August 6, 2002

11:36 PM

The Doomsday Argument Rarely does philosophy produce empirical predictions. The Doomsday argument is an important exception. From seemingly trivial premises it seeks to show that the risk that humankind will go extinct soon has been systematically underestimated...[more inside]
posted by crunchburger (50 comments total)

From Nick Bostrom, who got some notoriety a few months ago for his simulation argument.

The idea is this: imagine you are going to pick a number from a lottery. There will be either 10 or 100 numbers in the lottery - you don't know which.But in either case, every number has an equal chance of coming up. Your number is 7. Knowing this, do you think it is more likely that there were 10 numbers in the lottery, or 100?

Apparantly the rules of inference and probability compel you to conclude that it's more likely that there were 10 numbers than that there were 100 numbers.

Now suppose that your order of birth in all the human beings who will ever live is equal to, say, the 8 billionth. Estimate how long humanity will survive. Parity (if we accept the lottery argument) compels us to the conclusion that, humanity, rather than continuing for many thousands of years and growing to, say, a trillion people, will probably die out soon. From disease, nuclear war, God turning down the lights and winding up the show, aliens using us as a food animal, whatever.

Does this argument have any weight, or is it merely dorm room bong session fare?
posted by crunchburger at 11:41 PM on August 6, 2002

I have a copy of The End Of The World, and after four years of reading and re-reading it, I have yet to find a flaw in the Doomsday Argument.

The book states the argument much better than the article does, and also offers a few common rebuttals to it, and why they don't hold any weight.

Highly recommended reading for all.
posted by Jairus at 12:05 AM on August 7, 2002

It's that "all the human beings who will ever live" variable that makes this bong session fare for me. Who's to say that I'm Human being number 60 billion out of 61 billion total "human beings who will ever live", or out of 600 trillion "human beings who will ever live".

Just because our (human race) run to date appears to be luckily unbroken doesn't mean it's bound to end soon. That sounds like saying you're about to win the lottery because you've played it so many times that you're bound to hit the right one soon. This feels like a semantic quirk that arises out of the rules of inference and probability you mention, not something to really worry about in real life, IMHO.
posted by kokogiak at 12:08 AM on August 7, 2002

The reason this argument is flawed is because whether or not Doomsday will occur is not a random event. In the explanation of the argument, the thought experiments presented in Step I and II assume explicitly that the number of people in the cubicles is randomly determined. So, if you assume the same thing about Doomsday - that it will randomly either happen or not - then the argument is valid. Most people would not agree with this assumption, though.
posted by yoz420 at 12:09 AM on August 7, 2002

Nick Bostrom is, in my very humble opinion, a self-promoting idiot, a mountebank who other highly-credentialled academics as well as lay types like my ownself do not take particularly seriously.

His problem here, as it was in the "simulation" thing mentioned, is in the sloppiness of his premises.

At any rate, I was glad to see the article at least provided for a third possibility between "Doom Early" and "Doom Late," something my gut tells me actually has the highest probability associated with it, as strange as it may seem at the moment. That possibility is humanity evolving sufficiently radically, in the relatively near term, to make nonsense of such calculations such as these.

Don't get me wrong. I'm nobody's Extropian. I just pay attention. ; . )
posted by adamgreenfield at 12:17 AM on August 7, 2002

The simplest way to see what's wrong with it, or at least the simplest way I've come up with in the past five minutes, is this: there is no possible set of conditions under which the Doomsday Argument, as given, would not apply. That is, at any given time, even, say, 300,000 years ago, we can pick some random person, take their birth rank as given, and prove that they are most likely one of the last few human beings ever.

It predicts that any time is most likely near doomsday. The vast majority of times are not near doomsday. Therefore etc., Q.E. to the motherfuckin' D.
posted by moss at 12:27 AM on August 7, 2002

Bah.

Even Bostrom was never able to counter the assumption (which would wreck this argument) that humans will exist for infinity. (how can this happen you say if the universe is finite? Perhaps its not or we will learn in an advanced technological age how to create universes of our own?)

Since this cannot be dismissed then the entire argument loses its force. These Bayesian calculations assume not only that we have boundaries on the past and present but on the future which we do not. I'd add to the argument that we do not have a boundary in the past either. When exactly (date and time please?) did we become human?
posted by vacapinta at 12:32 AM on August 7, 2002

Here's a fun one:

Premise: I am an organism on planet Earth.
Conclusion: I am a bacterium!

What? I'm not. Damn probability.

kokogiak: Don't get greedy. Pass the bong.
posted by vacapinta at 12:38 AM on August 7, 2002

For what it's worth, all above mentioned objections to the doomsday argument are covered in the book, along with reasoning as to why they're not valid objections. The webpage doesn't even scratch the surface of the argument, and anyone interested in the topic should read The End Of The World.

Poking holes in the rough outline of the argument presented on the website is like shooting fish in a barrel. Finding flaws in the 300-page monograph is much more difficult, I assure you.
posted by Jairus at 12:55 AM on August 7, 2002

There were never any people whom I've ever had a bong session with that would have ever come up with this shite. This opens up a can of worms of the probability of a bong session tending toward the use of this kind of arbitrary premise for probability reasoning. Ah, it's late...
posted by Wizzle at 1:03 AM on August 7, 2002

moss: You beat me to it. Something like "Ugh, caveman Ogg tenth human. URGH! Caveman Ogg LAST human!"

vacapinta: Damn. I'm a nematode.
posted by Hieronymous Coward at 1:03 AM on August 7, 2002

If he's correct, then really all I have to say about it is an atheistic "thank God." If my personal perceptual universe is going to end with utter finality, I don't see why anyone else should be allowed to go on.

Nihilist hangups aside, I have this annoying sense of absolute justice that humanity has been raping forwards, backwards, and sideways lately. Good to see it all go, if you ask me. I don't have much hope even for humanity successfully eliminating itself, however. I expect it would just evolve into something worse, less intelligent, less just, and better at accumulating resource.

"The supernova bomb?" said Marvin. "It's a very, very small bomb."
"Yeah?"
"That would destroy the Universe completely," added Marvin. "Good idea, if you ask me. They won't get it to work though."
"Why not, if it's so brilliant?"
"It's brilliant," said Marvin, "they're not . . . they've spent the last five years building it. They think they've got it right but they haven't. They're as stupid as any other organic life form. I hate them."

--Life, the Universe, and Everything by Douglas Adams
posted by Ryvar at 1:13 AM on August 7, 2002

Poking holes in the rough outline of the argument presented on the website is like shooting fish in a barrel. Finding flaws in the 300-page monograph is much more difficult, I assure you

Fine but we have nothing to go on other than your assurances. Also, dont assume how much I have or havent read.

My first argument about infinities is addressed by nick bostrom on his site:

John Leslie thinks that the DA gives us supremely strong reasons against any hypothesis implying the existence of an infinity of observers. In fact, it seems to follows from Leslie’s view that the probability should be zero. But intuitively that seems too strong: could we really rule out the possibility of an infinite population from our armchair, so to speak, without even knowing our birth rank? (For we can assume that we knew that we had some finite birth rank, and for any finite birth rank we turned out to have, the argument would apply—it would have been "infinitely improbable" that we would have had that low a birth rank if the total population were infinite.

My opinion is that the infinite case has not yet been completely settled and that it might well represent a possible scenario (one of the more desirable ones) that is not ruled out by the DA.

And my second facetious argument is really about reference classes and the proper definition of. Bostrom says:

the problem of the reference class. Briefly stated, this is the problem of what class of entities that you should consider yourself a random sample from. Is it the class of all conscious entities? Or all entities that have a conception of their birth rank? Or all entities that are intelligent enough to be able to understand the Doomsday argument if it were explained to them? Or all entities who are in fact aware of the Doomsday argument? In my opinion, the problem of the reference class is still unsolved, and it is a serious one for the doomsayer

All this from Leslie's biggest defender.
posted by vacapinta at 1:19 AM on August 7, 2002

The problem I see with the analogy of "birth order number" to "cubicle numbers" is one of time and reproduction. The fact that I am observing myself as the 50 billionth person isn't meaningful since other observers have not had the chance yet to make the observation because they haven't been born yet and won't be for maybe thousands of years. The cubicles don't produce more cubicles in the future. The fact that I am in #7 now is meaningless if I now that these first ten will make more at some point. If the mathematical analogy scales to a reproducing population, the idea would work far better and I imagine be far more complex.

I find the link interesting, however, and I will check out the book.
posted by McBain at 1:35 AM on August 7, 2002

Please tell me if I am misunderstanding the concept, but I don't see how this is any different from common sense. For all humans alive today, it is certain that doomsday will not arrive before their birth. Any prediction beyond today, and you loose certainty as a function of how far you attempt to predict. A prediction on the scale of the numbered cubes example, that we have lived only one tenth as long as we will, introduces roughly ninety percent uncertainty (91%, as calculated). However, at any point during that time the probability is quite respectable that we will live half again as long as we already have.
posted by Nothing at 1:51 AM on August 7, 2002

Well then, at what evolutionary point can you say that humans are no longer humans? I mean, how can any species last infinitely if it keeps mutating over time.

Then, somewhat contradictory to my own post, why have simple organisms cockroaches been on earth so long and have changed so little? How have they beaten the lottery theory for so long?
posted by Aikido at 2:10 AM on August 7, 2002

Without having read the book I'm with the dorm-bong thesis.
Basically I just don't accept the move from Stage II to Stage III - the analogy between the numbered cubicles and the end of the world.
But the key thing here, surely, is this is not an 'empirical prediction' - it is a philosophical conundrum, like 'the paradox of the surprise examination.'
(http://www.wischik.com/lu/philosophy/surprise-exam.html)
It tells us NOTHING about the human race's chances of survival, the fun is to find the flaw in the argument - and like the 'surprise examination', it is much more difficult than it seems.
posted by rolo at 2:18 AM on August 7, 2002

vacapinta:

But there's the neccessary bottleneck of getting past the "Drake Equation", which is best exemplified by humanity and all of Earth's biota being planet-bound at the moment. One way or the other, as far as we know, anything at all that we do ends ultimately in existential futility. However, however, however, give humanity a bridge to circumvent the bottleneck and who knows what great capabilities we're good for. Not to mention the heretofore misunderstood and/or not ready for prime time intellects of other cousins in our mammalian kingdom.

I mean, what if we did discover a species far away in another star system eventually? No, no capability at radio yet. But still a species we would say, that could eventually, Prime Directive style, could on it's own, in 500,000 years, discover radio waves and how to communicate through them. Say they had the equivalent intelligence of the bonobo. Would we file them away in the way we terrestrially file away the bonobo as mere "endangered species" interest? Hell no. They'd be heralded as "Intelligent Beings Found Around Beta Pictoris". Even though we have specimens of "lesser" intelligence extant on Earth which warrant no such headline.

Where am I going with this? I do not know.
posted by crasspastor at 4:07 AM on August 7, 2002

I have a copy of The End Of The World, and after four years of reading [...], I have yet to find a flaw [...].

:-)

if you still can't find an argument against it after 10 years, does that make it more right? after 20, 30? i first heard this argument when i was postgrad - it appeared in a letter to nature in about 1992 - and civilisation is still here.

(one interesting thing to consider is the error estimate for the prediction - what is it? with a single observation ("you are born now") is an estimate possible? if not, is a prediction with no error estimate useful?)
posted by andrew cooke at 4:18 AM on August 7, 2002

and if you truly believed civilisation was about to end, would you bother wasting time writing a book about it?
posted by andrew cooke at 4:19 AM on August 7, 2002

(Warning: The poster of this comment is a third-year undergraduate philosophy major, and therefore thinks he knows a lot more than he does.)

Rolo's got it right. What makes the argument frightening is Bostrom's immediate declaration: "Rarely does philosophy produce empirical predictions. The Doomsday argument is an important exception." On the contrary, however, the argument is not empirical in the least.

It is nothing more than an epistemological theory -- it attempts to tell us with what certainty we can predict the fate of humanity, as well as what the most sensible prediction would be. The proof is, of course, flawless -- the real fallacy lies in its application, because the truth of the matter is that predictions don't affect reality at all, much less dictate the future. To use the example put forth in the original link, yes, you are more likely to be right if you guess that you are in the blue cubicle, but your probability of being correct obviously doesn't affect the actual color of your cubicle. On a roll of the dice, betting on a result of "seven" every time will turn out the best results, but it obviously doesn't follow that seven is therefore the only possible number to roll.

Now, one might argue that the Doomsday Argument is still useful, because it may show us that Doomsday is more likely than not, but in actuality it tells us nothing, because it's based on the assumption that we know nothing about the conditions which will produce the result. No matter what we predict, Doomsday will either happen or it won't, and there's nothing our prediction can do about it, nor is there anything we can do based on our prediction. To stop Doomsday we'd have to know if it were actually going to happen (not to mention how!), and that requires real, empirical data, not philosophy.

The argument may be inescapable, but it's also pointless.
posted by tweebiscuit at 4:35 AM on August 7, 2002

i'm no expert on this, but it makes sense to me.
other ppl said :
The reason this argument is flawed is because whether or not Doomsday will occur is not a random event

well its an event with a certain probabilty of occuring.

In the explanation of the argument, the thought experiments presented in Step I and II assume explicitly that the number of people in the cubicles is randomly determined.

yes but thats just to explain how the reasoning behind it all works, i dont think it has anything to do with the actual theory

The fact that I am observing myself as the 50 billionth person isn't meaningful

agreed, but the fact that you are observing yourself as the 50bth person out of n number of people is.

The cubicles don't produce more cubicles in the future
Basically I just don't accept the move from Stage II to Stage III - the analogy between the numbered cubicles and the end of the world.


why not?
Stage I = 1-100 people
StageII = 1-n people
exactly the same thing, innit? the cubilces are just used to explain how the reasoning works.
posted by zoid at 5:01 AM on August 7, 2002

Q1. Does the Doomsday argument not apply to any point in human history? Should Cro-Magnon man have lived in fear of extinction? The Doomsday argument is equally valid for him. Why, then, is the human race still going strong after all these years?

A1. The answer is, yes, early humans could have used the Doomsday argument and they would have been misled. However, it is in the nature of probabilistic reasoning that in untypical circumstances it will fail. The earliest humans were in highly untypical circumstances so it's not so remarkable that they would have been misled.

Uh, "untypical circumstances?" Like, maybe, life?

Don't make me quote Jeff Goldblum.
posted by monju_bosatsu at 6:41 AM on August 7, 2002

Carl Sagan presents in several of his books (I read it in Pale Blue Dot)what is basically the same exact argument, but with the opposite conclusion. Similar to what Moss said above, Sagan argues that from a probabilistic view, it's much more likely that we are in the middle of humanity's existence, rather than near the beginning or end.
posted by skwm at 6:55 AM on August 7, 2002

My problems with the theory as presented:

a) As moss pointed out, the same theory would apply to any point in the past. If you believe the theory, then the human race has been almost certainly been about to go extinct at almost every century (or decade) (or week) for the last million years. It hasn't. Kind of an odd definition of "almost certainly."

b) Bostrom acknowledges that "Exactly how much more probable (Doom Soon is) will depend on the precise numbers you use (for the likelihood of Doom Soon).". Yet he glosses over this to say "In the present example (5% probability of Doom Soon events), the posterior probability of Doom Soon will be very close to one. You can with near certainty rule out Doom Late. " How strange. In his cubicle example he makes the probability of the coin toss as 50%, yet discovery of which cube you are in only raises the possibility of tails to 91%. (Very likely, but not "almost certainly.") Now he states that changing the odds from 50% to 5% makes no appreciable difference? How about if we calculate that the empirical odds of a humanity-destroying event are .00002%. Does Bayes theorem still produce a number close to 1? Let's see some math please.

c) Blather about other intelligent races bring part of the same reference class is a red herring. I could adapt Bostrom's argument to say "This amoeba I am looking at is the 5,000, 000, 000, 000, 000, 001st amoeba to ever live. The Doomsday argument shows that amoebas will almost certainly go extinct before the 5,000,000,000,000,000,101st is born." Just as logical. Just as wrong.

d) crunchburger - With regards to the lottery analogy: Yes, a lottery with 10 numbers is more likely than one with 100. Let's look at this a bit more carefully now. Given that your number is 7 (and that you know nothing about the way lotteries are generally constructed) it is more likely that there are 10 numbers than 20, more likely that there are 20 numbers than 30, more likely that there are 30 than 100, more likely that there are 100 than 1000, more likely 1000 than 5000, etc... That does not mean that there are almost certainly 7 and only 7 numbers in the lottery. (In addition, an examination of real world lotteries will show that they almost never have only 10 or fewer numbers.)

e) Forget talking about doom in the next century. If Bostrom's argument is correct, it applies just as well to humanity becoming extinct before lunchtime today. I don't expect it will, and I don't think Bostrom expects it to either.

Probability & randomness is an odd subject. Even the mathematicians & philosophers who specialize in it have a hard time coming up with a satisfactory definition of it. In this case, it seems Bostrom has stretched "almost certainly" to mean something unrecognizable from what we take it to mean.

Jairus - If the book offers clear rebuttals to my points above, please post a summary of those rebuttals. Better do it before lunchtime today, though...
posted by tdismukes at 6:59 AM on August 7, 2002

"[Einstein] told us once: 'Life is finite. Time is infinite. The probability that I am alive today is zero. In spite of this, I am now alive. Now how is that?' None of his students had an answer. After a pause, Einstein said, 'Well, after the fact, one should not ask for probabilities.'"

For good refutations of the Doomsday argument, read this article about randomness, this article about sampling assumptions, this article about reference classes, and this article about probability theory.
posted by monju_bosatsu at 7:00 AM on August 7, 2002

This is the problem I see with the argument, correct me if I'm wrong: "you should think that with 91% probability there are only ten people." I repeat: "should think." What if I choose to not think that? I'm infinite, baby! Really, it's a decent logical argument, but that's the gap I see. The probability may say that we are at a certain point on the doomsday clock, but probabilities are often wrong.
posted by The Michael The at 7:28 AM on August 7, 2002

did this remind anyone else of the lottery paradox? for some reason it left me with the same feeling of sketchy-premise unease.
posted by callicles at 7:47 AM on August 7, 2002

How do you put 11 horses into 10 stables, 1 horse a stable?

Put a horse aside for now, and put 1 horse in the first stable. Put horse #3 into stable #2, horse #4 into stable #3, horse #5 into stable #4. horse #6 into stable #5, horse #7 into stable #6, horse #8 into stable #7, horse#9 into stable #8, horse #10 into stable #9, which leaves a stable open for horse #11 put aside at the beginning . Voila, a child's play.
posted by semmi at 8:19 AM on August 7, 2002

Semmi - That one will only fool people who are really bad at counting and/or reading. Either stable#1 has horse#1 and horse#2 is out in the cold, or vice-versa.
posted by tdismukes at 8:24 AM on August 7, 2002

This theory is based on far too many assumptions for me to lose any sleep over it.

Exactly how much more probable will depend on the precise numbers you use. In the present example, the posterior probability of Doom Soon will be very close to one. You can with near certainty rule out Doom Late.

It sounds as if you could also choose numbers that would make DL and DS much more closely equal in probability; for example, if humans decrease their birth rate, increase their life span, and never get around to colonizing the galaxy, the total number of humans in the DL scenario will be much closer to the number of humans in the DS scenario, and their relative probabilities will change accordingly. Or perhaps DS could occur as a result of massive overpopulation, so having a lower rank order is evidence for DL. An interesting topic, though.

There are other people out there who are also pessimistic about our chances for long-term survival as a species.
posted by TedW at 8:39 AM on August 7, 2002

I think that one of the earlier links pointed out that the theory assumes that one's position in the total number of people in history is random. This seems to be a pretty major flaw that invalidates the entire theory. Since ones birth order in history is sequential and non-random, then many of the statistical conclusions would seem not to apply.
posted by KirkJobSluder at 9:20 AM on August 7, 2002

...Carter discovered a previously unnoticed consequence of a version of the weak anthropic principle.

Weak anthropic principles...they'll get you every time.
posted by jaronson at 9:23 AM on August 7, 2002

Agreeing with much of the criticism above, and also unwilling to quote Jeff Goldblum, allow me to restate it. Life is not a random cellular automaton, but a biological machine with a will, as well as biological imperatives to survive and reproduce. Extinction, on the other hand, is a random event, not some kind of light being switched off. Only a few possible scenarios would involve the full destruction of humanity, like the sun exploding. Even a major asteroid hitting earth would, assuming it destroyed functional civilization, be unlikely to result in extinction. We'd fight like hell to survive, wouldn't we? And even into the foreseeable future, any holocaust we devise for ourselves would be unlikely to result in extinction, as a practical matter, for much the same reason.

I don't see this thought experiment having much real-world application, no matter how well argued.
posted by dhartung at 9:41 AM on August 7, 2002

moss had it perfectly, and Hieronymous Coward hit it out of the park.

One roll of the die isn't nearly enough information to speculate about how many faces the die has.

And what if the die is round (i.e., it has infinite sides)? You'll just assume that for any N you see, that the number of sides is at maximum N.

And what if it's a plaything of the gods, and just comes up with an arbitrary number to see how shortsighted you turn out to be?

It's stuff like this that makes me think that Philosophy with a capital P is some huge cosmic joke perpetrated in order to busy people who really should know better with spouting mumbo-jumbo at each other so they don't use their intellects for something more constructive, like say helping us all to find a more sustainable way to live.

Oh, wait, that in itself could be a doomsday argument....

Premise: the field of Philosophy exists, and consists mostly of people going round in circles and wasting time, effort and energy spouting unprovable nonsense at each other, and that often otherwise brilliant people manage to be sucked into this vortex of meaninglessness.

Conclusion: it's all a nefarious plot by hyper-intelligent Outside Forces who are trying to bring about Doomsday (at least for humans).

I mean, come on. At least bong-induced nonsense wears off when you get sober again.

And to further dhartung's point: what do we know about the odds of extinction of sentient creatures capable of inventing and using complex technology (including amazing feats of medicine and off-planet travel), acting cooperatively on a huge scale, and of altering their own genetic makeup rapidly through non-reproductive means?

Answer: not one goddamn thing.

We are to the animals as the animals are to inert matter. There's a huge honkin' discontinuity between what we're capable of to save ourselves, and what they're capable of.

The future remains to be written. Don't throw it in the garbage can just yet.
posted by beth at 9:54 AM on August 7, 2002

Right on, dhartung.
posted by tweebiscuit at 9:56 AM on August 7, 2002

Why not ask the Church of the SubGenius about the probability of the world coming to an end? They've been wrong 5 times already, so they're about coming due. Probabilitywise.
Personally, I could really go for being rescued from a dying Earth by the Sex Goddesses from Planet X in their pleasure saucers.
posted by kablam at 10:06 AM on August 7, 2002

beth -- please, please don't judge the entire field of philosophy based on this guy's stupid argument. This is just a game of analytic logic, not serious work of any kind -- philosophy itself is a hell of a lot more useful, interesting, and valuable to society, trust me. It's about us understanding ourselves, our place in the universe, making sense of experience, of consciousness, of existence. God... I don't even know where to begin. I love it, and it's worth our energy.
posted by tweebiscuit at 10:06 AM on August 7, 2002

The idea is this: imagine you are going to pick a number from a lottery. There will be either 10 or 100 numbers in the lottery - you don't know which.But in either case, every number has an equal chance of coming up. Your number is 7. Knowing this, do you think it is more likely that there were 10 numbers in the lottery, or 100?

Setting aside the problem that we're not talking about a random sample here, there is a problem with this logic as well. If a random number generator spits out the number 7, I can make no inferences regarding the population size. In the lottery 7 is as likely to win as one million.

But again, a key problem with this hypothesis is that it makes an assumption dependent on random sampling and applies it to a situation in which the sample is non-random. To use an example from my biology days, in order to determine the growth curve of a certain microorganism, what you do is you inoculate a fairly large volume of growth media (about a liter) with a fairly small (about one million) number of microorganisms. The exact number of microorganisms did not matter all that much. You then sampled the growth media over certain time intervals (such as one hour) and counted the microorganisms. The results usually fell along a nice exponential growth curve before leveling out.

Now here is the key problem with the doomsday hypothesis. The samples are random (or at least pseudorandom) in regards to which individuals at each time frame were sampled, but not random in regards to the time in which they were sampled. The sampling at each hour captured individuals that were n-generations removed from the original inoculation. The number of individuals at each hour can be predicted to a reasonable degree of precision if you know the time since inoculation and the doubling rates of the organism. Since population size at any given point in time is a nonrandom process, probabilistic analogies to lottery balls are misleading.

However this opens the door to a completely different doomsday hypothesis. Homo sapiens is faced with a certain population crunch at some time in the future not because we have no way of knowing what the total population of the human species will be over time, but because we have a fairly good understanding of organism growth curves. The big question is how and when.


posted by KirkJobSluder at 10:30 AM on August 7, 2002

Well, lunchtime has passed, so I guess Bostrom was wrong...

Thanks to monju_bosatsu for some really good articles refuting the Doomsday theory with more rigor than any of us had bothered to go into. One of those links gave the equation for Bayes theorem, which confirmed my suspicion (see my #2 above) that the empirical probabilities one assigns to extinction events makes a huge difference in the end result one calculates.

One weakness of the Doomsday theory which all of the linked refutations went into was the fallacy of assuming that the observer has been randomly selected. Going back to crunchburger's lottery analogy: My original quick analysis was wrong. (I should have known better, but it's been too long since I studied statistics - sorry.) Our holding lottery ticket #7 implies actually implies nothing about the number of total tickets in the lottery. If we randomly selected tickets multiple times and came up with ticket #7 more than once, we could estimate some probabilities, but coming up with the ticket once shows nothing. Also, the ticket would have to be randomly assigned. If the tickets were handed out sequentially and you said "I want that one!" when you got to number 7, you're no longer random. Getting back to the Doomsday theory, if you could select an human observer from the set of all humans that ever lived or will ever live, you'd have random selection. In the absence of the ability to do this, you can't assume that you are not in "untypical circumstances." (See monju_bosatsu's quote above.)

Hmmm... it's sometimes fun poking holes in a logical fallacy, because it forces you to think with more rigor than usual. It's also easy to fall into error making the effort. Most of my original effort was correct, but I blew it on my point #3. (I think some of the other posters also made some logical errors in their refutations, but I'll let them figure it out on their own.)

On preview - KirkJobSluder got the point on non-random sampling. (I should have originally but I was to quick to post and didn't spend enough time thinking.)
posted by tdismukes at 10:43 AM on August 7, 2002

tweebiscuit: I should have been more specific in my criticism.

What I don't like (and don't Get and don't particularly want to) is what I call hifalutin' Philosophy, with a capital P, where obfuscatory language is used instead of everyday concepts, and I can see no real application to the real world. It all sounds like mumbo jumbo to me (and I find what little I know about it to be distasteful enough that I really don't feel like spending my limited time on earth learning more about it).

I'm talking about the kind of Philosophy that is taught in college courses (which I've admittedly never taken any of) and which is presented in such a way that it's completely incomprehensible to the average reasonably intelligent person (i.e., me). If there's so much value in it, why can't someone bother to translate it into a format that makes some sense?

Every time I've tried to read something of this ilk, my eyes glaze over, my head feels swimmy, and all sense of actual meaning existing on the page is gone by the time I get through just a few paragraphs. Very soon, the idea of banging my head against a wall sounds more appealing that continuing to (attempt to) read, so I give up.

Now, little p philosophy is something different; to me, it means general musing, discussion, and fruitful thought about the big questions of life. Why are we here, how should we live, how do we determine what is true, how should we treat other people, etc.

I like little p philosophy. One of my favorite books, Goatwalking, is something I'd call very philosophical indeed. I'm also a lay student of Zen, Sufism, Quakerism, Buddhism, mythology (a la Joseph Campbell), connections (a la James Burke), memetics (a la Richard Dawkins), mental illness (particularly what this reveals about the self and how human minds work), cognitive science, linguistics, game theory, and on and on.

I like stuff that is comprehensible, that challenges me to think more deeply, to rethink my assumptions. I think this helps me to grow into a richer and wiser person. I also feel like I understand more of how the world really works, and this deepens my appreciation and sense of beauty, day-to-day.

It's the Philosophical discussions I've seen about whether existence really exists, incomprehensible-isms, and the pretensions of using so-called logic on things which are beyond our powers to test the truth of, that really leaves a bad taste in my mouth.

I mean, people get paid for this? To teach it, to write about it, and especially to argue about it (endlessly, it seems). The arguments never seem to get anywhere, either. It seems to just go round and round, as though the participants really enjoy arguing for the sake of arguing. Count me out - I see argument as a means to an end, that of discernment of which ideas should be kept, and which reconsidered or discarded.

I just don't get it - it smells like a scam to me. Perhaps I just haven't read the right books (but I am also wary of wasting my time when there are so many books I *know* I really want to read and will enjoy and benefit from).

I don't want to be mean. Perhaps I should just say that it's really not my cup of tea, and leave it at that.
posted by beth at 11:04 AM on August 7, 2002

beth - for an example of a decently written, clearly thought-out Philosophy article, check out the link from monju_bosatsu above regarding sampling assumptions.

I haven't spent enough time surveying the realms of academic philosophy to know what the ratio of useful stuff to crap is. (I had undergrad intro to philosophy plus randomly assorted readings since then.) The problem I see with certain brands of philosophy (academic and otherwise) is when they replace "philosophy" meaning "love of knowledge" with "philosophy" meaning "love of whatever I can come up with in my own head, ignoring the outside world." The problem with that approach is that the reality of the outside world is way more inventive than the most brilliant human that ever lived. If you don't check your conclusions against reality, you're wasting your time. If you claim to love knowledge, but don't pay close attention to what people studying the outside world have discovered, you're wasting your time. Personally, I think any philosopher concerned with the traditional domains of philosophy should be keeping a close eye on findings in cosmology, evolutionary biology, cognitive neuroscience and AI.
posted by tdismukes at 11:26 AM on August 7, 2002

I couldn't agree with you more, tdismukes. In that vein, see also this.
posted by monju_bosatsu at 11:33 AM on August 7, 2002

Somewhere I read a joke saying that cosmologists only need a pencil, paper, and a large waste basket. Philosophers can do without the waste basket.

Perhaps I'm a little bit skeptical of the claim that philosophy is disconnected from empirical study. At least in my field philosophy plays a pretty big implicit role in deciding what makes a good answer to a question, what makes a good question, and what are our political and ethical obligations along the way.
posted by KirkJobSluder at 11:38 AM on August 7, 2002

tdismukes: Our holding lottery ticket #7 implies actually implies nothing about the number of total tickets in the lottery

Thats not true. Bostrom uses Bayesian statistics correctly. Its only his assumptions that are flawed.

Here's a better example: You are about to draw balls from an urn which contains either 10 or one trillion balls. All the balls are white except one which is red.

The first ball drawn is white. And the second... The seventh ball is red! Bayesian analysis does tell you that, yes, there is more likely 10 balls in that urn than one trillion.

The flaw in your reasoning is that lottery ticket #7 is in no way 'special'. Lets make it special: #7 is the winning ticket. And you have been told that this 'lottery' is either a church raffle or a national lottery. I dont know whether you are a betting man, but I'd bet that you hold the winning ticket to a church raffle.
posted by vacapinta at 12:16 PM on August 7, 2002

tdismukes: your original assertion was right. What made you change your mind?
posted by vacapinta at 12:27 PM on August 7, 2002

vacapinta - Juggling too many thoughts at once got me confused. You're right, if the ticket is randomly selected and we have no other information about the relative likelihood of a lottery with 10 vs 100 numbers, then holding ticket number seven does give us information about the probability of which lottery it is. When I posted my follow-up comment I originally meant to just make the point about the ticket not being randomly selected. (Examining the Doomsday theory from the standpoint of a human living now is equivalent to saying "I'll take that one" when the ticket 7 is handed out in sequential order.)

So I was right (if not completely thorough) in my initial statements. Trying to add too much information on top of that got me tripping over my own mental shoelaces. Sigh...the story of my life.
posted by tdismukes at 12:48 PM on August 7, 2002

What suprises me the most about this thread is how indignant everyone is.

I think the theory makes a lot of sense. If we had no other way to predict whether the world will end soon, this would be a good one.

However, we do have ways of predicting such things, based on the current state of affairs and so on, so it's probably best to lean more on those instead.

The problem with any doomsday theory is that if it's correct, everybody dies and nobody's left to care. Perhaps in most alternate universes, cavemen DID die out. The only data we have to go on is our own past, and since it HAS TO contain our survival for us to be here worrying about it, it's not very usefull.
posted by Flimsy_Parkins at 1:19 PM on August 7, 2002

Here's a better example: You are about to draw balls from an urn which contains either 10 or one trillion balls. All the balls are white except one which is red.

The first ball drawn is white. And the second... The seventh ball is red! Bayesian analysis does tell you that, yes, there is more likely 10 balls in that urn than one trillion.

True, however this appears to be different from the argument presented by the doomsday hypothesis that because one just happens to be the seventh ball to come out of the urn, one can make an inference about the total number of balls within the urn.

The flaw in your reasoning is that lottery ticket #7 is in no way 'special'. Lets make it special: #7 is the winning ticket. And you have been told that this 'lottery' is either a church raffle or a national lottery. I dont know whether you are a betting man, but I'd bet that you hold the winning ticket to a church raffle.


The big question here is how is that number seven assigned? In the case of the doomsday argument we are asked to believe that number seven is special simply because it is the seventh in the sequence of balls pulled from the urn. I have serious doubts that you can establish the total number of balls of the urn by virtue of the fact that you have pulled out seven of them.

Again, both of these examples depend on random sampling (or at least in the case of the lottery ticket random assignment) which fail in the case of the doomsday argument because assignment to cubicles, ticket numbers, or sequence out of the urn is assigned in temporal order.


It is true that simply holding ticket No. 7 offers no information about the total number of tickets in the lottery, provided that ticket 7 is not in any other way different from tickets 77, 777, or 7777.

I think the theory makes a lot of sense. If we had no other way to predict whether the world will end soon, this would be a good one.

The problem with this theory is that it always predicts that the end of the world is around the next corner. The same reasoning can be just as easily be applied to the number of seconds since the universe was created as human lives. To use an analogy John the Goatherd decides to move from the city to the country and start a dairy goat farm. To start with he has five goats. We are forced to assume according to the doomsday hypothesis that disaster is just around the corner. The next season, he has nine goats and again we are forced to assume that because there are nine goats in his pen that the total number of goats in his history is only slightly larger than nine.
posted by KirkJobSluder at 1:35 PM on August 7, 2002

bayesian probability is weird, it's also called subjective probability. it's really a very different type than what i tend to think of as "real" probability. it takes into account things you know, or things that happen. it's good for some things, but as one of monju's links above says maybe not the Doomsday hypothesis.

for the red ball as ball #7, normal probability would tell you that is either an unlikly event [1 trillion balls], or a sorta likely one [10 balls], but it couldn't tell you at all whether the bottle has 10 or 1 trillion balls. you couldn't know with normal probability, because neither are impossible. bayesian says it probably has 10 because the probability of such an improbable occurance happening is low. sort of like second-order probability.

now if you just have a bottle with an unknown number of balls, and 1 red and you start pulling them out, you do get a sort of temporal ordering. when you pull out the red one bayesian prob will tell you that you don't have much longer to go. this is because it would be very improbable to pull out the only red ball out of 1 trillion first.

this whole doomsday thing though seems wacky, i see almost no resemblance between the cubicles example and/or the bottle example and human life on earth. what is the red ball? me. why am i special? why not joe, or jim, or caveman ook? also in the real world unlike both examples the time to doomsday may not be finite. there may be no doomsday. if there were an infinite number of white balls that would ruin the bottle experiment...

comsmology+bayesian theory+philosophy+weak analogy?empirical prediction
posted by rhyax at 2:27 PM on August 7, 2002