Re: Objection to the Doomsday Argument?

Hal Finney (hal@rain.org)
Mon, 17 Aug 1998 08:40:39 -0700

Doug Bailey, <Doug.Bailey@ey.com>, writes:
> What if we apply the DA to other classes of objects? If we apply
> the DA to stars, shouldn't conclude such stars are not atypical
> and thus conclude Doom Soon for stars as well? Does the DA give us
> any insight into the probability that Doom Soon for stars is any
> higher than before we considered the DA? I can't see how.

Reasoning analogous to the DA would suggest that when you look at a random star, you probably see it about halfway through its lifetime. I think this is exactly in accord with observations.

You don't get "doom soon" because there is no exponential growth phenomenon involved.

> If we apply the Bayesian analysis used by the DA to other classes
> of objects we get some fairly alarming results. What if we apply
> the DA to the number of scientific discoveries that have been
> made? Are we to conclude that virtually all of the discoveries
> that we have been made have already been made? To conclude
> otherwise would be to violate the "typical" assumption that's
> embedded in the current analysis of the DA.

If we look at "science" as a phenomenon which has been going on for a few hundred years, the DA suggests that it will probably continue for a few tens to a few thousand years.

> The only difference I see is that the "stars" and "scientific
> discoveries" examples have a third-person perspective (I'm
> considering them) whereas when I consider the DA in the context
> of humanity its a first-person perspective (I'm looking at my own
> relative position). I'm not sure how this might be a problem. I
> can see how the first-person perspective is significant to the DA
> but I don't see how our assessment of ourselves would be qualitatively
> different (for purposes of applying the DA) than our assessment of
> scientific discoveries or stars.

The difference is that the growth of "scientific discoveries" does not by itself give you a greater probability of observing it when there are more of them than when there are fewer. (To the extent that it does, improved technology causing larger population, then the regular DA argument does kick in.)

Consider something like a growing fad, say the Tamagotchi toys. By the time you see someone playing with one, chances are the fad is near the end of its lifetime. It may have grown quietly for years and you never heard of it. Your chances of seeing someone with one are greater, the more people have them. By the time you see one the fad has neared its peak. In fact, Tamagotchis are passe now, I am told by my kids.

This may also help explain the well known phenomenon where the average investor complains that an investment with a good track record turns bad as soon as he puts money into it. An investment may be doing well for a while without many people knowing about it, but over time information about it spreads and more people get into it. By the time the average person gets in the opportunity is gone. (Pyramid schemes are a perfect example of this.)

> The conclusions I am reaching using the DA is that I should expect
> everything (that has a bounded existence) to become "extinct" and
> other things (such as scientific discoveries) not to increase too
> much in size between now and forever. The DA appears to place the
> entire universe (at every conceivable level) on the execution block
> or in deep freeze. Shouldn't such a set of conclusions raise some
> concern about the viability of the DA as a reliable guide?

You have to look at history to see whether the DA has been accurate. Some phenomena, like life, have continued forever. Others, like species, have mostly gone extinct. Historically, most things have finite lifetimes. That's one principle on which the DA is based.

Hal