From: Robert J. Bradbury (bradbury@www.aeiveos.com)
Date: Thu Dec 23 1999 - 09:39:19 MST
On Wed, 22 Dec 1999, Eliezer S. Yudkowsky wrote:
> I wrote:
> >
> > The only solution that I can see is some kind of evolved value
> > (i.e. created self-perceived rewards) for either *no* reproduction
> > or *consenting* reproduction with an agreement to sacrifice or
> > invest resources. Dealing with the "joy of rape" is going to be
> > with us for a very long time.
>
> Absurd. Selfishness is no more inherent to minds-in-general than the
> joy of consenting reproduction. If we don't build in the motivation to
> rape, we don't need to build in countermotivation.
Eliezer, that assumes we are building in the motivations and not
starting with soft-squishy people or uploads derived directly from
them.
Bryan has got the issue exactly correct in that he is questioning
why something not ok in reality, *is* ok in VR. This goes back
to my entire discussion about when it is ok for an SI to pull the
plug on a sub-SI (after all the instantiation of *everything*
going on within the SI is "virtual").
Lets look at it another way using "lions", instead of "humans"
to make it a little less repulsive. An older male lion gets
driven from a pride by a younger male lion who then proceeds
to kill all of the young cubs fathered by the older male lion
to bring the females into estrus sooner. Simple, hard, cold,
brutal and *entirely* "rational" operating procedures for "nature"
trying to select the best genes.
Now, I tend to put this in the same class as "rape", because in
both circumstances you are trying to utilize the reproductive
resources of others for personal benefit. When I think in terms
of human "rape", I mean it more in the sense of classical rape
conducted in association with "pillaging" by tribes or armies.
(Keeping in mind that the anthropologists say the only two things
that induce violence in "primitive" humans is lack of food and
lack of women). Much modern "rape" seems to be of a different
nature involving issues of "control" (but one can extend the
anthropologists perspective by seeing that what is involved
is the "control" of the resources (food & gestational apparatus)
necessary for reproduction), so the "roots" of modern and classical
rape are probably the same -- and not dissimilar from that of the male lion.
So, infanticide and rape have their roots in nature's process of
natural selection, i.e. they are most likely genetic. The tendency
would be for the "infanticide" or "rape" genes to become stronger until
they are offset by those situations where a young lion tries to take on an
old lion too soon and gets killed or a young man rapes the sister of the
strongest warrior or chief's wife. In practice in humans, we probably
have some "need for social acceptance" and "cooperation" genes that work
against the violence/infanticide/rape genes.
Genes that have varying utility for reproductive success *will* be
polymorphic. You can see this if you have for example:
S - strong sex drive (w/rape tendencies), s - weak sex drive
C - strong cooperation drive, c - weak cooperation drive
Some examples of what you get in the population:
SS, cc - rapes too much, is asocial and gets driven off or killed,
but might have reproduced before that occurs
Ss, cc - sometimes rapes and but due to low cooperation probably
has < average reproduction
Ss, Cc - rarely rapes, generally cooperative, average reproduction
ss, CC - never rapes, highly cooperative, possibly < average reproduction
SS, CC - rapes only in asocial circumstances (i.e. classical rape)
so has slightly > average reproduction
[I'm sure the hard core geneticists or sociologists might quibble
on the likelyhood of success of various combinations -- this is for
illustration purposes.]
Given the differences in success these combinations produce and the
fact that humans have probably dozens to hundreds of variations for
the genes that are involved, there will be a wide range of behaviors
in the population. You can see that humans have social conventions and
laws to restrict people with some of the "bad" combinations from "stealing"
resources that belong to others.
The people who have these "bad" combinations may very well have
simply moved what they would like to do in reality from the
external world to their internal world. They don't "do"
rape or child molestation or infanticide, but they may
think about it. This seems to be the essence of Bryan's point --
you haven't really fixed the source of the problem, you have simply
forced it underground.
Now Eliezer solves it by getting rid of "S" entirely or perhaps
building in strong (nonremovable) constraints on the expression
of "S" (if you make "S" too weak, you might as well "burn" your
your computronium, because someone with a stronger "S" will figure
out how to take it from you). But that *doesn't* work with real
people.
What does work with real people is genotyping them and "mandatory"
gene therapy so the SS/cc combinations get turned into ss/CC.
If you don't like mandatory gene therapy (I can't imagine many
people on the list would), then you have to live with the fact
that sooner or later people *are* likely to encounter a situation
where the "extremes" of the population slip over the edge of their
internal self-restraint mechanisms and turn their internal reality
into an external reality.
Just as if you live long enough, you will suffer a fatal accident,
if you live long enough, you are likely to be raped, or perhaps
die trying to defend someone from being raped, or perhaps even
commit a rape yourself. [I'm generalizing a lot here because
it depends on the "rate" of these events and that for many individuals
may be very very low.]
So, I'm forced to conclude that if you *really* want to diminish
the probability of these things occuring to you over a 50,000
year lifespan (assuming you choose to remain soft-n-squishy)
that you must consider mandatory genome correction. Another
possibility might be enclaves of known "types". Feeling bored
after 2000 years in "Safe-city", take a 2-week, all-expenses
paid vacation to "Rape-city"... A third possibility might be
mandatory risk "broadcasting". However if the reactions to
child-molesters being released back into society are any example,
we can expect a lot of vilification of these "potential" offenders
simply because they got stuck with some bad genes.
Robert
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:06:11 MST