Talk:Inverse gambler's fallacy

Hint that the assumptions are wrong
The German article links to an article by Bostrom, which proves that the IGF is wrong:

N. Bostrom: Anthropic Bias, Observation Selection Effects in Science and Philosophy. Routledge, 2002, ISBN 0-415-93858-9 — Preceding unsigned comment added by 178.200.146.119 (talk) 13:16, 21 December 2014 (UTC)

The passage from the German Wiki:

''Bostrom hat zudem aufgezeigt, wie das von Hacking angegebene Beispiel, welches zum umgekehrten Spielerfehlschluss führt, modifiziert werden müsste, so dass es tatsächlich mit der anthropischen Argumentation vergleichbar wäre. Um selektiven Beobachtungseffekten Rechnung zu tragen, müsste demnach in Hackings Beispiel mit dem Würfelspiel ein Spieler solange außerhalb der Spielhalle warten, bis eine Doppel-Sechs geworfen wurde. Unter diesen modifizierten Bedingungen wäre der umgekehrte Spielerfehlschluss aber kein Fehlschluss mehr. Vielmehr könnte ein Spieler unter diesen Bedingungen, wenn er nach einer geworfenen Doppel-Sechs in die Spielhalle eingelassen wird, tatsächlich zu recht schließen, dass bereits eine mehr oder weniger große Anzahl von Würfen stattgefunden hat.''

Minor grammar correction
Changed in the sentence,

"A rebuttal paper[2] by John Leslie points out a difference between the observation of double sixes and the observation of fine tuning, namely that the former is not necessary (the roll could have come out differently) while the latter is necessary (our universe must support life, which means ex hypothesi that we must see fine tuning).

in the parenthetical statement, "(the roll could have come out differently)", "differently" to "different". The word is not modifying the "coming out" process, like the dice hitting the backboard harder and coming back farther, but the result. As it was, it's like writing, "I scored lowly on the test." It was awkward at best and I would argue, grammatically incorrect. "Different" describes (the results of) the "roll", doesn't modify anything verbal. — Preceding unsigned comment added by 83.183.102.137 (talk) 18:19, 25 May 2013 (UTC)

Not Real
Theoretically, the fallacy seems sound but unfortunately it does not represent reality. Toss a coin n times and count how many times it comes up heads. While the statistical probability is exactly 50% with each toss of the coin, over time the statistical likelihood of recurrence will diminish exponentially with each toss.

The Gambler Fallacy argument is invalid in multiverse theories.

- First, I don't think we need to be taking stances on whether inferences to multiple worlds do or do not commit the inverse gambler's fallacy. There is supposed to be a little neutrality, especially when it comes to controversial philosophical issues. That's why I originally just said "Defenders of the argument under criticism usually appeal to an observational selection effect that, allegedly, makes all the difference. But it would take us far afield to address this response." There is no need to try to solve everything.

Second, the stance taken doesn't even make sense to me. The article now says that, since the universe is a specified run, then it's not a fallacy to infer multiple universes. But the fact that the universe is a specified run supports the claim that the fallacy is committed, it doesn't undermine it.

>From the article: "The multiple universe hypothesis makes it more likely that life-permitting constants turn up in some universe or another. But it does not affect the probability that they turn up here in our universe (on this specified 'run'); that probability remains as low as it ever was."

A general, non-specified existence claim is confirmed. But the non-general, specified existence claim is not confirmed. Arguing that life-permission is just such a specified existence claim therefore undermines the inference to many universes.

DrRetard

God Related
I find the topic interesting, although somehow I find the discussion (both in the article and here on the talk page) difficult to follow. I studied probability and statistics in high school, and I know how to do simple calculations about lottery tickets and whether I'm safer in a car or an airplane.

Is this topic related somehow to the likelihood that God exists or not? --Ed Poor


 * No. My understanding of this article is this: there is a commonly-used argument that goes roughly "it's highly improbable that life came to be in the universe -- surely then, something must have put it in motion." This is fallacious reasoning. That does not mean god does not exist, or that god does exist -- it merely says that one argument is not valid. -- Tarquin


 * The original article as I understood it implied that said argument was valid. When you read it for the first time (a few hours ago) I had probably already rewritten its end and turned its conclusion upside down. Actually, I've had the impression that the whole original article just aimed at introducing and justifying that argument; hence the thing looses cohesion if the conclusion is reversed. But I'm quite sensitive on the invalidity of that argument, so my reading may be biased. -- FvdP 17:39 Oct 8, 2002 (UTC)
 * (but see below) -- FvdP 18:32 Oct 8, 2002 (UTC)

Overlap
There are two different arguments that can be said to fall under "inverse gambler's fallacy":
 * 1) a "proof" that multiple universes exist
 * 2) a counterargument to the following theist argument that God exists: "it's highly improbable that life came to be in the universe -- surely then, something must have put it in motion."

These things are quite different. The fallacy may or may not apply to 1 (multiple universes), it probably does, I'm not definitely sure. What I (FvdP) and (I thought) Hacking are after is its application to 2 (God's existence), and there I stand that the inverse gambler's fallacy clearly does not apply. My guess is that the misunderstaning between us (FvdP & DrRetard) is mainly a misunderstanding about whether the article's arguments address point 1 (multiple universes) or point 2 (God's existence).

Now point by point:
 * First, I don't think we need to be taking stances on whether inferences to [multiple worlds or God's existence] do or do not commit the inverse gambler's fallacy ? Yes we should, because by presenting Hacking's argument, the article is taking stance anyway.


 * Second, the stance taken doesn't even make sense to me. The article now says that, since the universe is a specified run, then it's not a fallacy to infer multiple universes. But the fact that the universe is a specified run supports the claim that the fallacy is committed, it doesn't undermine it.  I agree with you that applied to the multiple universes hypothesis, my stance does not make sense. But as explained above, it was really about God's existence (in my intent at least).

Actually, I must confess (after a few readings of the article) that I don't understand Hacking's point. Would he be actually fighting on 1 (against multiple worlds) rather than 2 (for the "fine-tuning" pro-God argument) ? (Or is he by chance equating the two questions ?)

My point is really about this: "and if there are, then life-permitting constants had to come up somewhere or another, why not here?. To which Hacking somehow answers "why would they appear here ?". I say "why here ? But simply because we are here ! the choice of our universe, because it's by definition the choice of a universe where life exists, is biased towards finding fine-tuned universe parameters (regardless of the existence of a designer)". It's like stopping on the next dice run that issues a 6-6. That surely increases the probability of someone seing a 6-6.

Probability arguments for or against God's existence lead us in muddy waters anyway ;-)

FvdP 18:32 Oct 8, 2002 (UTC)

--

First, the point of the original article: Hey, there's this fallacy. We have a tendency to commit this fallacy. We have an even stronger tendency in certain contexts. There's a popular response to an argument for theism, and this response might commit the fallacy.


 * I'll answer by inserting indented comments as they'll be (mostly) short. FvdP 20:36 Oct 11, 2002 (UTC).

Second, my personal convictions: there is no God, and the fine-tuning argument is pure codswallop. I still find the issues interesting, though.


 * Same here (but you know that already). FvdP

Third, some specific responses.

No. My understanding of this article is this: there is a commonly-used argument that goes roughly "it's highly improbable that life came to be in the universe -- surely then, something'' must have put it in motion." This is fallacious reasoning. That does not mean god does not exist, or that god does exist -- it merely says that one argument is not valid.''

The original article doesn't address such an argument. Rather, the article addresses a popular response to the fine-tuning argument for theism. Here's the response: "The obtaining of such wildly improbable life-permitting cosmological constants is good evidence that there are many other universes; and this explanation is at least as good as the God-explanation." The article doesn't say whether this response is good or bad. It says that the response might commit the fallacy, and that there's a debate about whether it does.

The original article as I understood it implied that said argument was valid"

The original article says nothing about the validity, soundness, or persuasiveness of the fine-tuning argument itself. Rather, it addresses, without deciding, a popular response to the argument. In any case, I'd never say that the fine-tuning argument is any good; I think it's codswallop.

"There are two different arguments that can be said to fall under "inverse gambler's fallacy":
 * 1) a "proof" that multiple universes exist
 * 2) a counterargument to the following theist argument that God exists: "it's highly improbable that life came to be in the universe -- surely then, something must have put it in motion."

These things are quite different. The fallacy may or may not apply to 1 (multiple universes), it probably does, I'm not definitely sure. What I (FvdP) and (I thought) Hacking are after is its application to 2 (God's existence), and there I stand that the inverse gambler's fallacy clearly does not apply. My guess is that the misunderstaning between us (FvdP & DrRetard) is mainly a misunderstanding about whether the article's arguments address point 1 (multiple universes) or point 2 (God's existence)."

Hacking thinks that the "oscillating universe" version of the popular "multiple universe" response commits the fallacy. White and Dowe think that all versions of the popular response commit the fallacy. Nobody, so far as I know, think that the fine-tuning argument commits the fallacy. I mean, sure, there's a hell of a lot wrong with the fine-tuning argument, in my opinion. But it doesn't fallaciously infer from an improbable result that there are or have been many trials of a similar nature. And that fallacious inference, after all, is the inverse gambler's fallacy.


 * OK. FvdP

*First, I don't think we need to be taking stances on whether inferences to [multiple worlds or God's existence] do or do not commit the inverse gambler's fallacy'' ? Yes we should, because by presenting Hacking's argument, the article is taking stance anyway.''

This is just bizarre. In any philosophy article, you have to present arguments. The key is to try to give the arguments a fair shake, even if you disagree with them; and you have to try to give popular criticisms a fair shake, even if you think the argument is good and the criticisms lousy. None of this is 'taking a stance', the way that, say, an Everything2 article might do. When an issue is controversial, with bright people on both sides, then try not to take a stance. Simple enough, I would think.


 * Simple enough ? Simple enough if we agree on words, and on the fact that the question is indeed disputable. About words, please take into account the fact that I'm not a native english speaker. By "the article is taking stance" I meant "the article takes position in favour of Hacking's argument that (some) proof a the multiple worlds hypothesis commits the i.v.f.". And this because (1) Hacking is taking stance this way; (2) the article seems to take Hacking's argument as granted and does not even present the contradictory view ("the argument to the multiple worlds does not commit the inverse gambler's fallacy"); (3) hence the article is taking stance.
 * I believe that if the answer is obvious and undisputed, even a philosophy article should take stance. (This has more to do with NPOV than with philosophy, actually). My error was believing that the answer was fairly obviously opposite to Hacking's (I no longer think so), so that we should take stance accordingly (even while at the same time presenting Hacking's argument).
 * And finally, no need to patronize me with "Simple enough", thanks. Don't forget that in this dispute too, there are relatively bright people on both sides, even though I might be dense at times. FvdP

''Actually, I must confess (after a few readings of the article) that I don't understand Hacking's point. Would he be actually fighting on 1 (against multiple worlds) rather than 2 (for the "fine-tuning" pro-God argument) ? (Or is he by chance equating the two questions ?)''

Hacking thinks that someone who looks at the (alleged) improbability of life-permitting constants and thence infers the existence of many previous oscillating universes is committing the inverse gambler's fallacy. Just because the improbable result would show up on some run doesn't mean it would show up on this run. So it's improper to infer multiple runs.

My point is really about this: "and if there are, then life-permitting constants had to come up somewhere or another, why not here?''. To which Hacking somehow answers "why would they appear here ?". I say "why here ? But simply because we are here ! the choice of our universe, because it's by definition the choice of a universe where life exists, is biased towards finding fine-tuned universe parameters (regardless of the existence of a designer)". It's like stopping on the next dice run that issues a 6-6. That surely increases the probability of someone seing a 6-6.''

But no one disagrees with this. If there are many universes, then there's nothing surprising about us living in a universe with the improbable bundle of life-permitting constants. The question is whether we can infer that there are indeed many universes, from the observation that we live in such a universe. Just because a hypothesis, if known true, would render an improbable result unsurprising doesn't make it OK to infer the hypothesis from the observation of the improbable result.


 * I have no problem with that argument expressed in that way.


 * Could you also please identify yourself ? You did not sign, you were not logged-in, so all I know is your IP address. A nickname would be better (e.g. are you DrRetard or not ?) This would give a better sense of continuity in the discussion, in case you already participated or will participate more.
 * FvdP 20:36 Oct 11, 2002 (UTC)

Yeah, sorry, it logged me out and I was in a hurry to be done with the post! I apologize for any patronizing tone. And if my original article caused all this confusion, then it probably does need some rewriting! For one thing, I'll try to present the response to Hacking, et al. in greater detail than before. --DrRetard

Weak Anthropic
As I understand it, the many-worlds theory and inverse gambler's fallacy do not apply to the Weak Anthropic Principle (if that's what you are referring to with the fine-tuning argument). Instead, what it says is the probability of a universe favorable to intelligent life, conditional on the probability that intelligent life exists, is high.

How likely is it that you would have been born? Your parents might not have met, one of them might have died, a different spermatozoon might have fertilized your progenitor egg cell. You could commit the inverse gambler's fallacy and say that you were very lucky to be born. But nobody could claim that they were unlucky to not be born -- because people that were not born don't exist! Everyone who exists is "lucky"; but when we only sample from the "lucky" population, we should not be surprised to find out that everyone got lucky.

From the article:
 * today's best cosmologies allow for the possibility that there are many universes besides ours; and if there are, then life-permitting constants had to come up somewhere or another, why not here?

But it's not here in particular -- if we think we are special, then we have committed a fallacy. All the living universes (that may or may not exist) may be surprised that they are alive; none of the dead universes (that may or may not exist) will notice that they are dead.

I hope I made sense. --hb

ps. similar argument in quantum suicide. --hb

Not Many Worlds
I don't see the gambler's fallacy as a counterargument to the many worlds theory. A better analogy would be to think of yourself as the $50 bill in the machine. You don't know how many times the machine has been run, but you know the number 17 has come up at least once, since you have been spit out of the machine. From your perspective, there are three possibilities:
 * 1) The machine has been run only once, and the one player got lucky.
 * 2) The programmer rigged the machine so 17 would come up every time.
 * 3) The machine has been run many times.

Which seems more likely?

--erall

"dice have no memories"
Should it not be "dice have no memory"?

Gah...

I really don't know.

Acegikmo1 05:14, 12 Oct 2004 (UTC)

Google suggests both are used, but "dice have no memory" is more popular. That's also the version I would prefer. Memories are what you remember, a memory is what you remember with (or in, if you prefer). Multiple "memories" in the last sense are weird. 82.92.119.11 23:33, 21 Dec 2004 (UTC)

Cosmogony
If this is all about a many worlds/creationism debate, it should say so in the intro. Alternatively, having followed the link from gambler's fallacy, I would have expected it to start with something like:
 * You see a pair of fair dice rolled once, and the result is double-sixes. This is a quite improbable result, so you conclude that the dice were probably fixed to always roll six.

That seems a more natural reaction (and inverse) than the rolled many times before position in the article as of now. Indeed, there are Bayesian arguments why you might be slightly more inclined to believe the dice have a bias than you were before the roll. --Henrygb 18:17, 31 Mar 2005 (UTC)

Okay, don't know where to put this, but why the "scare quotes" around intelligent design (and, for that matter, fine-tuning and worlds) in the multiple-world hypothesis section? --Matt O'Connor

Similar reasoning
Here is an argument similar but not equivalent to the one the article discusses.


 * Someone brings you a pair of fair dice which he rolled, revealing a result of double-sixes. This is a quite improbable result, so you conclude that the person probably rolled the dice many times before choosing to show them to you.

Logic or fallacy? (The person makes no claim as to why he brought the dice to you, or whether or not it is his first result. The diffence is that the dice are not rolled in front of your eyes, so the person may in fact have only come running to you when he got an interesting result.)  -- Milo


 * You can't even claim that he rolled them :)
 * Anyway, if your observation was not conditional to the dice reading 12, then you cannot make any inferences about past rolls. If he has something to gain from rolling the sixes, then you have reason to suspect your observational status and the outcome are not independent, and can accuse him of cheating with some significant probability :)
 * --Andy —Preceding unsigned comment added by 24.233.77.207 (talk) 15:50, 30 October 2009 (UTC)

Encyclopedic style
This article doesn't seem to be written in a style coherant to that of an encyclopedia in places. Would a cleanup be in order? 80.42.209.188 22:05, 7 March 2007 (UTC)
 * I agree (and came looking here for others who might have thought that) and I have now tagged it. Specifically, the Socratic dialogue is not very encyclopedic, and even occasionally seems to stray into POV. &mdash; brighterorange  (talk) 17:12, 9 April 2007 (UTC)
 * I replaced the tag with the rewrite one. It seems to read more like a textbook than an article, therfore it needs a complete rewrite.--Orthologist 21:56, 17 April 2007 (UTC)

Multiple worlds hypotheses
I agree that this is an unnecesary section of the article but what should we do ? I think that we should create a link to it on a seperate page, because it is of only moderate relevance, not worrth mension on the article it's self. I would like to hear some opinions on this subject. --Bobnessownage

Mmm, there no longer seems to be any discussion of how this relates to a multiverse / world ensemble theory anywhere in Wikipedia that I could find.

The Multiverse idea isn't motivated by fallacious thinking - there's a key difference between this fallacy and the inference to a multiverse. The multiverse case is more like this:


 * Someone tells you that he's going to roll a dice a certain number of times (but he doesn't say how many) and if he rolls 6-6 at any point he'll stop and come show them to you. And indeed, he does come and show you a 6-6. So should you conclude that he rolled more than once?

Yes, probably. The point is that 6-6 is the only result you could possibly have seen. Likewise, an apparently fine-tuned universe is the only result you could possibly have seen if you exist to see anything at all.

If we were to select a random universe and look at it, and it turned out that it contained intelligent observers, that wouldn't indicate how many universes there are. But that's not what we're doing. Evercat 23:01, 1 August 2007 (UTC)

Error in article
On a much more mundane level, is there not a fallacy in the John Leslie example. No doubt a statistician will correct me if I am wrong, but do we not need to use a Bernoulli distribution root(p)(1-p) where p=1/36 for the expected number of throws before a double six is obtained, giving an "at most" of about one in 20.8 not 1/36. I do not have access to the John Leslie article, so I don't know whether the error is in his article or in the editor's interpretation, but I suspect the former. Can anyone check? Shall I add a note that there is a mathematical error here, to avoid perpetuating the minor fallacy? Dbfirs (talk) 21:00, 15 January 2008 (UTC)


 * I'm not sure where you get 1/20.8. If you want the number of throws until the first success (inclusive), then that would be governed by the geometric distribution, but the expected value of that is simply 1/p = 36 here. What the article talks about is the probability that it took only one throw to get a double six, which is p = 1/36. Since there's the additional twist that what we see is not necessarily the first success but just some success, these values turn into lower and upper bounds, respectively. -- Coffee2theorems (talk) 21:59, 18 October 2008 (UTC)

Lots of fallacies about the fallacy
The fallacy is utterly unrelated to the argument from design.

Philosopher number 1 is using a straw man. Philosopher number 2 has come to the wrong conclusion.

Both are incredibly wrong, because they have yet to consistently define a roll in the metaphor.

Thus they are engaging in the fallacy of equivocation. Does a roll represent the process caused by the big bang? Or does a roll indicate a sub-process pertaining to the big bang?

The other assumption that is being made is that fine-tuning is required for living creatures.

If you don't know what the dice look like or what causes them to roll or how often they roll, (the dice could be misshapen, labeled with negative numbers as well, and there may be one or a million), how can you possibly make any inferences about what the results would normally be?

You can't. Both philosophers should consider becoming lawyers.

Deepstratagem (talk) 13:15, 24 September 2008 (UTC)


 * Not very much can be said, true, but something can be (taken from the external links of Fine-tuned universe). Might even be worth mentioning in the article as another attempt at saying something definite about the problem from a Bayesian viewpoint (or maybe not, it's unfortunately only a web page AFAIK - maybe as statisticians, as opposed to philosophers, the authors considered the subject as a mere curiosity). -- Coffee2theorems (talk) 22:41, 18 October 2008 (UTC)

Bayes theorem in the lead
I haven't been able to download Hacking's original paper, but does he explicitly say that the inverse gambler's fallacy is a fallacy of a misinterpretation of Bayes theorem? Or, does he just say that it is a fallacy (not explicitly mentioning Bayes)? Right now I find the discussion in the lead, where Bayes theorem is invoked, to be confusing. It almost reads like there is a fundamental problem with Bayes theorem, when I think the point is that a misinterpretation is being applied to Bayes theorem. Can someone help on this? Thanks, Grandma (talk) 16:03, 7 December 2014 (UTC)

Couple things: untangling the bloated 'real life examples' section and cleaning this talk page
First, the real life examples section has grown into a big, unfocused morass of example, criticism, counter-criticism(?), and further support for the idea. I don't think it's too controversial to suggest parsing things down a bit. Even just adding a 'criticism' section, to just pull that stuff out into its own area would make the article more understandably arranged. I would happily start on this myself, but I'll gladly defer the task to those wiser than I (which is probably most of you) or has alternative ideas on this. I'm on standby on that until/unless it stalls out or someone offers a decent reason not to or an alternative, which I will gladly discuss.

Secondly, the age of the comments on this talk page are insane. 2002 and likely before? Just like the article, this place could use some archiving, or at least a tidy up. Anyone want to tackle that? I'm guessing no, so, anyone care if I work on it? Let me know. I'll wait a while, and clearly this page will too...! Buddy23Lee (talk) 02:04, 17 March 2016 (UTC)

Clarification of "dice had been rolled many times before" needed
For example, if one observes...double sixes, it is wrong to suppose that...the dice have been rolled many times before.

The text ought to explain the reasoning which would lead to such a supposition. It was unclear to me, and I cannot express it in the correct tone.

One might suppose that the dice had been rolled many times before because—why? Because the "bad luck had all been exhausted by previous rolls"? No, that's not it.

This seems to be the Gambler's fallacy redux, if an observer supposes that the double sixes have appeared because of the influence of past rolls. catsmoke (talk) 18:15, 7 March 2022 (UTC)