Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Why does Olcott continue to ignore the finite/infinite sequence formulation of the halting problem?

99 views
Skip to first unread message

immibis

unread,
Jan 30, 2024, 3:55:31 PMJan 30
to
Is it because he cannot find an excuse to argue that a sequence can be
both finite and infinite?

wij

unread,
Jan 30, 2024, 6:16:54 PMJan 30
to
He ignores a lots, not just now. You are arguing with a liar. To be
a true liar, you have to lie to yourself first. olcott did it.
How liars create the ‘illusion of truth’
https://www.bbc.com/future/article/20161026-how-liars-create-the-illusion-of-truth

For him, truth::= a lie telling one thousand times.
(Richard should know the details)

But, on the other hand, lots had been tried, something is true there.

Mikko

unread,
Jan 31, 2024, 5:39:23 AMJan 31
to
According to that article, people tend to believe what they hear often.
But not what contradicts what they have heard even more often.
Olcott fails because he cannot avoid contradicting what is heard
more often, apparently because he doesn't know enough about what is
often heard.

--
Mikko

olcott

unread,
Jan 31, 2024, 10:35:04 AMJan 31
to
On 1/30/2024 5:16 PM, wij wrote:
> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>> Is it because he cannot find an excuse to argue that a sequence can
>> be
>> both finite and infinite?
>
> He ignores a lots, not just now. You are arguing with a liar. To be
> a true liar, you have to lie to yourself first. olcott did it.

Nothing that I said is an intentional falsehood thus you commit libel
against me. Much of what a say is self-evident truth thus impossibly
false.

--
Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
hits a target no one else can see." Arthur Schopenhauer

wij

unread,
Jan 31, 2024, 11:09:38 AMJan 31
to
On Wed, 2024-01-31 at 09:34 -0600, olcott wrote:
> On 1/30/2024 5:16 PM, wij wrote:
> > On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
> > > Is it because he cannot find an excuse to argue that a sequence
> > > can
> > > be
> > > both finite and infinite?
> >
> > He ignores a lots, not just now. You are arguing with a liar. To be
> > a true liar, you have to lie to yourself first. olcott did it.
>
> Nothing that I said is an intentional falsehood thus you commit libel
> against me.

I know what kind of risk I am involved. But yes, many things you said
are intentionally false, you know POOH has many flaws that do not fit
the reality.

> Much of what a say is self-evident truth thus impossibly
> false.
>

Just on the contrary, much of what you said (about POOH) are false.

olcott

unread,
Jan 31, 2024, 11:38:04 AMJan 31
to
On 1/31/2024 10:09 AM, wij wrote:
> On Wed, 2024-01-31 at 09:34 -0600, olcott wrote:
>> On 1/30/2024 5:16 PM, wij wrote:
>>> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>>>> Is it because he cannot find an excuse to argue that a sequence
>>>> can
>>>> be
>>>> both finite and infinite?
>>>
>>> He ignores a lots, not just now. You are arguing with a liar. To be
>>> a true liar, you have to lie to yourself first. olcott did it.
>>
>> Nothing that I said is an intentional falsehood thus you commit libel
>> against me.
>
> I know what kind of risk I am involved. But yes, many things you said
> are intentionally false,

*You can't provide any examples of such*
All of the examples that were claimed were simply anchored in
the insufficient understanding of the notion of self-evident truth.

Also Richard has the dippy idea that unintentionally false
statements count as lying. He seems to think that a difference
of opinion with his own misconceptions counts as me lying.

Below I reference an infinite set of simulating termination
analyzers that each correctly aborts its simulation of D
and correctly rejects D as non-halting.

*PREMISE*
*When one understands that simulating termination analyzer H*
*is always correct to abort any simulation that cannot possibly*
*stop running unless aborted*

01 int D(ptr x) // ptr is pointer to int function
02 {
03 int Halt_Status = H(x, x);
04 if (Halt_Status)
05 HERE: goto HERE;
06 return Halt_Status;
07 }
08
09 void main()
10 {
11 H(D,D);
12 }

*IS LOGICALLY ENTAILED FROM PREMISE*
Then every simulating termination analyzer H specified by
the above template correctly aborts its simulation of D
and correctly rejects D as non-halting.

Pages 661 to 696 of Halt7.c specify the H that does this
https://github.com/plolcott/x86utm/blob/master/Halt7.c

immibis

unread,
Jan 31, 2024, 12:09:58 PMJan 31
to
On 1/31/24 16:34, olcott wrote:
> On 1/30/2024 5:16 PM, wij wrote:
>> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>>> Is it because he cannot find an excuse to argue that a sequence can
>>> be
>>> both finite and infinite?
>>
>> He ignores a lots, not just now. You are arguing with a liar. To be
>> a true liar, you have to lie to yourself first. olcott did it.
>
> Nothing that I said is an intentional falsehood thus you commit libel
> against me. Much of what a say is self-evident truth thus impossibly
> false.
>
You intentionally ignore the infinite sequence formulation of the
halting problem.

a...@muc.de

unread,
Jan 31, 2024, 12:56:56 PMJan 31
to
In comp.theory olcott <polc...@gmail.com> wrote:
> On 1/31/2024 10:09 AM, wij wrote:
>> On Wed, 2024-01-31 at 09:34 -0600, olcott wrote:
>>> On 1/30/2024 5:16 PM, wij wrote:
>>>> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>>>>> Is it because he cannot find an excuse to argue that a sequence can
>>>>> be both finite and infinite?

>>>> He ignores a lots, not just now. You are arguing with a liar. To be
>>>> a true liar, you have to lie to yourself first. olcott did it.

>>> Nothing that I said is an intentional falsehood thus you commit libel
>>> against me.

>> I know what kind of risk I am involved. But yes, many things you said
>> are intentionally false,

> *You can't provide any examples of such*

How about you posting on this group that you had coded up turing machines
which demonstrated the falsity of the halting theorem? Did you make such
a post or didn't you? If you did, it was a lie.

> All of the examples that were claimed were simply anchored in
> the insufficient understanding of the notion of self-evident truth.

I know about self evident truth, having a degree in mathematics. In your
posts over the years, you have ignored self evident truths (i.e.
mathematically proven results) and lied about them being falsehoods.

> Also Richard has the dippy idea that unintentionally false
> statements count as lying. He seems to think that a difference
> of opinion with his own misconceptions counts as me lying.

No. Unintentionally false statements are not lying. But deliberately
remaining ignorant of the truth does indeed point to lying. With
mathematically proven results, there's no such thing as "a difference of
opinion". Proven is proven and wrong is wrong.

> Below I reference an infinite set of simulating termination
> analyzers that each correctly aborts its simulation of D
> and correctly rejects D as non-halting.

I suspect very much this is a lie, too. There's no sign of an infinite
set. There's no such thing as a "termination analyser", simulating or
otherwise.

You know full well that it's not truthful.

[ .... ]

> --
> Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
> hits a target no one else can see." Arthur Schopenhauer

--
Alan Mackenzie (Nuremberg, Germany).

olcott

unread,
Jan 31, 2024, 4:58:36 PMJan 31
to
I don't ACM much and definitely not in de.
I do ACM a little.

*Three PhD computer science professors agree*
Does the halting problem place an actual limit on computation?
https://www.researchgate.net/publication/374806722_Does_the_halting_problem_place_an_actual_limit_on_computation


*Maybe you can try your libel on them*

olcott

unread,
Jan 31, 2024, 6:25:16 PMJan 31
to
On 1/31/2024 11:56 AM, a...@muc.de wrote:
It is an objective fact that is a ridiculously stupid thing to say
There are many to be found on Google Scholar [termination analyzer]

*Here is the best one that I found*
https://aprove.informatik.rwth-aachen.de/interface/v-AProVE2023/c

Richard Damon

unread,
Jan 31, 2024, 8:34:15 PMJan 31
to
On 1/31/24 10:34 AM, olcott wrote:
> On 1/30/2024 5:16 PM, wij wrote:
>> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>>> Is it because he cannot find an excuse to argue that a sequence can
>>> be
>>> both finite and infinite?
>>
>> He ignores a lots, not just now. You are arguing with a liar. To be
>> a true liar, you have to lie to yourself first. olcott did it.
>
> Nothing that I said is an intentional falsehood thus you commit libel
> against me. Much of what a say is self-evident truth thus impossibly
> false.
>

It is not label to state a truth.

The fact that you claim false statements to be self-evident truths shows
that you are a pathological liar, whihc is a form of liar, even if you
don't think you are.

wij

unread,
Jan 31, 2024, 9:16:45 PMJan 31
to
On Wed, 2024-01-31 at 09:34 -0600, olcott wrote:
> >

> *You can't provide any examples of such*
> All of the examples that were claimed were simply anchored in
> the insufficient understanding of the notion of self-evident truth.
>
> Also Richard has the dippy idea that unintentionally false
> statements count as lying. He seems to think that a difference
> of opinion with his own misconceptions counts as me lying.
>
> Below I reference an infinite set of simulating termination
> analyzers that each correctly aborts its simulation of D
> and correctly rejects D as non-halting.
>
> *PREMISE*
> *When one understands that simulating termination analyzer H*
> *is always correct to abort any simulation that cannot possibly*
> *stop running unless aborted*
>
> 01 int D(ptr x) // ptr is pointer to int function
> 02 {
> 03 int Halt_Status = H(x, x);
> 04 if (Halt_Status)
> 05 HERE: goto HERE;
> 06 return Halt_Status;
> 07 }
> 08
> 09 void main()
> 10 {
> 11 H(D,D);
> 12 }

Q1. Does the H at line 11 return? What value?
Q2. Does the H at line 03 return? What value?

This basic question has been asked many times before.
Everybody understands C knows exactly what such short piece of codes
will behave. But you always present MADE-UP report, saying things not
what the program actually does.

Most importantly, you know you are not telling the truth because you
deliberately+very carefully EDIT the false reports (many, and
repeatly).


Richard Damon

unread,
Jan 31, 2024, 9:27:07 PMJan 31
to
Bigger quesition, why don't you make main be:

int main()
{
H(D,D);
D(D);
return 0;
}


So we can see the actual behavior of D(D) like Halting is asking about.
If a D(D) doesn't actually "halt" if a simulation of it is aborted, that
will show up in the trace generated by x86UTM of this input.


You also need int main(), as void main() is undefined behavior unless
the implementation is SPECIFICALLY "Free Standing" and defines that void
main is ok.

olcott

unread,
Jan 31, 2024, 9:44:25 PMJan 31
to
Q2 proves otherwise.
It is like asking: Does this code print "Equals":
if (3 == 5)
printf("Equals\n");

> But you always present MADE-UP report, saying things not
> what the program actually does.
>

*H examines the execution trace of the x86 code of D*
H simulates D in debug-step mode and as soon as H sees D call H with
its same parameters and there are no conditional branch instructions
between the beginning of D and its call to H(D,D) then H has its proof
that D never halts.

Pages 661 to 696 of Halt7.c specify the H that does this
https://github.com/plolcott/x86utm/blob/master/Halt7.c

If you can't infer the value that H returns on the basis of the
above explanation (and code) then you lack sufficient technical
skill to correctly review my work.

> Most importantly, you know you are not telling the truth because you
> deliberately+very carefully EDIT the false reports (many, and
> repeatly).
>
>

wij

unread,
Jan 31, 2024, 9:47:49 PMJan 31
to
I tried this before, and IIRC you responded approximately the same as
you say now.

>
> You also need int main(), as void main() is undefined behavior unless
> the implementation is SPECIFICALLY "Free Standing" and defines that
> void
> main is ok.

Technically, the HP has been silently modified to addressing
'function'.

If you are talking about "C". I think "void main()" is valid,
but not sure.

olcott

unread,
Jan 31, 2024, 10:16:05 PMJan 31
to
On 1/31/2024 8:53 PM, wij wrote:
> Your strategy won't work for me. Good luck, liar.
>

In other words you admit that you don't know these things well enough.

Richard Damon

unread,
Jan 31, 2024, 10:26:57 PMJan 31
to
He did it once, and when pointed out that HE had shown that the correct
simulation of D (by x86UTM) came to and end, he seems to have realized
he can't let himself do this.

>>
>> You also need int main(), as void main() is undefined behavior unless
>> the implementation is SPECIFICALLY "Free Standing" and defines that
>> void
>> main is ok.
>
> Technically, the HP has been silently modified to addressing
> 'function'.
>
> If you are talking about "C". I think "void main()" is valid,
> but not sure.
>

5.1.2.2.1p1

The function called at program startup is named main. The implementation
declares no prototype for this function. It shall be defined with a
return type of int and with no parameters:

int main(void) { /* ... */ }

or with two parameters (referred to here as argc and argv, though any
names may be used, as they
are local to the function in which they are declared):

int main(int argc, char *argv[]) { /* ... */ }

or equivalent;10)


Richard Damon

unread,
Jan 31, 2024, 10:32:08 PMJan 31
to
WHy do you say that?

Are you LYIONG again that H(D,D) can do two different things?

YOu have admitted otherwise by refusing to show the trace that proves it
can.

>
>> But you always present MADE-UP report, saying things not
>> what the program actually does.
>>
>
> *H examines the execution trace of the x86 code of D*
> H simulates D in debug-step mode and as soon as H sees D call H with
> its same parameters and there are no conditional branch instructions
> between the beginning of D and its call to H(D,D) then H has its proof
> that D never halts.

Nope. UNSOUND LOGIC.

Proven so, and thus you show yourself AGAIN to just be a pathological liar.

>
> Pages 661 to 696 of Halt7.c specify the H that does this
> https://github.com/plolcott/x86utm/blob/master/Halt7.c

It can do the actions, doesn't make the logic correct.

Since the logic is PROVEN incorrect, it just shows that you believe in
unsound logic, just as your seem to be unsound in mind.

>
> If you can't infer the value that H returns on the basis of the
> above explanation (and code) then you lack sufficient technical
> skill to correctly review my work.


Oh, we can infer that H uses that "logic", and tha tis why it gets the
wrong answer, as you have admitted.

You are just proving that you are just an ignorant hypocritical
pathological lying idiot.

Jeff Barnett

unread,
Feb 1, 2024, 12:17:14 AMFeb 1
to
You might have a degree in Mathematics but it's clear that you (and most
of the people contributing to these inane threads) have never read
"Proofs and Refutations" by Imre Lakatos. The book is immensely
enjoyable and is highly recommended.

>> Below I reference an infinite set of simulating termination
>> analyzers that each correctly aborts its simulation of D
>> and correctly rejects D as non-halting.
>
> I suspect very much this is a lie, too. There's no sign of an infinite
> set. There's no such thing as a "termination analyser", simulating or
> otherwise.
>
> You know full well that it's not truthful.--
Jeff Barnett

Alan Mackenzie

unread,
Feb 1, 2024, 1:57:06 PMFeb 1
to
Jeff Barnett <j...@notatt.com> wrote:
> On 1/31/2024 10:56 AM, a...@muc.de wrote:
>> In comp.theory olcott <polc...@gmail.com> wrote:
>>> On 1/31/2024 10:09 AM, wij wrote:

[ .... ]

>>>> I know what kind of risk I am involved. But yes, many things you said
>>>> are intentionally false,

>>> *You can't provide any examples of such*

>> How about you posting on this group that you had coded up turing machines
>> which demonstrated the falsity of the halting theorem? Did you make such
>> a post or didn't you? If you did, it was a lie.

>>> All of the examples that were claimed were simply anchored in
>>> the insufficient understanding of the notion of self-evident truth.

>> I know about self evident truth, having a degree in mathematics. In your
>> posts over the years, you have ignored self evident truths (i.e.
>> mathematically proven results) and lied about them being falsehoods.

>>> Also Richard has the dippy idea that unintentionally false
>>> statements count as lying. He seems to think that a difference
>>> of opinion with his own misconceptions counts as me lying.

>> No. Unintentionally false statements are not lying. But deliberately
>> remaining ignorant of the truth does indeed point to lying. With
>> mathematically proven results, there's no such thing as "a difference of
>> opinion". Proven is proven and wrong is wrong.

> You might have a degree in Mathematics but it's clear that you (and most
> of the people contributing to these inane threads) have never read
> "Proofs and Refutations" by Imre Lakatos. The book is immensely
> enjoyable and is highly recommended.

Maybe. Supposing I were to study this book earnestly, what would I learn
that is relevant to the current inane thread?

[ .... ]

> Jeff Barnett

Alan Mackenzie

unread,
Feb 1, 2024, 2:08:15 PMFeb 1
to
> I don't ACM much and definitely not in de.
> I do ACM a little.

That's incoherent and meaningless.

> *Three PhD computer science professors agree*
> Does the halting problem place an actual limit on computation?
> https://www.researchgate.net/publication/374806722_Does_the_halting_problem_place_an_actual_limit_on_computation

The plain straight answer is yes it does. More nuanced answers would say
that practical computation is far more limited by other things.

On what do these anonymous alleged computer science professors agree?
That the question exists?

> *Maybe you can try your libel on them*

Get this straight: I don't libel and I don't tell lies on Usenet.

Yesterday evening, I challenged you to deny having lied, and you failed
to address the point. You write falsehoods on Usenet, and do so
knowingly. There's a word for that.

> --
> Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
> hits a target no one else can see." Arthur Schopenhauer

olcott

unread,
Feb 1, 2024, 2:55:06 PMFeb 1
to
*Such a stupid things to say*
https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer

>
>>> You know full well that it's not truthful.
>
>>> [ .... ]
>
>> I don't ACM much and definitely not in de.
>> I do ACM a little.
>
> That's incoherent and meaningless.
It a a joke based on your email address.

>
>> *Three PhD computer science professors agree*
>> Does the halting problem place an actual limit on computation?
>> https://www.researchgate.net/publication/374806722_Does_the_halting_problem_place_an_actual_limit_on_computation
>
> The plain straight answer is yes it does. More nuanced answers would say
> that practical computation is far more limited by other things.
>
> On what do these anonymous alleged computer science professors agree?
> That the question exists?
>

*They are not anonymous Dumbo. Read before you critique nitwit*
They are listed in the paper's references.

Jeff Barnett

unread,
Feb 1, 2024, 3:53:06 PMFeb 1
to
On 2/1/2024 11:57 AM, Alan Mackenzie wrote:
> Jeff Barnett <j...@notatt.com> wrote:
>> On 1/31/2024 10:56 AM, a...@muc.de wrote:
>>> In comp.theory olcott <polc...@gmail.com> wrote:
SNIP
>>> No. Unintentionally false statements are not lying. But deliberately
>>> remaining ignorant of the truth does indeed point to lying. With
>>> mathematically proven results, there's no such thing as "a difference of
>>> opinion". Proven is proven and wrong is wrong.
>
>> You might have a degree in Mathematics but it's clear that you (and most
>> of the people contributing to these inane threads) have never read
>> "Proofs and Refutations" by Imre Lakatos. The book is immensely
>> enjoyable and is highly recommended.
>
> Maybe. Supposing I were to study this book earnestly, what would I learn
> that is relevant to the current inane thread?
The book is considered by many as a classic. It is a pretend
conversation between a teacher and a rather bright class of budding
mathematicians. Several classic theorems are covered that were believed
proved and used to do mathematical work proving other theorems in text
books, lectures and journals. The fly in the ointment is that none of
these theorems were true! Yet the whole math community went on with
business as usual. Two examples of such theorems: Euler's theorem
relating the number of faces, edges, and corners of a polygon depends on
how one defines polygon - it can get quite complicated with the whole
class of non-convex examples, e.g., those with tunnels. Another example
is from elementary high school calculus were we learn about series of
continuous functions that converge to a continuous function under some
simple epsilon/ delta conditions. That theorem was not only false but
was used by world class mathematicians when Fourier series were "fresh"
and all the rage! One could converge to discontinuous functions such as
step functions with standard sin/cos examples. Everyone ignored these
anomalies and went on with business as usual. One day the difference
between convergence and uniform convergence was discovered and the world
was right again. But a lot of text books needed to be rewritten.

Like I said: this book is a fun read and tries (successfully) to be
humorous. As for category. place it in philosophy of mathematics.

Enjoy.

PS What you would learn is that proof of real theorems is really really
difficult and knowing when something is really really proved is also
really really difficult.
--
Jeff Barnett

olcott

unread,
Feb 1, 2024, 4:41:13 PMFeb 1
to
On 1/31/2024 11:56 AM, a...@muc.de wrote:

> There's no such thing as a "termination analyser", simulating or
> otherwise.
>
> You know full well that it's not truthful.

*Alan Mackenzie proves a callous disregard for the truth*
https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer

Alan Mackenzie

unread,
Feb 1, 2024, 5:14:46 PMFeb 1
to
In comp.theory olcott <polc...@gmail.com> wrote:
> On 1/31/2024 11:56 AM, a...@muc.de wrote:

>> There's no such thing as a "termination analyser", simulating or
>> otherwise.

>> You know full well that it's not truthful.

> *Alan Mackenzie proves a callous disregard for the truth*
> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer

I've asked you before not to abuse my name in post Subjects. It doesn't
belong there.

And no, I'm not going to look up vague references, particularly on
google.com, to which I've got no access anyway.

I think I have indeed proved a "callous disregard for the truth", namely
yours. In the last few posts, you have declined to deny an earlier
posting asserting you had coded turing machines which refute the halting
theorem. That was a lie when you posted it, and you are not going to
repeat the lie by denying it now.

You have a dishonest disregard for proven truth, such as the halting
theorem, or Gödel's incompleteness theorems. The lack of understanding
you show for them doesn't excuse you, given the number of times people
have attempted to put you right.

> --
> Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
> hits a target no one else can see." Arthur Schopenhauer

olcott

unread,
Feb 1, 2024, 5:40:44 PMFeb 1
to
On 2/1/2024 4:14 PM, Alan Mackenzie wrote:
> In comp.theory olcott <polc...@gmail.com> wrote:
>> On 1/31/2024 11:56 AM, a...@muc.de wrote:
>
>>> There's no such thing as a "termination analyser", simulating or
>>> otherwise.
>
>>> You know full well that it's not truthful.
>
>> *Alan Mackenzie proves a callous disregard for the truth*
>> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer
>
> I've asked you before not to abuse my name in post Subjects. It doesn't
> belong there.
>
> And no, I'm not going to look up vague references, particularly on
> google.com, to which I've got no access anyway.
>
> I think I have indeed proved a "callous disregard for the truth", namely
> yours. In the last few posts, you have declined to deny an earlier
> posting asserting you had coded turing machines which refute the halting
> theorem. That was a lie when you posted it, and you are not going to
> repeat the lie by denying it now.

As long as you continue to libel me with callous disregard for
easily verified facts I will continue to call you out on this.

*WST 2023: 19th International Workshop on Termination*
https://easychair.org/cfp/WST2023

*Termination analysis without the tears*
https://dl.acm.org/doi/10.1145/3453483.3454110

*Termination Analysis of Higher-Order Functional Programs*
https://link.springer.com/chapter/10.1007/11575467_19

*Termination Analysis with Calling Context Graphs*
https://www.khoury.northeastern.edu/home/pete/pub/cav-ccgs.pdf

Mike Terry

unread,
Feb 1, 2024, 7:09:30 PMFeb 1
to
On 01/02/2024 22:14, Alan Mackenzie wrote:
> In comp.theory olcott <polc...@gmail.com> wrote:
>> On 1/31/2024 11:56 AM, a...@muc.de wrote:
>
>>> There's no such thing as a "termination analyser", simulating or
>>> otherwise.
>
>>> You know full well that it's not truthful.
>
>> *Alan Mackenzie proves a callous disregard for the truth*
>> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer
>
> I've asked you before not to abuse my name in post Subjects. It doesn't
> belong there.

Yeah, he does that to everone sooner or later. It's just the way he is. Perhaps he thinks he's
shaming you into behaving better, I don't know, but of course anybody reading the thread just thinks
"that PO - what a jerk...!" (...which doesn't bother PO...)

>
> And no, I'm not going to look up vague references, particularly on
> google.com, to which I've got no access anyway.
>
> I think I have indeed proved a "callous disregard for the truth", namely
> yours. In the last few posts, you have declined to deny an earlier
> posting asserting you had coded turing machines which refute the halting
> theorem. That was a lie when you posted it, and you are not going to
> repeat the lie by denying it now.
>
> You have a dishonest disregard for proven truth, such as the halting
> theorem, or Gödel's incompleteness theorems. The lack of understanding
> you show for them doesn't excuse you, given the number of times people
> have attempted to put you right.

Well, to play devil's advocate, I'd say PO /honestly/ believes he has refuted all those theorems!

Yes, people have explained to him why he's wrong, but he is genuinely intellectually incapable of
understanding those explanations - they just wash over him like a babbling brook, and I doubt he
even gets that the arguments are "logical", or that they differ in character from his own endless
repetitions of his intuitions. To PO both are just people "arguing their case".

[A bit like a blind person who doesn't understand other people can "see" or comprehend what that
involves, so believes he is as good an archer as other seeing people. Worse the person has somehow
convinced himself he's a world-class archer due to his supreme power of concentration, or whatever!!
:) ]

You may say, but if all that were really the case, what would be the point of engaging him in
arguments like this?


Mike.

olcott

unread,
Feb 1, 2024, 7:18:43 PMFeb 1
to
https://www.liarparadox.org/Wittgenstein.pdf
The philosophical underpinnings of analytical truth
prove that mathematical incompleteness is a misconception.

https://www.liarparadox.org/Haskell_Curry_45.pdf
When we understand that Haskell Curry proposes the notion
of True in a formal system means provable from the axioms
of this formal system it doesn't take a genius to see that
unprovable in PA simply means untrue in PA.

olcott

unread,
Feb 1, 2024, 10:06:25 PMFeb 1
to
On 2/1/2024 6:09 PM, Mike Terry wrote:
> On 01/02/2024 22:14, Alan Mackenzie wrote:
>> In comp.theory olcott <polc...@gmail.com> wrote:
>>> On 1/31/2024 11:56 AM, a...@muc.de wrote:
>>
>>>> There's no such thing as a "termination analyser", simulating or
>>>> otherwise.
>>
>>>> You know full well that it's not truthful.
>>
>>> *Alan Mackenzie proves a callous disregard for the truth*
>>> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer
>>
>> I've asked you before not to abuse my name in post Subjects.  It doesn't
>> belong there.
>
> Yeah, he does that to everone sooner or later.

Mike: you didn't bother to pay attention that he
committed libel against me.

*He said that my claim that termination analyzers exist is a lie*

*WST 2023: 19th International Workshop on Termination*
https://easychair.org/cfp/WST2023

*Termination analysis without the tears*
https://dl.acm.org/doi/10.1145/3453483.3454110

*Termination Analysis of Higher-Order Functional Programs*
https://link.springer.com/chapter/10.1007/11575467_19

*Termination Analysis with Calling Context Graphs*
https://www.khoury.northeastern.edu/home/pete/pub/cav-ccgs.pdf


Richard Damon

unread,
Feb 1, 2024, 10:39:35 PMFeb 1
to
Nope.

Proves YOU don't understand what truth is.

>
> https://www.liarparadox.org/Haskell_Curry_45.pdf
> When we understand that Haskell Curry proposes the notion
> of True in a formal system means provable from the axioms
> of this formal system it doesn't take a genius to see that
> unprovable in PA simply means untrue in PA.
>
>

Except that wasn't what Haskell Curry was proposing.

Your Idol just lies, like you. Perhaps because he is just badly misinformed.

Richard Damon

unread,
Feb 1, 2024, 10:39:47 PMFeb 1
to
On 2/1/24 10:06 PM, olcott wrote:
> On 2/1/2024 6:09 PM, Mike Terry wrote:
>> On 01/02/2024 22:14, Alan Mackenzie wrote:
>>> In comp.theory olcott <polc...@gmail.com> wrote:
>>>> On 1/31/2024 11:56 AM, a...@muc.de wrote:
>>>
>>>>> There's no such thing as a "termination analyser", simulating or
>>>>> otherwise.
>>>
>>>>> You know full well that it's not truthful.
>>>
>>>> *Alan Mackenzie proves a callous disregard for the truth*
>>>> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer
>>>
>>> I've asked you before not to abuse my name in post Subjects.  It doesn't
>>> belong there.
>>
>> Yeah, he does that to everone sooner or later.
>
> Mike: you didn't bother to pay attention that he
> committed libel against me.
>
> *He said that my claim that termination analyzers exist is a lie*

But it IS a lie, since you claim it to be in relation to Halt Deciders.

So, you are just lying by misuse of context.

>
> *WST 2023: 19th International Workshop on Termination*
> https://easychair.org/cfp/WST2023
>
> *Termination analysis without the tears*
> https://dl.acm.org/doi/10.1145/3453483.3454110
>
> *Termination Analysis of Higher-Order Functional Programs*
> https://link.springer.com/chapter/10.1007/11575467_19
>
> *Termination Analysis with Calling Context Graphs*
> https://www.khoury.northeastern.edu/home/pete/pub/cav-ccgs.pdf
>
>

And all of these admit to the limitations of Termination analysis, you
don't.

So, you DO lie.

It has been proven.

Jeff Barnett

unread,
Feb 2, 2024, 1:07:04 AMFeb 2
to
On 2/1/2024 11:57 AM, Alan Mackenzie wrote:
> Jeff Barnett <j...@notatt.com> wrote:
>> On 1/31/2024 10:56 AM, a...@muc.de wrote:
SNIP
>> You might have a degree in Mathematics but it's clear that you (and most
>> of the people contributing to these inane threads) have never read
>> "Proofs and Refutations" by Imre Lakatos. The book is immensely
>> enjoyable and is highly recommended.
>
> Maybe. Supposing I were to study this book earnestly, what would I learn
> that is relevant to the current inane thread?
Alan, now that I answered the above question in a sister post, are you
going to find out more about the book on your own?
--
Jeff Barnett

Alan Mackenzie

unread,
Feb 2, 2024, 7:07:34 AMFeb 2
to
In comp.theory olcott <polc...@gmail.com> wrote:
> On 2/1/2024 4:14 PM, Alan Mackenzie wrote:
>> In comp.theory olcott <polc...@gmail.com> wrote:
>>> On 1/31/2024 11:56 AM, a...@muc.de wrote:

>>>> There's no such thing as a "termination analyser", simulating or
>>>> otherwise.

>>>> You know full well that it's not truthful.

>>> *Alan Mackenzie proves a callous disregard for the truth*
>>> https://scholar.google.com/scholar?hl=en&as_sdt=0,28&q=termination+analyzer

>> I've asked you before not to abuse my name in post Subjects. It doesn't
>> belong there.

>> And no, I'm not going to look up vague references, particularly on
>> google.com, to which I've got no access anyway.

>> I think I have indeed proved a "callous disregard for the truth", namely
>> yours. In the last few posts, you have declined to deny an earlier
>> posting asserting you had coded turing machines which refute the halting
>> theorem. That was a lie when you posted it, and you are not going to
>> repeat the lie by denying it now.

> As long as you continue to libel me with callous disregard for
> easily verified facts I will continue to call you out on this.

You can call me out if you like, but harrassing people by putting their
names in the Subject: line is not a nice thing to do. I wonder what
Eternal September would think about that.

I'm also not libelling you. You have had the opportunity to set the
record straight over that alleged post from some years ago, but have
failed to do so. I think I'm justified in concluding that you lied in
that post, knowingly and deliberately.

As for termination analysers, you have used the term merely as an
imprecise synonym for halt deciders. These, indeed, do not exist, and
that was what I meant when I said the termination analysers do not exist.
If you really meant something different in your use of that term, then
please accept my apologies, and explain precisely what you did mean.

That you can can enter "termination analyzer" into a search engine and
come up with some matches from other contexts is clear.

[ .... ]

> --
> Copyright 2024 Olcott "Talent hits a target no one else can hit; Genius
> hits a target no one else can see." Arthur Schopenhauer

Alan Mackenzie

unread,
Feb 2, 2024, 7:15:21 AMFeb 2
to
Almost certainly not, I'm afraid. It's a long time since I seriously
read any maths; I'm doing other things, now.

But forgive me when I get annoyed when people disparage learning and
expertise. That has been going on on this newsgroups for a long time,
now.

> --

olcott

unread,
Feb 2, 2024, 10:53:58 AMFeb 2
to
On 1/31/2024 11:56 AM, a...@muc.de wrote:
> In comp.theory olcott <polc...@gmail.com> wrote:
> No. Unintentionally false statements are not lying. But deliberately
> remaining ignorant of the truth does indeed point to lying. With
> mathematically proven results, there's no such thing as "a difference of
> opinion". Proven is proven and wrong is wrong.
>
>> Below I reference an infinite set of simulating termination
>> analyzers that each correctly aborts its simulation of D
>> and correctly rejects D as non-halting.
>
> I suspect very much this is a lie, too. There's no sign of an infinite
> set.

> There's no such thing as a "termination analyser", simulating or
> otherwise.
>

*As long as you continue to libel me with callous disregard for*
*easily verified facts I will continue to call you out on this*

*WST 2023: 19th International Workshop on Termination*
https://easychair.org/cfp/WST2023

*Termination analysis without the tears*
https://dl.acm.org/doi/10.1145/3453483.3454110

*Termination Analysis of Higher-Order Functional Programs*
https://link.springer.com/chapter/10.1007/11575467_19

*Termination Analysis with Calling Context Graphs*
https://www.khoury.northeastern.edu/home/pete/pub/cav-ccgs.pdf

Richard Damon

unread,
Feb 2, 2024, 11:10:23 AMFeb 2
to
It is not LIBEL if the claim is true,

Read the papers you point to.

NONE claim what you claim for your "termination analyser", so are not
applicable to your claim.

YOU FAIL.

olcott

unread,
Feb 2, 2024, 11:18:07 AMFeb 2
to
*Your ADD prevents you from paying attention*
He claimed that there is no such thing as any termination
analyzer at all and I was lying for saying this.

immibis

unread,
Feb 5, 2024, 2:37:02 PMFeb 5
to
On 31/01/24 17:37, olcott wrote:
> On 1/31/2024 10:09 AM, wij wrote:
>> On Wed, 2024-01-31 at 09:34 -0600, olcott wrote:
>>> On 1/30/2024 5:16 PM, wij wrote:
>>>> On Tue, 2024-01-30 at 21:55 +0100, immibis wrote:
>>>>> Is it because he cannot find an excuse to argue that a sequence
>>>>> can
>>>>> be
>>>>> both finite and infinite?
>>>>
>>>> He ignores a lots, not just now. You are arguing with a liar. To be
>>>> a true liar, you have to lie to yourself first. olcott did it.
>>>
>>> Nothing that I said is an intentional falsehood thus you commit libel
>>> against me.
>>
>> I know what kind of risk I am involved. But yes, many things you said
>> are intentionally false,
>
> *You can't provide any examples of such*
> All of the examples that were claimed were simply anchored in
> the insufficient understanding of the notion of self-evident truth.
>
> Also Richard has the dippy idea that unintentionally false
> statements count as lying. He seems to think that a difference
> of opinion with his own misconceptions counts as me lying.
>
> Below I reference an infinite set of simulating termination
> analyzers that each correctly aborts its simulation of D
> and correctly rejects D as non-halting.
>
> *PREMISE*
> *When one understands that simulating termination analyzer H*
> *is always correct to abort any simulation that cannot possibly* > *stop running unless aborted*

When one understands that 1+1=3 it is obvious that 2+2=6.

immibis

unread,
Feb 5, 2024, 2:41:06 PMFeb 5
to
It's needless pedantry in this case. The worst that happens with "void
main()" on any serious compiler is that the return value is unspecified.
And in this case, nobody cares which value it returns.

immibis

unread,
Feb 5, 2024, 2:42:06 PMFeb 5
to
On 1/02/24 03:44, olcott wrote:
>>> 01 int D(ptr x)  // ptr is pointer to int function
>>> 02 {
>>> 03   int Halt_Status = H(x, x);
>>> 04   if (Halt_Status)
>>> 05     HERE: goto HERE;
>>> 06   return Halt_Status;
>>> 07 }
>>> 08
>>> 09 void main()
>>> 10 {
>>> 11   H(D,D);
>>> 12 }
>>
>> Q1. Does the H at line 11 return? What value?
>> Q2. Does the H at line 03 return? What value?
>>
>> This basic question has been asked many times before.
>> Everybody understands C knows exactly what such short piece of codes
>> will behave.
>
> Q2 proves otherwise.
> It is like asking: Does this code print "Equals":
>   if (3 == 5)
>     printf("Equals\n");
>
>> But you always present MADE-UP report, saying things not
>> what the program actually does.
>>
>
> *H examines the execution trace of the x86 code of D*
> H simulates D in debug-step mode

Wrong, actually. When D gets to DebugStep, H fails to look at the code
inside DebugStep and simulate that code. Instead, it simulates some
other code which is not the actual code.

Richard Damon

unread,
Feb 5, 2024, 10:58:20 PMFeb 5
to
Actually, some compilers, when invoke in certain ways, will reject the
program.

Yes, on most normal systems, if the compiler accepts it, it will work,
but there can be some systems with unusual ABIs that could have bad efects.
0 new messages