really want).
>
> Hey, Matt, look. The way I read this, it does seem that you mixed up
> several rather distinct points, from separate posts, into one
> mish-mash; and then attributed the mish-mash to me.
It was one post, one paragraph. If they were separate points you
should not have worded it the way you did.
> I'm *trying* to be nice.
I suspect you are telling the truth here. More the same for you.
> But you seem to have a real talent for taking
> things I've said,mashing them, and repeating the mash being what i
> *acutally* said. I will take it on faith that you really *think*
> that's what I said. Well, I didn't, and I find the misrepresntation
> damn' offensive.
I don't particularly care what you find offensive. Considering they
way you treat me and my posts your feelings are not exactly my top
concern. I do care whether or not I mispresent something. If you see
an actual mispresentation point it out. So far you have not done so.
>
> >And that since they could not do them in real domains, that said
> >something about whether it was an algorithm.
>
> No, I never said any such thing. And again, I find your paraphrase and
> its attribution to me, to be really, really, *really* offensive.
I point again to the paragraph quote above. If I have misunderstood
please make your point clearer.
> My point was quite explicitly about `toy domain', and how that
> referred more to AI domains which were well-enough understood that
> they become problems in textbooks; and more, that once a problem
> becomes a `toy domain', it stops being AI (_sens Jonathan_).
The paragraph in question said *nothing* about AI in particular. If
you meant to comment on AI and not on design as an algorithm you did a
terrible job. I suspect you have actually forgotten what you wrong.
> I don't know *why* you misread me so grossly, but I really *do* find
> it offensive. Also, I don't see John Wilkins making misunderstandings
> like that and attributing them to me. I really don't.
> > Appealing to the current
> >state of the art in computing does not tell us whether or not design
> >(or art or whatever) is an algorithm.
>
> No, it does not. but it does reflect the current scientific state of
> the art on such subjects.
Agreed. This is not what you have said before and is irrelevant to
Bobby's point and, AAMOF, rather trivial.
[snip]
> >That was why I made the point
> >about UTM vs computers. In the context you clearly meant real
> >computers that exist today running existent software.
>
> Yes, I did mean real computers. I dont know of any other kind of
> computers -- _mutatis mutandis_, that is, as technology changes
> for different points in time as `today'.
The Church-Turing thesis does not apply to real computers.
[snip]
Matt Silberstein
> john.w...@bigpond.com (John Wilkins) writes:
> > I do not have a place for abstract entities in my poverty-stricken
> > ontology. They do not "exist". My argument is that the properties
> > ascribed to abstract symbolisms and procedural languages are in fact
> > not causal. They do nothing.
>
> Could you slow this one down and run it by me again? Why, in your
> ontology, do properties have to do something rather than simply be
> something? To a physicist it would appear that you just said that
> mass does not exist because it does not do something...
I suspect that John will object to the idea that things have
properties. (And if he does not, I will.) Properties are abstractions,
things just are. Your sweater does not have the property of being red
and the property of being warm. We humans abstract aspects of the
stuff and give those names.
> > We abstract the (usually syntactic) properties of systems in terms
> > of formalisations (ie, what we find similar in the way we describe
> > them), but these are (useful) fictions.
>
> OK. I sort of follow you. 'mass' is an abstraction in physics, yet the
> abstraction exists because the universe contains a 'real' property
> that the abstraction is well suited to describing.
>
> In this sense, then, are you saying that abstractions don't 'exist'
> independently of the abstractor? IE, there might be ETs who's physics
> doesn't have 'mass' as a proprety because they've sliced the problem
> up differently, and so 'mass' doesn't really exist, even though the
> 'real property it describes does?
I could imagine, at the dangling end of my imagination, some ET who
could (successfully?) do this. I agree that it does not make much
sense. But more to the point I can imagine some ET who does not make
the life/not life distinction we make. Yet life seems to be a
"property" of some stuff. That the Universe seems to have some easy to
find groupings does not mean that the stuff has natural categories.
> [snip]
>
> > English is, as I said, a historical object. If we leave the room to
> > - say - Hindi speakers, then it is bounded elsewhere (where we, and
> > the rest of the English language community exist as physical objects
> > in relation to each other). But do to it what was done to Estruscan,
> > and it may very well vanish.
>
> This is approaching the edge of the problem that interests me in this
> area. English is an historical accident. It need not have existed. But
> is the same thing true about pi? Is it really possible for the ETs in
> my example to have generated a physics without 'mass'?
There are all kinds of ratios between all kinds of things. (See
http://www.satirewire.com/charts/aolpie.shtml and
http://www.satirewire.com/charts/madrigals.shtml. Ok, so those are bad
examples.) Pi and geometric circles are properties, so to speak, of
human minds. Absent the existence of humans (and other equivilant
thinkers) pi does not exist. Absent the existence of humans etc. stuff
will, we all suspect, act pretty much the same, but the particular
aspect we call mass will not be abstracted and name. The stuff is the
same whether or not we talk about.
> It is not clear to me that some properties that we think of as
> abstractions and non causal are not in that sense 'real'. The sense,
> to be redundent being that any observer that came along with the right
> sensory equipment could discover them. ('color', for example is
> subjective, but 'reflecting photons of a certain frequency' is not)
That does not make it "real". Likely, important, yes, real no.
[snip]
Marty Fouts
mat...@ix.netcom.com (Matt Silberstein) writes:
> Marty Fouts <usene...@usa.net> wrote in message
> news:<usn9ey...@usa.net>...
> > john.w...@bigpond.com (John Wilkins) writes:
>
> > > I do not have a place for abstract entities in my
> > > poverty-stricken ontology. They do not "exist". My argument is
> > > that the properties ascribed to abstract symbolisms and
> > > procedural languages are in fact not causal. They do nothing.
> >
> > Could you slow this one down and run it by me again? Why, in your
> > ontology, do properties have to do something rather than simply be
> > something? To a physicist it would appear that you just said that
> > mass does not exist because it does not do something...
>
> I suspect that John will object to the idea that things have
> properties. (And if he does not, I will.) Properties are
> abstractions, things just are. Your sweater does not have the
> property of being red and the property of being warm. We humans
> abstract aspects of the stuff and give those names.
So sheep don't have feet, then, because 'foot' is an abstraction? And
water doesn't contain oxygen because 'oxygen' is an abstraction?
[snip]
> I could imagine, at the dangling end of my imagination, some ET who
> could (successfully?) do this. I agree that it does not make much
> sense. But more to the point I can imagine some ET who does not make
> the life/not life distinction we make. Yet life seems to be a
> "property" of some stuff. That the Universe seems to have some easy
> to find groupings does not mean that the stuff has natural
> categories.
I don't think that showing an example that doesn't really qualify as a
property (the life/non-life) distinction makes the idea of property go
away.
> > This is approaching the edge of the problem that interests me in this
> > area. English is an historical accident. It need not have existed. But
> > is the same thing true about pi? Is it really possible for the ETs in
> > my example to have generated a physics without 'mass'?
>
> There are all kinds of ratios between all kinds of things. (See
> http://www.satirewire.com/charts/aolpie.shtml and
> http://www.satirewire.com/charts/madrigals.shtml. Ok, so those are bad
> examples.) Pi and geometric circles are properties, so to speak, of
> human minds. Absent the existence of humans (and other equivilant
> thinkers) pi does not exist. Absent the existence of humans etc. stuff
> will, we all suspect, act pretty much the same, but the particular
> aspect we call mass will not be abstracted and name. The stuff is the
> same whether or not we talk about.
Your next to last sentence seems to contradict your last sentence. The
property that we care to label 'mass' will not go away because there
is no intelligence around to invent a label for it.
[snip]
> > It is not clear to me that some properties that we think of as
> > abstractions and non causal are not in that sense 'real'. The sense,
> > to be redundent being that any observer that came along with the right
> > sensory equipment could discover them. ('color', for example is
> > subjective, but 'reflecting photons of a certain frequency' is not)
>
> That does not make it "real". Likely, important, yes, real no.
What is 'unreal' about the demonstrable observation that light
reflects?
Jonathan Stone
>
>mat...@ix.netcom.com (Matt Silberstein) writes:
>
[...]
>So sheep don't have feet, then, because 'foot' is an abstraction? And
>water doesn't contain oxygen because 'oxygen' is an abstraction?
This leads, I think, to one saying that any paricular sheep has its
own feet, but that sheep in the abstract don't have feet in the
abstract. Particulars yes, generalizations no.
>[snip]
>
>> I could imagine, at the dangling end of my imagination, some ET who
>> could (successfully?) do this. I agree that it does not make much
>> sense. But more to the point I can imagine some ET who does not make
>> the life/not life distinction we make. Yet life seems to be a
>> "property" of some stuff. That the Universe seems to have some easy
>> to find groupings does not mean that the stuff has natural
>> categories.
What's this "we"? I dont think plasmas are a little bit alive. :)
>I don't think that showing an example that doesn't really qualify as a
>property (the life/non-life) distinction makes the idea of property go
>away.
This leads to saying that all properties are abstract, but some are
more abstract than others.
[wilkins continues:]
>>
>> There are all kinds of ratios between all kinds of things. (See
>> http://www.satirewire.com/charts/aolpie.shtml and
>> http://www.satirewire.com/charts/madrigals.shtml. Ok, so those are bad
>> examples.) Pi and geometric circles are properties, so to speak, of
>> human minds. Absent the existence of humans (and other equivilant
>> thinkers) pi does not exist. Absent the existence of humans etc. stuff
>> will, we all suspect, act pretty much the same, but the particular
>> aspect we call mass will not be abstracted and name. The stuff is the
>> same whether or not we talk about.
>
>Your next to last sentence seems to contradict your last sentence. The
>property that we care to label 'mass' will not go away because there
>is no intelligence around to invent a label for it.
I did tease John that he was confusing me with someone who confused
the word "two" and the numeral "2" with the number two. And John
replied, no, that was himself. And I think he was semi-serious.
Then again, I said [to paraphrase] that perhaps the ideas of
philosophers, that ontology and epistemology are `prior' to knowledge
about the world, had -- like so much of Greek metaphiscs --- been
overtaken by experiment and found to be a heap of foteid dingo's kidneys.
I dont think John took that seriously, either.
>[snip]
>
>> > It is not clear to me that some properties that we think of as
>> > abstractions and non causal are not in that sense 'real'. The sense,
>> > to be redundent being that any observer that came along with the right
>> > sensory equipment could discover them. ('color', for example is
>> > subjective, but 'reflecting photons of a certain frequency' is not)
>>
>> That does not make it "real". Likely, important, yes, real no.
>
>What is 'unreal' about the demonstrable observation that light
>reflects?
If I understood John's ontology, he's saying that without
intelligences around to articulate some sharable concept of
"light and "reflection" and "frequency", then no, we cannot say that
light of a certain frequency reflects.
My own view (and I suspect that John shares it) is that photons
in Aristotle's day behaved just as they did for Planck.
Perhaps John will explain how this works, in his ontology, without
resorting to Mr. Plato.
(Have you ever read the _Republic_? It's a farce, a laugh a minute.)
David Jensen
mat...@ix.netcom.com (Matt Silberstein) wrote in
<76998029.02011...@posting.google.com>:
>Marty Fouts <usene...@usa.net> wrote in message news:<uu1tv1...@usa.net>...
>
>[snip]
>
>>If 'red' is not a property of my red sweather, what is it?
>
>It is the context, culture, and psychological state dependent name
>given to an particular subset of aspects of the sweater. It is "my"
>name for "my" perception, not a property of the sweater itself.
Well, there is a property of the sweater, its ability to reflect light
of <knowable range of> angstroms, that we call red.
>Or so it seems to me. I am, AFAICT, a rather severe nominalist myself.
>However, unlike some/many/most/all of the people posting to this
>thread I do not expect strict consistency of position. Our
>thoughts/ontologies/science/epistemologies will not ever map perfectly
>to "reality". Neither the ones in your brain nor the ones you utter.
>OTOH that may well be obvious and/or irrelevant.
Certainly the map is not the place and the words are not the things but
there are places and things that are real.
John Wilkins
> john.w...@bigpond.com (John Wilkins) writes:
>
> > I do not have a place for abstract entities in my poverty-stricken
> > ontology. They do not "exist". My argument is that the properties
> > ascribed to abstract symbolisms and procedural languages are in fact
> > not causal. They do nothing.
>
> Could you slow this one down and run it by me again? Why, in your
> ontology, do properties have to do something rather than simply be
> something? To a physicist it would appear that you just said that
> mass does not exist because it does not do something...
Nooo... not exactly. Properties are effectively the same as predicables,
ie, they are verbal "things". Mass exists. "Mass" does not, except as a
term in someone's natural language [ObTolkein: Not more f***ing natural
languages].
Abstract properties (not properties in general) are not causal. The
*idea* of information does nothing (not even in genetics). The
properties of the genetic system that we *call* "information" is causal.
This is the use-mention distinction: what's in the quotation marks
doesn't do anything.
Incidentally, in Aristotle's metaphysics (which I am becoming
increasingly impressed by) a substance is anything that does not require
the existence of something else to have a property. This strikes me as a
good way to speak of things in this field. A cat is a substance, where
red is not. Properties are "had" by substances. Abstract things, not
existing, do not have causal properties because they are not substances
(which was A's mistake). A passing thought.
>
> > We abstract the (usually syntactic) properties of systems in terms
> > of formalisations (ie, what we find similar in the way we describe
> > them), but these are (useful) fictions.
>
> OK. I sort of follow you. 'mass' is an abstraction in physics, yet the
> abstraction exists because the universe contains a 'real' property
> that the abstraction is well suited to describing.
>
> In this sense, then, are you saying that abstractions don't 'exist'
> independently of the abstractor? IE, there might be ETs who's physics
> doesn't have 'mass' as a proprety because they've sliced the problem
> up differently, and so 'mass' doesn't really exist, even though the
> 'real property it describes does?
Possibly (in fact, IMO very likely). This is a representational theory
of theories, in effect, but the *theory* does very little except
represent things in the minds of abstractors. I have a little back story
about abstraction and representation; it seems to me that we are trying
to capture as much useful information in the smallest amount of brain
resources as possible. Any representation that on average does this
better than an existing one will tend to be regarded as better in
itself. But in the end, it's an economic tradeoff. We do have this
prevailing tendency to mistake the theory for the reality, we humans. I
wonder why that is...
>
> [snip]
>
> > English is, as I said, a historical object. If we leave the room to
> > - say - Hindi speakers, then it is bounded elsewhere (where we, and
> > the rest of the English language community exist as physical objects
> > in relation to each other). But do to it what was done to Estruscan,
> > and it may very well vanish.
>
> This is approaching the edge of the problem that interests me in this
> area. English is an historical accident. It need not have existed. But
> is the same thing true about pi? Is it really possible for the ETs in
> my example to have generated a physics without 'mass'?
Could they generate some other constant that makes their own
calculations of the areas and volumes of idealised geometric objects
simpler? Yes, depending on what "simpler" means in their context. But if
they have a language that approaches many of the same things we do, it
seems likely they will trick onto the same approaches we do. But ...
suppose they live their lives dealing with objects of high relativistic
mass. What would *they* think was obvious that we do not?
>
> It is not clear to me that some properties that we think of as
> abstractions and non causal are not in that sense 'real'. The sense,
> to be redundent being that any observer that came along with the right
> sensory equipment could discover them. ('color', for example is
> subjective, but 'reflecting photons of a certain frequency' is not)
I'm of the view that maths is unrealistically successful because we
select the logic that works well for us. If they live in the same
general "physical niche" that we do, they will very likely end up with
the logical that works well for both of us, and hence is
intertranslatable. The solution to the Lebensformen problem of
Wittgenstein is that we humans share a lebensform that we do not share
with lions. Hence we share the same general features of the universe,
cognitively speaking. If we do this with ETs, intertranslatability will
be feasible. [This was W's very heavily disguised answer, too.]
>
> > >
> > > Praps you should go back to bed until you are feeling less Searle-ey.
> >
> > I've been accused of many sins, but being in agreement with Searle ain't
> > one of them. I think you are misreading me through some spectacles that
> > filter out my egregious and extreme nominalism.
> > >
> > > (PS: I do hope that gets a laugh. I'm no Harter, but I am not quite
> > > as unsophisticated as you think :)
> >
> > It rated 1.2 on the groan scale.
> >
>
> Well, He could, nominally, have written 'Searl-ey' you must be joking.
>
> Marty
Such puns are universally atrocious. And don't call me Surly...
John Wilkins
....
> I've never seen anyone in this group suggest that *John* needs to
> check his medication levels.
But I do. Every six months or so. The doctor has suggested I give it
another six months.
[I have a *certificate* to prove I'm sane. How many people can say
that?]
Marty Fouts
john.w...@bigpond.com (John Wilkins) writes:
> Matt Silberstein <mat...@ix.netcom.com> wrote:
>
> ....
> > I've never seen anyone in this group suggest that *John* needs to
> > check his medication levels.
>
> But I do. Every six months or so. The doctor has suggested I give it
> another six months.
>
> [I have a *certificate* to prove I'm sane. How many people can say
> that?]
Nearly anyone without a speech defect can *say* it.
Marty
Martin Crisp
wrote (in message
<1f5ubh7.1rv73f910oi3i8N%john.w...@bigpond.com>):
>> Well, He could, nominally, have written 'Searl-ey' you must be joking.
>>
>> Marty
>
> Such puns are universally atrocious. And don't call me Surly...
<flashback from _The Naked Vicar Show_>
Sessile?
Have Fun
Martin
--
Kinky:
What I do that you wouldn't
Perverted:
What you do that I wouldn't
Richard Harter
wrote:
Hmmm, let's see. "atrocious" would come from "trochee" (having two
feet) and therefore must signify the property of not having two feet.
"Universe", of course, refers to a poetic form with only one verse.
The puns in question, the "such puns", evidently are recited whilst
standing on one foot and are delivered with but a single breath.
Dunk
wrote:
>Marty Fouts <usene...@usa.net> wrote:
>
>> john.w...@bigpond.com (John Wilkins) writes:
>>
>> > I do not have a place for abstract entities in my poverty-stricken
>> > ontology. They do not "exist". My argument is that the properties
>> > ascribed to abstract symbolisms and procedural languages are in fact
>> > not causal. They do nothing.
>> > > Praps you should go back to bed until you are feeling less Searle-ey.
Searle stuns markets
7510d319.0201...@posting.google.com
Dunk
John Wilkins
But do they come when you do call them? Oh, sorry, wrong context.
Matt Silberstein
> In article <u66694...@usa.net>, Marty Fouts <usene...@usa.net> wrote:
> >
> >mat...@ix.netcom.com (Matt Silberstein) writes:
[snip]
> >
> >> I could imagine, at the dangling end of my imagination, some ET who
> >> could (successfully?) do this. I agree that it does not make much
> >> sense. But more to the point I can imagine some ET who does not make
> >> the life/not life distinction we make. Yet life seems to be a
> >> "property" of some stuff. That the Universe seems to have some easy
> >> to find groupings does not mean that the stuff has natural
> >> categories.
>
> What's this "we"? I dont think plasmas are a little bit alive. :)
Nor do I. You really need to distinguish between your carachature of
my ideas and my ideas.
[snip]
> (Have you ever read the _Republic_? It's a farce, a laugh a minute.)
OTOH read Aristotle's Ethics and Politics. Brilliant works. Not
relevant to this conversation, but brilliant.
Matt Silberstein
> mat...@ix.netcom.com (Matt Silberstein) writes:
>
> > Marty Fouts <usene...@usa.net> wrote in message
> > news:<usn9ey...@usa.net>...
> > > john.w...@bigpond.com (John Wilkins) writes:
>
> > > > I do not have a place for abstract entities in my
> > > > poverty-stricken ontology. They do not "exist". My argument is
> > > > that the properties ascribed to abstract symbolisms and
> > > > procedural languages are in fact not causal. They do nothing.
> > >
> > > Could you slow this one down and run it by me again? Why, in your
> > > ontology, do properties have to do something rather than simply be
> > > something? To a physicist it would appear that you just said that
> > > mass does not exist because it does not do something...
> >
> > I suspect that John will object to the idea that things have
> > properties. (And if he does not, I will.) Properties are
> > abstractions, things just are. Your sweater does not have the
> > property of being red and the property of being warm. We humans
> > abstract aspects of the stuff and give those names.
>
> So sheep don't have feet, then, because 'foot' is an abstraction? And
> water doesn't contain oxygen because 'oxygen' is an abstraction?
Not what I said. I said they don't have the property of feet. A
different statement. The thing we call a sheep has stuff we call feet.
But feet is not a property of sheep. Nor is foot a property of a
sheep. (Nor is hoof for those who follow halacha.)
> [snip]
>
> > I could imagine, at the dangling end of my imagination, some ET who
> > could (successfully?) do this. I agree that it does not make much
> > sense. But more to the point I can imagine some ET who does not make
> > the life/not life distinction we make. Yet life seems to be a
> > "property" of some stuff. That the Universe seems to have some easy
> > to find groupings does not mean that the stuff has natural
> > categories.
>
> I don't think that showing an example that doesn't really qualify as a
> property (the life/non-life) distinction makes the idea of property go
> away.
Ok, so life is not a property. I wonder if "pretty" is a property? It
is more obviously culture and situation dependent than red. I will try
to think of a better example, but my point is that "stuff" does not
have "properties",
"stuff", for lack of a better term, is. We group aspects of that stuff
and give it names. But the categories named are our categories.
> > > This is approaching the edge of the problem that interests me in this
> > > area. English is an historical accident. It need not have existed. But
> > > is the same thing true about pi? Is it really possible for the ETs in
> > > my example to have generated a physics without 'mass'?
> >
> > There are all kinds of ratios between all kinds of things. (See
> > http://www.satirewire.com/charts/aolpie.shtml and
> > http://www.satirewire.com/charts/madrigals.shtml. Ok, so those are bad
> > examples.) Pi and geometric circles are properties, so to speak, of
> > human minds. Absent the existence of humans (and other equivilant
> > thinkers) pi does not exist. Absent the existence of humans etc. stuff
> > will, we all suspect, act pretty much the same, but the particular
> > aspect we call mass will not be abstracted and name. The stuff is the
> > same whether or not we talk about.
>
> Your next to last sentence seems to contradict your last sentence. The
> property that we care to label 'mass' will not go away because there
> is no intelligence around to invent a label for it.
I don't see a contradiction because I seem to see things differently.
We agree that the stuff does not (AFAWCT) change whether or not
we/anyone observes it. (If someone wants to take up that line of
discussion they are welcome to it.) And I will gladly assert that in
normal conversation, when trying to simply communicate, I have no
problem with taking about properties of things. (As John says, these
are convenient fictions.) What I am trying to say is that the names
we use are names for aspects we find important. Red/Orange
distinctions are culture and/or physiology dependent. The light
reflected is situationally dependent. These do not belong to the
sweater nor are they indpendent things. (I assume these are all things
we agree about. The question is whether they have sufficient imporant
for incorporation in an ontology.)
> [snip]
>
> > > It is not clear to me that some properties that we think of as
> > > abstractions and non causal are not in that sense 'real'. The sense,
> > > to be redundent being that any observer that came along with the right
> > > sensory equipment could discover them. ('color', for example is
> > > subjective, but 'reflecting photons of a certain frequency' is not)
> >
> > That does not make it "real". Likely, important, yes, real no.
>
> What is 'unreal' about the demonstrable observation that light
> reflects?
What is unreal is the thought that redness is a property of the
sweater. It is not. If you say that the sweater has reflected light of
a particular frequency then you will get no argument from me. But that
is, ontologically speaking, a different claim.
Matt Silberstein
> On 10 Jan 2002 10:05:15 -0500, in talk.origins
> mat...@ix.netcom.com (Matt Silberstein) wrote in
> <76998029.02011...@posting.google.com>:
>
>
> >Marty Fouts <usene...@usa.net> wrote in message news:<uu1tv1...@usa.net>...
> >
> >[snip]
> >
> >>If 'red' is not a property of my red sweather, what is it?
> >
> >It is the context, culture, and psychological state dependent name
> >given to an particular subset of aspects of the sweater. It is "my"
> >name for "my" perception, not a property of the sweater itself.
>
> Well, there is a property of the sweater, its ability to reflect light
> of <knowable range of> angstroms, that we call red.
I would rather say, in this context, that the sweater, under some set
of reasonably knowable conditions, reflects light in a particular
range and we, under some set of conditions, call that read. But red is
not a property of the sweater. Red is a condtion involving the
sweater, the viewer, the light, the surroundings, etc.
> >Or so it seems to me. I am, AFAICT, a rather severe nominalist myself.
> >However, unlike some/many/most/all of the people posting to this
> >thread I do not expect strict consistency of position. Our
> >thoughts/ontologies/science/epistemologies will not ever map perfectly
> >to "reality". Neither the ones in your brain nor the ones you utter.
> >OTOH that may well be obvious and/or irrelevant.
>
> Certainly the map is not the place and the words are not the things but
> there are places and things that are real.
No one is denying that, at least not in this thread.
David Jensen
>David Jensen <da...@dajensen-family.com> wrote in message news:<64rs3u0gdfg7f623b...@4ax.com>...
>> On 10 Jan 2002 10:05:15 -0500, in talk.origins
>> mat...@ix.netcom.com (Matt Silberstein) wrote in
>> <76998029.02011...@posting.google.com>:
>>
>>
>> >Marty Fouts <usene...@usa.net> wrote in message news:<uu1tv1...@usa.net>...
>> >
>> >[snip]
>> >
>> >>If 'red' is not a property of my red sweather, what is it?
>> >
>> >It is the context, culture, and psychological state dependent name
>> >given to an particular subset of aspects of the sweater. It is "my"
>> >name for "my" perception, not a property of the sweater itself.
>>
>> Well, there is a property of the sweater, its ability to reflect light
>> of <knowable range of> angstroms, that we call red.
>
>I would rather say, in this context, that the sweater, under some set
>of reasonably knowable conditions, reflects light in a particular
>range and we, under some set of conditions, call that red.
>But red is not a property of the sweater.
>Red is a condition involving the sweater, the viewer, the light,
>the surroundings, etc.
I think that red is both. While red is not a property that can be
inherently discovered in the sweater without an external context (can we
discover any properties of anything without an external context?), it is
a property that we recognize as being of the sweater, not of the light
or the air or our eyes. The reflectivity curve of the sweater does not
depend on any of the outside stimuli. It is inherent in the sweater.
>> >Or so it seems to me. I am, AFAICT, a rather severe nominalist myself.
>> >However, unlike some/many/most/all of the people posting to this
>> >thread I do not expect strict consistency of position. Our
>> >thoughts/ontologies/science/epistemologies will not ever map perfectly
>> >to "reality". Neither the ones in your brain nor the ones you utter.
>> >OTOH that may well be obvious and/or irrelevant.
>>
>> Certainly the map is not the place and the words are not the things but
>> there are places and things that are real.
>
>No one is denying that, at least not in this thread.
But we occasionally have a problem with the first half of my statement
in other areas of t.o. There seem to be many, creationists
especially[1], who conflate the description and the reality, while
almost all of us will treat the two as the same in casual conversation.
[1] Second law of Thermodynamics, for example.
Mitchell Coffey
> Matt Silberstein <mat...@ix.netcom.com> wrote:
>
> ....
> > I've never seen anyone in this group suggest that *John* needs to
> > check his medication levels.
>
> But I do. Every six months or so. The doctor has suggested I give it
> another six months.
>
> [I have a *certificate* to prove I'm sane. How many people can say
> that?]
Non sequitur. Matt's point wasn't whether you checked your meds, but
whether anyone in this group ever suggested you needed to check them
(or, rather, whether Matt had ever SEEN anyone in this group suggest
you needed to check them; by which I suspect he really means he's
never READ anyone in this group suggest you needed to check them)! By
all evidence you titrate your meds effectively, thank you, without any
need for your fellow posters to tip you off on when a med modulation
might be in order.
This was a stupid mistake on your part, you moron. I hate you I hate
you I hate you
MRC
Marty Fouts
mat...@ix.netcom.com (Matt Silberstein) writes:
> Marty Fouts <usene...@usa.net> wrote in message news:<u66694...@usa.net>...
> Ok, so life is not a property. I wonder if "pretty" is a property?
> It is more obviously culture and situation dependent than red. I
> will try to think of a better example, but my point is that "stuff"
> does not have "properties", "stuff", for lack of a better term,
> is. We group aspects of that stuff and give it names. But the
> categories named are our categories.
The topic, I think, becomes interesting when you take it one step
further. Suppose that 'we' never came along, but that sheep still were
what they are. Suppose that instead some other 'intelligent' beings
came along and observed the sheep. Is it at all possible that such
creatures would not notice that sheep have feet?
This is the crux of the matter. If any possible intelligence that
might arise in the universe would always notice relationships like
'sheep have feet', then the -property- of having feet exists
independent of the observer and your argument isn't about whether the
property of footedness exists, but rather about whether there is an
observer around to notice and describe the property.
[snip]
> > What is 'unreal' about the demonstrable observation that light
> > reflects?
>
> What is unreal is the thought that redness is a property of the
> sweater. It is not. If you say that the sweater has reflected light
> of a particular frequency then you will get no argument from me. But
> that is, ontologically speaking, a different claim.
You've not answered the question. What is 'unreal' about the
demonstrable observation that light reflects? If the ability of a
substance to always reflect light exists, is not that ability a
property of the substance?
The hair you seem to be splitting is over whether specific examples
are properties or not, not over whether properties can exist without
observers present to describe them. I will freely grant that
properties fall into (at least) two classes, and that the subjective
ones are observer dependent. But what of the objective ones?
Steve Schaffner
> Matt Silberstein <mat...@ix.netcom.com> wrote:
>
> ....
> > I've never seen anyone in this group suggest that *John* needs to
> > check his medication levels.
>
> But I do. Every six months or so. The doctor has suggested I give it
> another six months.
>
> [I have a *certificate* to prove I'm sane. How many people can say
> that?]
My wife, who's an economist, once received a mailing from some guy who
went on at great length about how his ideas had been stolen by the
economics establishment, and how he'd been cheated out of a Nobel
prize and billions of dollars of income in the process. He actually
started out sounding fairly rational; the paranoia only emerged when
you got onto the second or third page. Point to note: his packet
included _two_ certificates(*), both from doctors, no less. So don't
go waving your certificate at me.
(*) To be sure, his certificates attested to the healthy state of his
penis rather than to his sanity, but I'm sure it all fits together
somehow.
--
Steve Schaffner s...@genome.wi.mit.edu
Immediate assurance is an excellent sign of probable lack of
insight into the topic. Josiah Royce