Re: Promoting High Bandwidth AGI Discussion [WAS: Re: Not the only way to build an AI]

From: Ben Goertzel (
Date: Wed Dec 14 2005 - 11:01:50 MST


This is an interesting suggestion, and it basically sounds like a
special kind of subject-specific Wikipedia, doesn't it?

I guess that what you describe could fairly easily be implemented
using Wiki technology.

I don't think your suggestion fully addresses the concerns Jef was
raising, but I think it would be a valuable contribution.

Basically, I think that email lists and websites are good, but that
there would also be value in a variety of other
communication/interaction media, and the one you describe could be a
useful part of the mix.

-- Ben G

> What about a web site organized around "evolving essays", of the
> following sort:
> 1) Someone starts with an encyclopaedia-like, hierarchical topic list,
> which is supposed to be the set of categories within which essays (which
> will end up being encyclopaedia entries) are stored.
> 2) This framework is debated, but the debate takes the form of
> restatements of the topic list, which are then voted on in some way. So
> for example, if I think that a branch of the topic list headed by the
> topic "AGI Goal Systems" should be split into
> AGI Goal Systems
> Decision-Theory-Based
> Motivational-System-Based
> ... then it is up to me to suggest this, and to provide two headline
> paragraphs (no more than one para for each subtopic) that describe the
> two subtopics.
> 3) If an aelaboration of the topic tree is sanctioned, it first comes in
> as a "provisional" branch, and it is then up to someone to suggest an
> essay that describes the topic.
> 4) People give feedback to the poster of the essay, and the feedback
> goes privately (no back and forth argument at this stage) to the poster.
> This person is supposed to incorporate the suggestions into the essay
> if possible.
> 5) When the essay writer has done editing the essay it is resubmitted to
> the list, and all, including the original reply folks, are invited to
> say whether this is coherent enough to stand. (Coherent according to
> various criteria TBD).
> 6) If someone strongly disagrees with the coherence of the essay, they
> must say so at this point, but there should be a fairly high threshold
> for rejection (you cannot simply disagree with it, you have to posit
> incoherence).
> 7) Dissenting opinions are also allowed. If an individual or a group
> believes that a particular entry is incomplete or wrong or bad in some
> way, it can write a coherent reply in the form of an essay, and send it
> through the same process.
> What I have in mind through this process is the collaborative writing of
> an extremely hierarchical book.
> Another note: I think that we should make some allowance for background
> knowledge, but I am not sure how to do this. For example, if someone
> well studied in philosophy makes comments about an issue that is, say,
> about hardware design, there ought to some way to give those comments
> less weight than if they come from a software or hardware engineer. And
> vice-versa of course. This is a little difficult, as it might imply
> people voluntarily submitting their claimed areas of expertise, which
> might not go over too well.
> Respectfully,
> Richard Loosemore

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:54 MDT