When is something believable?
Phideaux
don't want to describe in plain terms because I really don't believe
that some of the apparent parts are truly related, and I don't want to
be lumped with the crackpots.
In general terms:
Event A occurs in a definite cycle. (As reliable as sunrise.)
Event B can only occur when Event A happens, and the chance of it
happening has been worked out to slightly more than 2.08% (this is not
observation, but calculated. My personal observation puts it slightly
less, but that's using less than a four year timeline).
Event C is totally unpredictable and the conditions surrounding it are
beyond the scope of study, so there is no hope of reproducing it
under controlled conditions. It has happened >.214% of the time. There
are no near-misses where C could be said to have happened: it either
did or it didn't, and always between one A and the next.
However, and here comes the sticky bit, every time C has happened, B
has also occurred. (C can't be triggering B unless you believe in
astrology, telekineses, little green men, or other nonsense.)
Now I would like to continue to observe this and make no definite
conclusions until I've got hundreds of examples of concurrance, but
that would take thousands of years.
Now here's my question:
At what point do you start to believe that C is a reliable predictor
of B when there is no known science that could possibly link the two?
Phideaux
____________
e...@ekj.vestdata.no
> Event A occurs in a definite cycle. (As reliable as sunrise.)
>
> Event B can only occur when Event A happens, and the chance of it
> happening has been worked out to slightly more than 2.08% (this is not
> observation, but calculated. My personal observation puts it slightly
> less, but that's using less than a four year timeline).
Timelines in number of days are irrelevant to your question, the
question is number of occurances. Are you saying once a day for a few
years ? Or like 1000 or so occurences ? 2% of 1000 occurances is only
20, not very much.
> Event C is totally unpredictable and the conditions surrounding it are
> beyond the scope of study, so there is no hope of reproducing it
> under controlled conditions. It has happened >.214% of the time.
With 1000 occurences of A, that means C has happened twice ?
> There are no near-misses where C could be said to have happened: it
> either did or it didn't, and always between one A and the next.
Since A is periodic that's like saying everyone dies between one sunrise
and the next sunrise -- a truism.
> However, and here comes the sticky bit, every time C has happened, B
> has also occurred. (C can't be triggering B unless you believe in
> astrology, telekineses, little green men, or other nonsense.)
If B and C are *totally* unrelated, and C are indeed random as you
claim, then you'd expect B to happen 2% of the times regardless of if C
happens or not. Or 1/50 if you like.
Thus the chanse that it'll happen 100% of the time is roughly 1 to 50^n
where n is the number of occurences of C. For 2 that works out to 1:2500
But keep in mind that that's not at all as high odds as you'd think,
because there's probably *many* different "candidate C" happenings. The
human mind is wonderful at noticing the odd occurences that happen, and
forgetting about all the ones that don't. Put differently -- perhaps
I'll notice if all three times I've broken a bone I've been earing a
blue t-shirt (even though only about 1/5th of my t-shirts are blue)
while failing to notice that my pants have been completely different at
the three occasions.
> Now here's my question:
> At what point do you start to believe that C is a reliable predictor
> of B when there is no known science that could possibly link the two?
There's no absolutes in science. You can do *millions* of experiments
"confirming" Newton, and yet it takes only *one* to show that his
equations do not hold, and infact are only a good approximation aslong
as the speeds involved are low compared to c.
Beyond the mere statistics, which are rather simple really, is the much
harder problem of consistent and uninfluenced observations. Without you
saying anything about what this is about it's impossible to comment on
that most important part.
Sincerely,
Eivind Kjørstad
Erik Max Francis
> Now I would like to continue to observe this and make no definite
> conclusions until I've got hundreds of examples of concurrance, but
> that would take thousands of years.
> Now here's my question:
> At what point do you start to believe that C is a reliable predictor
> of B when there is no known science that could possibly link the two?
Without context it is impossible to say; it takes a judgement call. How
many correlated events have you witnessed? You say that to get hundreds
of trials would take thousands of years, which means that the frequency
has to be something on the order of ten years. So unless you're a very
old man, you've only seen a handful of occurrences. Given that, it
would presumptuous to suggest that there is some meaningful correlation
given such a tiny set of data.
Without the context it is hard to say whether or not a correlation is
likely or even possible. You say you don't think there is one, but
maybe there is and you're just not seeing it. Or perhaps there is a
correlation, but it is from an indirected cause and effect that is hard
to see.
Remember, a crank isn't one who comes up with weird ideas, it's one who
comes up with weird ideas, it's one who persists in retaining those
weird ideas even when they've unambiguously been shown to be wrong.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ Honesty has nothing to hide
\__/ Joi
Alcyone Systems' CatCam / http://www.catcam.com/
What do your pets do all day while you're at work? Find out.
Phideaux
>On Tue, 17 Apr 2001, Phideaux wrote:
>
>> Event A occurs in a definite cycle. (As reliable as sunrise.)
>>
>> Event B can only occur when Event A happens, and the chance of it
>> happening has been worked out to slightly more than 2.08% (this is not
>> observation, but calculated. My personal observation puts it slightly
>> less, but that's using less than a four year timeline).
>
>Timelines in number of days are irrelevant to your question, the
>question is number of occurances. Are you saying once a day for a few
>years ? Or like 1000 or so occurences ? 2% of 1000 occurances is only
>20, not very much.
That's the problem with trying to give only part of the info -- since
I know too well the period, I forgot that I wasn't giving it.
Since I've started keeping notes, there have been 436 occurances.
According to pure statistics, this means B should have happened 8
times, and it's happened 7.
>> Event C is totally unpredictable and the conditions surrounding it are
>> beyond the scope of study, so there is no hope of reproducing it
>> under controlled conditions. It has happened >.214% of the time.
>
>With 1000 occurences of A, that means C has happened twice ?
My first 'observation' was over 1400 A's ago, but I attributed it to
pure coincidence. When it happened 436 A's ago, it made me curious.
It's third occurance has started me wondering.
I follow the general rule:
Once is happenstance.
Twice is coincidence.
Three times shows that you're not paying attention.
>> There are no near-misses where C could be said to have happened: it
>> either did or it didn't, and always between one A and the next.
>
>Since A is periodic that's like saying everyone dies between one sunrise
>and the next sunrise -- a truism.
True -- I should have said that when C has happened, at the next A, B
also happened. There was no past-posting or 'within a few occurances
of A'.
>> However, and here comes the sticky bit, every time C has happened, B
>> has also occurred. (C can't be triggering B unless you believe in
>> astrology, telekineses, little green men, or other nonsense.)
>
>If B and C are *totally* unrelated, and C are indeed random as you
>claim, then you'd expect B to happen 2% of the times regardless of if C
>happens or not. Or 1/50 if you like.
Right on all counts. If B and C _are_ related, I'm throwing out my
calculus books and taking up oriental mysticism.
>Thus the chanse that it'll happen 100% of the time is roughly 1 to 50^n
>where n is the number of occurences of C. For 2 that works out to 1:2500
>
>But keep in mind that that's not at all as high odds as you'd think,
Exactly -- it actually calculates to 1:110,592, not nearly as high as
the odds of being struck by lightening, of which dozens of people die
every year.
>because there's probably *many* different "candidate C" happenings. The
>human mind is wonderful at noticing the odd occurences that happen, and
>forgetting about all the ones that don't. Put differently -- perhaps
>I'll notice if all three times I've broken a bone I've been earing a
>blue t-shirt (even though only about 1/5th of my t-shirts are blue)
>while failing to notice that my pants have been completely different at
>the three occasions.
There are so many variables, most of which cannot be identified, let
alone controlled, I'm sure a little looking would yield an entire
field of candidates. However, C is such a notable event (and by
notable I mean I actually wrote it down _before_ A-B happened, and
those were the only occurances so noted), and occurs so seldom, that
it is beginning to push the boundaries of my believability.
>> Now here's my question:
>> At what point do you start to believe that C is a reliable predictor
>> of B when there is no known science that could possibly link the two?
>
>Beyond the mere statistics, which are rather simple really, is the much
>harder problem of consistent and uninfluenced observations. Without you
>saying anything about what this is about it's impossible to comment on
>that most important part.
>
I guess what I'm really asking is at what point does the average
person (not to suggest that anyone here is average) start thinking
that C and B are actually related in ways that cannot be readily
explained? When the odds of C then B in a series are 1:50,000,000,
1:9,000,000,000, 1:9,999,999,999,999,999,999,999,999?
Phideaux
____________
Charles R Martin
When you have a falsifiable theory about the connection. This somes
up fairly often in the epistomology of science sorts of
investigations, starting with Kant and Hegel. Kant (I think) had a
"clock parable" to explain the problem: let's say you have two clocks,
slightly out of sync, so that the one we call (arbitrarily) clock A
ticks first. We _cannot_ infer that clock A "causes" clock B to tick,
even though there's an obvious apparent causal connection. (And in
fact, from the conditions of the thought experiment, we know that A
and B really are independent.)
In statistics, this gets expressed as the truism "correlation doesn't
imply causation."
>
>
> Phideaux ____________
--
No one is patriotic about taxes. -- George Orwell
------------------------------------------------------------------------------
Charlie Martin, Broomfield, CO USA 40 N 105 W
Jesse_Mazer
are? It's possible there could be some hidden cause you just haven't
thought of...it would help to know what these events actually are.
Anyway, now I'm all curious!
Jesse
Brett Evill
> I guess what I'm really asking is at what point does the average
> person (not to suggest that anyone here is average) start thinking
> that C and B are actually related in ways that cannot be readily
> explained? When the odds of C then B in a series are 1:50,000,000,
> 1:9,000,000,000, 1:9,999,999,999,999,999,999,999,999?
Experients with first-year stats classes seem to show that the average
person starts to believe at 1:10 or thereabouts.
You have to get people quite sophisticated before they can understand
the concept of PETWAC, or understand the dangers inherent in looking at
thousands of natural experiments and marvelling at those with levels of
significance of 99.5%
Regards,
Brett Evill
Stan
> Now here's my question:
> At what point do you start to believe that C is a reliable predictor
> of B when there is no known science that could possibly link the two?
As a high school teacher was fond of saying, proximity is not proof.
You have amassed statistics, which should be used for illumination,
not for support. Here's a story which might help:
A man walking down the street notices a peculiar fellow snapping his
fingers rapidly. When asked why, the fellow says "It's to keep the
rampaging elephants away!" When told by the man that there aren't
any elephants for hundreds, maybe thousands of miles the fellow says
"Well, then, it's working, isn't it!"
Stan.
"Doubt is not a pleasant condition, but certainty is absurd." Voltaire
Phideaux
1) It falls too close to an area that the crackpots and charlatans
have been mining for years.
2) I want to see comments based on the rate of coincidence, not on the
validity of a possible connection. In other words -- at what point do
the numbers start meaning more than the appearance of being
ridiculous.
3) If it is reasonable to assume that C predicts B, for whatever
reason, a small investment could yield a huge return in a few years.
Although having the possibility widely known could not affect the
outcome, it could affect the profits.
Phideaux
________
The early bird may get the worm,
but the second mouse gets the cheese.
Erik Max Francis
> On Tue, 17 Apr 2001 16:50:59 -0400, Jesse_Mazer
> <Jesse...@alumni.brown.edu> wrote:
>
> >If you don't mind my asking, why don't you want to say what A, B, and
> > C
> >are? It's possible there could be some hidden cause you just haven't
> >thought of...it would help to know what these events actually are.
> >Anyway, now I'm all curious!
>
> 1) It falls too close to an area that the crackpots and charlatans
> have been mining for years.
But that means we're nearing the end of the discussion, because context
is important.
> 2) I want to see comments based on the rate of coincidence, not on the
> validity of a possible connection. In other words -- at what point do
> the numbers start meaning more than the appearance of being
> ridiculous.
It requires a judgement call based on the context and the processes
involved. Without that it is very hard to say, particularly in total
abstract terms.
> 3) If it is reasonable to assume that C predicts B, for whatever
> reason, a small investment could yield a huge return in a few years.
Oh, dear.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ I want to know God's thought; the rest are details.
\__/ Albert Einstein
Physics reference / http://www.alcyone.com/max/reference/physics/
A physics reference.
Jesse_Mazer
(Phideaux) wrote:
> On Tue, 17 Apr 2001 16:50:59 -0400, Jesse_Mazer
> <Jesse...@alumni.brown.edu> wrote:
>
> >If you don't mind my asking, why don't you want to say what A, B, and C
> >are? It's possible there could be some hidden cause you just haven't
> >thought of...it would help to know what these events actually are.
> >Anyway, now I'm all curious!
> >
>
> 1) It falls too close to an area that the crackpots and charlatans
> have been mining for years.
Maybe you could give an analogy, or just the general "area?" Psi,
telekenesis, premonition, what?
Scott Hedrick
"Phideaux" <phid...@mcmsys.com> wrote in message
news:3adcdd89...@news.newsguy.com...
> On Tue, 17 Apr 2001 16:50:59 -0400, Jesse_Mazer
> <Jesse...@alumni.brown.edu> wrote:
>
> >If you don't mind my asking, why don't you want to say what A, B, and C
charlatans
> have been mining for years.
So does holding back information that may be necessary. Keeping secrets-
refusing to let the mechanism of the secret antigravity drive be analyzed,
for example- is indicative (though not definitive) of the crackpot.
Scott Hedrick
"Charles R Martin" <crma...@indra.com> wrote in message
news:m31yqrj...@localhost.localdomain...
> phid...@mcmsys.com (Phideaux) writes:
> In statistics, this gets expressed as the truism "correlation doesn't
> imply causation."
How's this analogy- sharks and dolphins may have the same general shape, but
that doesn't mean they're both fish.
I seem to recall an experiment that was done to see the effects of changing
work conditions on productivity. I'd have to dig up my college psych book to
name it. Lighting was increased, then decreased. Music was played loud, then
soft. Many different changes were made. Productivity seemed to improve with
every change. No one element could be pinpointed to explain why-there seemed
to be no correlation with the type of change. It turned out to be something
unrelated to any element- the workers were happy (more or less) that
management was paying attention to them, and when some change occured, they
knew they were being watched.
Brian Davis
>> 3) If it is reasonable to assume that C predicts B, for whatever
>> reason, a small investment could yield a huge return in a few years.
>
> Oh, dear.
Since this is rasfs, I'll quote Niven: TANSTAAFL.
--
Brian Davis
Erik Max Francis
> Since this is rasfs, I'll quote Niven: TANSTAAFL.
I think you mean Heinlein.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ What a crime to waste [youth] on children.
\__/ George Bernard Shaw
Alcyone Systems' Daily Planet / http://www.alcyone.com/planet.html
A new, virtual planet, every day.
Phideaux
e...@ekj.vestdata.no
> When you have a falsifiable theory about the connection. This somes
> up fairly often in the epistomology of science sorts of
> investigations, starting with Kant and Hegel. Kant (I think) had a
> "clock parable" to explain the problem: let's say you have two
> clocks, slightly out of sync, so that the one we call (arbitrarily)
> clock A ticks first. We _cannot_ infer that clock A "causes" clock
> B to tick, even though there's an obvious apparent causal
> connection. (And in fact, from the conditions of the thought
> experiment, we know that A and B really are independent.)
> In statistics, this gets expressed as the truism "correlation
> doesn't imply causation."
They are independent. But I'm still not so sure it'd be a bad idea to
bet that B would tick just following the next A. Indeed, barring a
breakdown in either clock, that is the likely outcome.
If I got the question correctly, the poster was wondering aloud at which
point it becomes reasonable to expect that B is going to tick just after
A. If A *causes* B to tick is a quite different question.
When is it reasonable to conclude with correlation is a quite different
question from when is it reasonable to conclude that there's causation.
regards,
Eivind
e...@ekj.vestdata.no
> On Tue, 17 Apr 2001 11:19:39 +0200, <e...@ekj.vestdata.no> wrote:
> >On Tue, 17 Apr 2001, Phideaux wrote:
> >> Event A occurs in a definite cycle. (As reliable as sunrise.)
>>> Event B can only occur when Event A happens, and the chance of it
>>> happening has been worked out to slightly more than 2.08% (this is not
>>> observation, but calculated. My personal observation puts it slightly
>>> less, but that's using less than a four year timeline).
> Since I've started keeping notes, there have been 436 occurances.
> According to pure statistics, this means B should have happened 8
> times, and it's happened 7.
> My first 'observation' was over 1400 A's ago, but I attributed it to
> pure coincidence. When it happened 436 A's ago, it made me curious.
> It's third occurance has started me wondering.
So. Your C has happened three times, in each case followed by B which
"normally" only happens in 2% of the cases or something.
> True -- I should have said that when C has happened, at the next A,
> B also happened. There was no past-posting or 'within a few
> occurances of A'.
Is C and B tied to A's period, or is it conceivable that B migth for
example happen *twice* between two A's ?
> There are so many variables, most of which cannot be identified, let
> alone controlled, I'm sure a little looking would yield an entire
> field of candidates. However, C is such a notable event (and by
> notable I mean I actually wrote it down _before_ A-B happened, and
> those were the only occurances so noted), and occurs so seldom, that
> it is beginning to push the boundaries of my believability.
3 times isn't a lot of statistical material to go by... My example about
the blue t-shirt wasn't random, it refers to my brother and not me but
otherwise it happened. On three occasions has he broken a bone, twice
his left arm, and once his right leg. On all three occasions has he been
wearing a blue t-shirt. This inspite the fact that he wears blue
t-shirts no more than say 1/5th of the time at most. The chanse of that
happening by chanse is less than 1%. But there's a major caveat: the
chanse that *some* of his clothing would have *something* in common on
the three occasions are much higher. And allthough I've not checked
this, migth it be possible that he's got some blue t-shirts which he
likes to use while engaging in sports ?
> I guess what I'm really asking is at what point does the average
> person (not to suggest that anyone here is average) start thinking
> that C and B are actually related in ways that cannot be readily
> explained? When the odds of C then B in a series are 1:50,000,000,
> 1:9,000,000,000, 1:9,999,999,999,999,999,999,999,999?
There's no absolutes in science. But in general the more radical the
claim, the more solid evidence does it need to be believed. If I claimed
I had a plant at home, and showed a photo from my apartment as evidence
most people would be inclined to believe me. If I claimed to have 11
aliens living in my home, and showed a photo as evidence most people
would start asking more questions.
However, you have a hypothesis, roughly something like:
* Whenever C happens, B will always also happen before 2*A has happened.
Since you've said A is periodic it's actually superfluous and can be
replaced with:
* Whenever C happens, B will always also happen within 2*X
Where X is the period of A.
Absent any information about what all these letters of yours are for
there's notthing much to be said about this.
Sincerely,
Eivind
Charles R Martin
Phideaux
IIRC, that was Bell Labs back in the 30s.
At least in that case, there was an identifiable correlation, just not
one they thought of in advance.
Phideaux
My reticence is based on the old adage: it is better to remain silent
and be thought a fool, than to speak and remove all doubts.
Some time ago (in the 60s?) someone with way too much time on their
hands found that of X number of hurricanes every decade, Y% hit North
Carolina. They also found that if the New York Times (or so other big
paper -- it's been a long time since I heard about this) had exactly
three typographical errors on their front page, the _next_ hurricane
hit NC. Their 'research' showed that of something like 5 occurances of
exactly three typos, each was followed by NC getting hit.
To suggest that there is a correlation between the two is obviously
lunatic (according to my definition). But without a lengthy
explanation of exact circumstances, what I'm observing would probably
be classed as very similar -- and it probably is. I don't believe it
yet, but the coincidences are getting to the point that they are
interesting, and I'd appreciate some feedback on just how many such
coincidences have to happen before you start to seriously question
whether there are factors that you can't possibly know about which are
influencing events.
BTW, if you have plans for an anti-graivity device, I can help. For
only a small down payment (cash, in small bills), I can sell you a
bridge, the tolls from which can finance your project for years.
Monte Davis
>I seem to recall an experiment that was done to see the effects of changing
>work conditions on productivity. I'd have to dig up my college psych book to
>name it.
The Hawthorne effect. See, for example,
http://www.burtonreport.com/InfHealthCare/Info&UseHawthorne.html
Allen Thomson
"Phideaux" <phid...@mcmsys.com> wrote
> I'm taking a very informal look at a strange little phenomenon which I
> don't want to describe in plain terms because I really don't believe
> that some of the apparent parts are truly related, and I don't want to
> be lumped with the crackpots.
>
> In general terms:
[deletia]
I recently asked sci.physics about a somewhat similar case (actually a
flipside: respectable arguments indicated that an A should cause a B with a
high degree of probability, but the data contained lots of As and few/none
Bs), and received a very useful response. Do a Usenet search on "causation
without correlation" to find the thread.
Also try s.p. or one of the .stat groups.
Bill Snyder
wrote:
>
>BTW, if you have plans for an anti-graivity device, I can help. For
>only a small down payment (cash, in small bills), I can sell you a
>bridge, the tolls from which can finance your project for years.
>
float a loan.
--
Bill Snyder [This space unintentionally left blank.]
Brian Davis
> I think you mean Heinlein.
<sigh>... nothing like making a point. sorry 'bout that, you're both
right of course.
--
Brian Davis
Scott Hedrick
> My reticence is based on the old adage: it is better to remain silent
> and be thought a fool, than to speak and remove all doubts.
Perhaps, but you've reached the point where what you are hiding is now
necessary to continue with the conversation. Keeping it a secret now does
mark you for a crank.
Keeping quiet makes you a crank. You're afraid that revealing will mark you
as a crank, but may also allow the answer you want to be determined. Since
you seem to be destined to be a crank either way, but only revelation
increases your chance of success, revelation would appear to be the logical
choice.
Jesse_Mazer
<rebi...@email.msn.com> wrote:
> "Phideaux" <phid...@mcmsys.com> wrote in message
> news:3addc053...@news.newsguy.com...
> > My reticence is based on the old adage: it is better to remain silent
> > and be thought a fool, than to speak and remove all doubts.
>
> Perhaps, but you've reached the point where what you are hiding is now
> necessary to continue with the conversation. Keeping it a secret now does
> mark you for a crank.
Uh, didn't he just reveal it in the very post you're quoting? The thing
about typos and hurricanes?
I'd like to know more about this pattern...how many "hits" has it had
*since* the original pattern was found? Can you verify that the
original prediction was made in the 60's rather than more recently?
It's always possible to find crazy patterns in retrospect by sifting
through large amounts of data...just look at the predictions in the
"Bible Code."
Erik Max Francis
> Uh, didn't he just reveal it in the very post you're quoting? The
> thing
> about typos and hurricanes?
No, he was giving that as an example.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ Only the ephemeral is of lasting value.
\__/ Ionesco
Polly Wanna Cracka? / http://www.pollywannacracka.com/
The Internet resource for interracial relationships.
Jesse_Mazer
<m...@alcyone.com> wrote:
> Jesse_Mazer wrote:
>
> > Uh, didn't he just reveal it in the very post you're quoting? The
> > thing
> > about typos and hurricanes?
>
> No, he was giving that as an example.
Whoops, you're right--sorry about that. But I think it goes to show
that there's no reason not to reveal it, since hopefully no one would
think he was an idiot for being fascinated by the typo/hurricane thing,
as long as he didn't take it *too* seriously...and his real example
can't be much more out there than that, can it?
Brett Evill
>
> But I think it goes to show
> that there's no reason not to reveal it, since hopefully no one would
> think he was an idiot for being fascinated by the typo/hurricane thing,
> as long as he didn't take it *too* seriously...and his real example
> can't be much more out there than that, can it?
No. But he *does* believe it.
Regards,
Brett Evill
Phideaux
<Jesse...@alumni.brown.edu> wrote:
>In article <uPo$66IyAHA.273@cpmsnbbsa07>, "Scott Hedrick"
><rebi...@email.msn.com> wrote:
>
>> "Phideaux" <phid...@mcmsys.com> wrote in message
>> news:3addc053...@news.newsguy.com...
>> > My reticence is based on the old adage: it is better to remain silent
>> > and be thought a fool, than to speak and remove all doubts.
>>
>> Perhaps, but you've reached the point where what you are hiding is now
>> necessary to continue with the conversation. Keeping it a secret now does
>> mark you for a crank.
>
>Uh, didn't he just reveal it in the very post you're quoting? The thing
>about typos and hurricanes?
>
>I'd like to know more about this pattern...how many "hits" has it had
>*since* the original pattern was found? Can you verify that the
>original prediction was made in the 60's rather than more recently?
That was just an example of just how far out some coincidences are,
and has nothing to do with my interest, except that to suggest there
is a causal link in what I'm looking at would be just as far-fetched.
>It's always possible to find crazy patterns in retrospect by sifting
>through large amounts of data...just look at the predictions in the
>"Bible Code."
I'm perversely amused when things like the hemline/aspirin/stock
market 'relationship' can be graphed so accurately over the decades.
And someone once made a mint selling a book that purported to list
price shifts in the commodities markets based on dates -- after all,
if soybeans have gone up on every April 17th for the last 9 years, why
wouldn't they this year?
Actually, the context doesn't matter -- it only provided the impetus
for me to ask the original question: At what point do you start to
believe in something even when there is no rational explanation for
it? The history of mankind is filled with things people believe in
without being able to understand -- my primary interest is in knowing
at what depth of observation does that belief come into being.
Obviously it will differ from person to person -- some people will
believe just about everything, others don't believe anything that
isn't provable (well, at least that's their functional illusion). But
there should be a wide middle-ground where belief in that which cannot
be explained does not fall into the categories of delusion or
religion.
Erik Max Francis
> Whoops, you're right--sorry about that. But I think it goes to show
> that there's no reason not to reveal it, since hopefully no one would
> think he was an idiot for being fascinated by the typo/hurricane
> thing,
> as long as he didn't take it *too* seriously...and his real example
> can't be much more out there than that, can it?
Actually, the typos/hurricans example is a good one. Of course there is
no real correlation between the two events, but there's almost certainly
a selection effect going on. This kind of thing happens a lot when
you're looking for a correlation between two unrelated events. If
you're consciously looking for a connection, you'll tend to notice
connections that aren't really there. When the supposedly correlated
event happens, you'll look harder. When it doesn't, you won't look as
hard. It would be easy to unconsciously miss typos when reading the
front page of a newspaper (a process which is in and of itself easy to
do bad proofreading on), and other errors could be used to tweak the
final result: Is using the wrong a typo? bad grammar? a word or idiom
that the proofreader doesn't know? etc.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ We'll have to make our own luck from now on.
\__/ Louis Wu
Interstelen / http://www.interstelen.com/
A multiplayer, strategic, turn-based Web game on an interstellar scale.
Jesse_Mazer
b.e...@tyndale.apana.org.au wrote:
Well, theres' a difference between being very intrigued and
*believing*...for example, I'm very intrigued by all the recent psi
experiments using long runs of random-number generators which seem to
show a statistically significant effect. But I don't "believe" it,
although I don't disbelieve it either. Here's a nice summary of these
results from a sceptic's point of view:
http://www.btinternet.com/~neuronaut/webtwo_features_psi_two.htm
Phideaux
wrote:
>Erik (and others) commented:
I've made so many similar mistakes that it was gleeful perversity that
I could see someone else's.
But I suspect I'm going to have to go back and reread a couple of
books. I know Niven created tanj, but a niggling part of my brain says
he mentioned tanstaafl in _Ringworld_ or _Flight of the Horse_.
A few years ago, in another newsgroup, I mentioned the Yardbird's song
_Stairway to Heaven_. Even with a poor news server, I saw dozens and
dozens of replies to that one.
Phideaux
____________
My insurance agent complains that I drive like lightning.
I've never gotten a speeding ticket, but I've
hit a lot of trees.
John Kensmark
> Actually, the typos/hurricans example is a good one. Of course
> there is no real correlation between the two events, but
> there's almost certainly a selection effect going on. This kind
> of thing happens a lot when you're looking for a correlation
> between two unrelated events. If you're consciously looking for
> a connection, you'll tend to notice connections that aren't
> really there. When the supposedly correlated event happens,
> you'll look harder. When it doesn't, you won't look as hard.
> It would be easy to unconsciously miss typos when reading the
> front page of a newspaper (a process which is in and of itself
> easy to do bad proofreading on), and other errors could be used
> to tweak the final result: Is using the wrong a typo? bad
> grammar? a word or idiom that the proofreader doesn't know? etc.
I used to play selective perception tricks to make my then nearly
two-hour commute less dull. Frex, because I saw a co-worker eating
candy one day, I decided to see how many people that I saw on my way
home were eating candy. You would not believe how many people on
Boston's subways are eating candy, once you look for it, but I'd
never noticed it before.
A lot of things that folks argue over as to whether they're
scientific or not seem to come down to selective perception--are
those chimps using sign language, or are people fooling themselves,
etc. I've often wondered if, in the reasonably near future, AI
systems might resolve a lot of these issues by being truly impartial
observers not affected by selective perception.
Seem reasonable? Or am I just fooling myself?
--
John Kensmark kensmark#hotmail.com
Today's Word -- agelast: one who never laughs.
Brett Evill
>
> But I suspect I'm going to have to go back and reread a couple of
> books. I know Niven created tanj, but a niggling part of my brain says
> he mentioned tanstaafl in _Ringworld_ or _Flight of the Horse_.
Maybe he did. But 'The Moon is a Harsh Mistress' was published in 1966,
so I doubt Heinlein was quoting Niven.
Regards,
Brett Evill
Erik Max Francis
> But I suspect I'm going to have to go back and reread a couple of
> books. I know Niven created tanj, but a niggling part of my brain says
> he mentioned tanstaafl in _Ringworld_ or _Flight of the Horse_.
Yes, Niven used it a few times. He got it from Heinlein, however.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ The great artist is the simplifier.
\__/ Henri Amiel
Maths reference / http://www.alcyone.com/max/reference/maths/
A mathematics reference.
John David Galt
> Since this is rasfs, I'll quote Niven: TANSTAAFL.
Not Niven. The best known appearance of this is a Heinlein story
already mentioned, but I believe Sturgeon is the original source.
Joseph Hertzlinger
wrote:
>Since I've started keeping notes, there have been 436 occurances.
>According to pure statistics, this means B should have happened 8
>times, and it's happened 7.
>>With 1000 occurences of A, that means C has happened twice ?
>
>My first 'observation' was over 1400 A's ago, but I attributed it to
>pure coincidence. When it happened 436 A's ago, it made me curious.
>It's third occurance has started me wondering.
Let's consider how many events you could find possible correlations
between. If there are 1000 possible events resembling B (e.g., a
hurrican hitting North Carolina or a large uptick in the stock market)
and 100 possible events resembling C (e.g., three typos in _The New
York Times_ or your cat throwing up) and you look for all possible
correlations between those types of events you're likely find at least
one correlation with odds of 100,000 to 1 against.
I have read that critics of the "efficient market hypothesis"
frequently find non-random correlations in stock prices. For some
reason, most of those correlations stop working after a while.
>I follow the general rule:
>Once is happenstance.
>Twice is coincidence.
>Three times shows that you're not paying attention.
I thought the last line was: "Three times is enemy action."
pervect
> Now here's my question:
> At what point do you start to believe that C is a reliable predictor
> of B when there is no known science that could possibly link the two?
Well, first of all you have to decide what confidence level a "reliable"
predictor is. How often are you willing to be wrong about a "reliable"
predictor? A lot depends on the consequences of being wrong, in some
situations any probability greater than 50% is good enough to take a gamble
on, in other situations much more confidence is required.
It would appear that event C occurs before event B - this is implied though
not directly stated from what I could tell.
The statistical approach is to assume that B and C are totally random and
uncorrelated, and to use that as a model to derive the probabilites. The
result of this is simply a statement of the probability that this assumption
is wrong - that there is some sort of correlation between B and C.
Unfortunately, this result, though helpful, is not the same thing as saying
that C is a predictor of B. In general, when B and C are correlated, it's
possible that B influences the probability of C, that C influences the
probability of B, or that some set of events D influences the probability of
both B and C.
Other assumptions are needed for the statistical approach to work. One is
assuming that the statistical processes involved are "stationary" - not
changing with time. This assumption may not always be the case.
There are other potential pitfalls as well. It's easy to examine a large
number of predictors and find one that appears to have a probability of >
50% by pure luck. It strikes me as a wise idea not to draw any conclusions
about the validity of B as a predictor until one has examined data collected
after the hypothesis was formed - though I don't know how to analyze this
issue formally.
It goes without saying that one has to be confident in one's data collection
process, and look closely at the methodological issues of how events B and C
are determined - are these determinations really objective?
Out of all these problems, the first one I mentioned, the unrecognized
factor D, is probably the main reason that science demands controlled
experiments, and does not speculate about scenarios like this one. By
controlling the experiments, one can remove (or attempt to remove) the
troublesome category of D events - this is important when one is seeking
causes and effects, rather than just correlation's. Dealing only with
repeatable results is another important way that science attempts to
minimize the problem of category D events - if category D events are
important, but not controlled, the results will most likely not be
repeatable.