Re: The great filter

N.Bostrom@lse.ac.uk
Tue, 13 Aug 96 19:00:04 GMT

I recommend every transhumanist to read Robin Hansons short
document at http://hss.caltech.edu/~hanson/greatfilter.html.
It gives a clear presentation of an argument that should be
taken very seriously.

(First a minor comment:
>But if single-cell life started in some distant molecular
>cloud and spread here via a wider panspermia, then that
>does help, in proportion to the
>volume of space between here and there. The chance of life
>starting in any one small volume
>can be pretty low and still be consistent with that data.

But on the other hand, if the distance were very great, then
it would be unlikely that the "molecular cloud" would not
have infected many other planets in its surroundings, some
of which would have evolved life that expanded into the
universe and which we should have discovered. Wouldn't it be
more reassuring if Earth and Mars had been infected by a
_local_ panspermia (from another planet in our solar system,
or from a nearby solar system)?)

If life developed independently on Earth and on Mars, what
could block the conclusion that our far future is probably
doomed? I can see only three potential answers:

1. As you suggest, there could be a great filter at some
later stage in the evolution of high intelligence.
2. Higher life forms continue to prosper but do not cause an
"explosion" into cosmos. (I think that Michael Wiik favoured
this alternative.)
3. Higher life forms do explode into cosmos, but in ways
that are invisible to us. This would presumably mean that
they do not engage in galactic scale constructions, and that
they are not interested in contacting human level life.

As for (1), it is dubious that such a filter would be
sufficiently effective to bring down the probability from
the value we would give it knowing that life evolved
independently on two planets in our solar system (say p=2/9)
down to the value that is equal the inverse of the number of
planets less than a few million light years away from us.
However, our present biological knowledge is not sufficient
to settle this issue, so it leaves some room for hope.

Alternative (2) might be more likely. Why should any
intelligent being explore the details of cold and monotonous
space if they can live in a much richer VR? And why should
they care about reality at all, whether real reality or
virtual reality, if they can achieve satisfaction from drugs
and electrical stimulation? If this is the case, they might
deem that no or only relatively minor space missions are
sufficient. I don't know of any conclusive psychological,
cybernetic or political arguments that this will that it
won't happen. Also there is the possibility that the
relevant technologies for extensive space colonisation
simply aren't feasible.

Alternative (3) does not seem extremely improbable either.
Humanoid civilisations might not be at all interesting to
posthumans. And it is hard to tell what sort of
manipulations of the universe a posthuman culture would find
suitable. Also, there could perhaps be ethical/strategic
reasons for not interfering to much with the natural order
in cosmos.

It should be noted that only alternative (1) is compatible
with typical transhumanist accounts of the far future of
mankind.

Then, of course, there is the other possibility: that
intelligent life almost always put an end to its own
existence once it has reached a certain level of
sophistication. It would be interesting to list the ways in
which this could happen.


I don't know whether it would be significant, but mankind
has been sending out signals into space for some time now.
Even if humanoid life tended to annihilate itself 50 or 100
years after it began to search for extraterrestrial life,
there would still be a chance that the civilisations on two
planets could overlap so that one would find out about the
other. If we have been signalling into space for some time
t, and the average time a species is actively searching for
life in the universe is T, then the probability that we
should have had an overlap with a given hi-tech life on a
planet would be P(overlap)=(t+T)/5*10^9, where we have
estimated the period in which such life is likely to occur
to 5*10^9. If we multiply this with the number of planets n
such that we would have discovered a signal from them had it
been sent, and with r, the average number of times hi-tech
life occurs on a planet given that it occurs there at all,
and with the likelihood P(hi-tech evolves) that high-tech
life evolves on a given planet, we get an estimate of the a
priori probability that we should have discovered
extraterrestrial life

P(high-tech found)=P(hi-tech evolves)*n*r*(t+T)/5*10^9

r is presumably a small number (say 1.8), but if n were very
great, then, since we have reason to believe that P(high
tech found) is not much bigger that 0.5, we would have to
conclude that P(hi-tech evolves) is small. If this were the
case, then that would be reassuring, for it would indicate
that hi-tech life is rare, so that the fact that we have not
seen traces of any life explosion need not mean that there
has to be a great filter. However, I doubt that n is great
enough today. Does anybody know the relevant facts about the
search for extraterrestrial life to make a reasonable
estimate of n?

Niklas Bostrom n.bostrom@lse.ac.uk