Re: Recursive self-improvement curves

From: Mitchell Porter (
Date: Thu Jan 08 2004 - 22:16:22 MST


>Unless you are speaking of more than weak anthropic principles at work I
>fail to take your point. What does this have to do with relatively weak
>mechanisms of self-improvement leading to us being here versus relatively
>strong means of self-improvment we can easily envision? I don't see what
>a supposed anthropic element has to do with it.

Well, the question (as raised by Paul Fidika) was, how do we
know that recursive self-enhancement is an accelerating process?
You could just as well argue that it slows down, because each
new improvement is harder to discover. Eliezer gave examples
from history, which suggest an accelerating tendency stretching
across biological, cultural and technological evolution. My idea was
indeed that this might all just be an artefact of the weak anthropic
principle: Any sentient being, when it investigates its origins, has
to find a world that went through all those transitions; but sentience
having been attained, WAP says nothing about further evolutionary
leaps being likely. We may be wrong to generalize about the
powers of evolution, on the basis of how it unfolded here on Earth.

The paper by Robin Hanson on this subject ("Must Early Life
Be Easy?") doesn't suggest any way that *acceleration* would
show up as an artefact, however. If there are hard and unlikely
steps on the path to sentience, they should be more or less
evenly distributed throughout the history leading up to it - that's
what his model says. A more complicated model might make a
different prediction, but then we are getting into Drake-equation
territory, making guesses about quantities whose actual values
we have no idea about.

E-mail just got a whole lot better. New ninemsn Premium. Click here

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:43 MDT