Re: META: To my Colleague Extropians

From: Anders Sandberg (asa@nada.kth.se)
Date: Thu May 03 2001 - 01:18:42 MDT


On Wed, May 02, 2001 at 02:18:06PM -0700, Spike Jones wrote:
> > Natasha Vita-More wrote:
>
> > I'd rather see a system implemented that follows more mind-bending criteria
> > than whether a person is popular or not-:) If a rating system relied upon
> > the Extropian principles and a variety of essays that embody content
> > reflecting the goals, thoughts and feelings of "Extropy" rather than the
> > personal likes and dislikes...
>
> This is a waaaay better idea, thanks Natasha!

Yes, it definitely is. The only problem is implementing it - that is far
trickier, nearly AI-complete.

> How would we do it so that we rate the posts and not the posters?
> Then create a matrix such as was described previously? Then
> have the computer reconnect the post ratings with the person,
> so as to make an ERA to allow Anders' filtered list of faves?
> We could eventually write a subroutine that would keep track
> of our favorite and most extropic posters and threads. We
> would tell the machine we have half an hour to read email, for
> instance, and it would automatically delete the junk and give
> you half an hour's worth of stuff from your family, friends and
> favorite extropians, who got that way from writing cool stuff,
> not just from being nice people offlist. spike

Using people as extropy-detectors is of course the best way today. One
way of detecting that at thread is interesting is that many extropic
people participate, so the thread could be weighted by the sum of the
ratings of the participating people, with some scaling for the length of
the thread. Unfortunately it is hard for humans today to distinguish
between a poster and his posts; even when judging the latter judgements
about the first creep in.

There seems to be some instability in such a ratings system: first, a
new thread will just have the score of the originator, so it might be
hard to detect even if it is good. Also, high ranked people become more
visible and have a greater chance of being highly ranked. Here we need
some form of "conscience" in the system that balances the concentration
of ratings.

What we must avoid is having a lot of parameters to tweak. In my
opinion, the best systems are always those that have no free parameters
because they adapt well on their own.

-- 
-----------------------------------------------------------------------
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


This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:07:27 MST