Re: poly: ESS for HPLD

From: Anders Sandberg <>
Date: Mon Dec 08 1997 - 10:01:17 PST

Hal Finney <> writes about the interior of the growing

> However on reconsideration it is less clear what the units of replication
> should be. With the levels of control which should be possible there may
> not be any analog to individuals. An over-simplified model suggests that
> all matter will be engaged in computation and associated tasks, with the
> thinking being done in virtual form.
> It might be better to look at it as competition among memes than among
> organisms. Combinations of memes, strategies if you will, control how
> matter is manipulated to implement whatever goals are determined by
> the meme sets. (A familiar problem is how to rationally adopt goals when
> you have direct control over your own motivations. I will just assume
> for now that goals are memes.)

I think you can model selfish replicators in general, since it appears
likely they will be the dominant systems (a side-thread: is there any
likeliehood that any other kind of replicator, like say an uniform
parallel AI program, will become dominant and ESS?).

When the genes appeared, they in time evolved phenotypes such as
bodies and eventually nervous systems in order to gain fitness
advantages. The properties of nervous systems were such that selfish
replicators (memes) could evolve in populations of nervous systems,
and the memes appear to be developing systems such as technology and
information processing systems to further their own fitness.

This suggests to me that new selfish replicators can or have appeared
in these systems; a simple example may be computer viruses and
computer-meme viruses like "Good Times", but more advanced entities
can doubless exist in the infrastructure for meme transmission,
protection and development we have created. Maybe there will shortly
be a new takeover, similar to the "Kemes" mentioned in Gregory
Benford's _Sailing Bright Eternity_.

This process could quite likely continue indefinitely, although the
levels appear less and less clear. So it appears to me that there
would be a complex web of replicators co-evolving with each other on
many levels. Maybe it would be useful to look more at how ecosystems
interact rather than simple replicators. Ecological succession might
be the future of war.

> It is questionable whether an HPLD can really exist. It could be that
> the laws of physics are complex enough that it is intractable to discover
> all useful technologies in a reasonable period of time, or it could even
> be that the laws themselves are infinitely complex and new technological
> discoveries can never be ruled out.

This is an open question. If the assumption of a simple basic
structure is true, then HPLD is a possibility. Otherwise we might get
strong divergence, where many civilizations are utterly incompatible
with each other. That would intensify the differences but also make
many "ecological niches" possible; a civilization based on
manipulation of quantum noise could co-exist with a civilization based
on (say) philosophical engineering of the platonic forms - neither
would have anything useful for the other.

> Ignoring technological advancements, it seems likely that space will
> crystalize into one or more regions, each with its own meme set which
> represents a local maximum in terms of expansion/defensive capability.
> It may be that one meme set is dominant, in which case this region will
> expand to include all space. Or it may be that there is more than one
> local maximum, each close enough that they are not able to conquer the
> other, and so there are boundaries between these regions.

There is no reasons why not several layers of conflict could exist;
while memes A and B are mutually incompatible and have fenced of the
universe into two regions, memes C,D and E can exist regardless of A
and B, and themselves compete. To make matters even trickier, there
may be couplings between the levels, creating a complex mixture. This
state does not need to be stable, but would likely be self-organized
critical. think this is worth studying closer.

Anders Sandberg                                      Towards Ascension!                  
GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
Received on Mon Dec 8 17:52:59 1997

This archive was generated by hypermail 2.1.8 : Tue Mar 07 2006 - 14:45:29 PST