Re: Transhumanist Kids

Anders Sandberg (asa@nada.kth.se)
10 Dec 1998 17:35:09 +0100

Michael Lorrey <retroman@together.net> writes:

> I would love to reveiw your model.

Youre welcome, I can send you a draft of my paper.

> Have you tested a model where you test MIR only
> individuals, Co-op only individuals, and MIR/Coop capable individuals? I bet that
> Individuals who can do both MIR and Co-op, depending on the situation and the
> behavior of the others will fare best.

MIR doesn't do well against itself, since around half of the time the agent meets somebody who is bigger than himself; the total payoff is just 0.5 - and half of the agents does much less. CWC works fine with itself, each agent makes 1.0 units of payoff per time unit (one interaction initiated by the agent, another initiated by another agent). Mixed strategies that randomly mix MIR and CWC evolve so that they in general are cooperating, there is no particular bias to cheat more or less.

I'm running a version of the prisoner's dilemma with coercion now, and the dynamics is much more complex. It looks like speciation occurs, and there are strange shifts in strategy I haven't understood yet.

> > The problem is that humans in general aren't as rational as they could
> > be, and MIR is easy to explain (just use a gun) while coperative
> > strategies require more thinking, communication and education. Which
> > is why I think we should introduce young people more to game theory,
> > the prisoners' dilemma and the theory of cooperation.
> >
>
> Yes, I beleive that youngsters should be exposed to much more of this.

Maybe one could make some interesting toys from this?

> However I
> wouldn't leave out the MIR principle in its entirety. As 20th century history
> shows, MIR individuals are usually only best dealt with in an MIR manner.

Something I haven't yet tested is group coercion, where groups can go together and coerce others. This seems to be the most likely way of dealing with MIR in this manner (it is hard to find heroes on white steeds these days to overthrow dictators or Bill Gates - essentially a bigger MIR against the current MIR), but another way might be to simply ignore them if they do not do too much damage - other agents can institute a "MIR insurance" and let the rare MIRs coerce; as long as the cooperators can make the spread of more MIR unlikely (e.g. by make being cooperator more profitable than being a MIR) this would work well.

-- 
-----------------------------------------------------------------------
Anders Sandberg                                      Towards Ascension!
asa@nada.kth.se                            http://www.nada.kth.se/~asa/
GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y