Re: Sysop vs. Liberty (was Re: How To Live In A Simulation)

From: James Higgins (
Date: Fri Mar 16 2001 - 09:40:44 MST

At 05:58 PM 3/15/2001 -0500, you wrote:
>"Eliezer S. Yudkowsky" wrote:
> >
> > Gordon, we discussed this in the "Beyond evolution" thread, and we heard
> > your own opinions in the "Learning to be evil" thread. I don't think we
> > need to reopen the discussion at this point.
> >
> >
> >
>(Of course, it's kind of ironic that this post comes in a thread entitled
>"Sysop vs. Liberty". It's just that certain other mailing lists I know
>have a pattern wherein the same discussions are just repeated over and
>over. I think the solution is Archive Awareness; knowing where all the
>best of the previous discussions are, and being able to reference them; if
>anything new does come along, it can build on the previous discussion.
>Maybe someday we'll even have an SL4 FAQ, with archive references.)

Yeah, and someday (post SI) we'll have background processes to go wade
though all those archives automatically and highlight all the interesting
bits. But for now I only have a single stream of consciousness and it is
very, very busy.

I have a simple related question. How is it at all possible that this
Sysop API will know what you should & shouldn't do? Two main points:

1. Assuming that the Sysop rules are not hard-coded and monolithic. In
order to decide what you can do, the Sysop API would have to be
significantly smarter than you are in order to "judge" your actions. How
likely is that?

2. Assuming they are hard-coded and simple. It is very easy to get around
virtually any rule given a reasonably amount of thinking (look at today's
laws). Also, it is sometimes necessary to break the rules for the greater
good. This is the same thing that scares me with today's fly-by-wire jumbo
jets. The pilot can not exceed certain pre-set angles of turn or
ascent/descent. What if there is an emergency and they will die if they do
not? Answer, they die. I don't like that at all...

James Higgins

P.S. Hi everyone! I'll get to writing an introduction RSN (hopefully). ;)

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT