Re: anonymous collaborative filtering

From: Simon McClenahan (SMcClenahan@ATTBI.com)
Date: Mon Sep 23 2002 - 08:48:03 MDT


----- Original Message -----
From: "Cliff Stabbert" <cps46@earthlink.net>

> Here's a reformulation: a new discussion board-type technology where
> the anonymous contributions become, through some form of collaberative
> filtering/reordering/reformulation, a _single train of thought_.

Your wording doesn't make sense to me. A collaborative technique or action
has the end result of a single train of thought. And if it human
collaboration that is employing the technique, the only benefit from using a
particular technology is simply as a tool for the individual humans to enter
and filter/reorder/reformulate the stored information. If the collaboration
is performed by an AI, then the discussion board technology simply becomes
an issue of human user interface design on how to enter and render
information.

> My main thrust isn't the anonymity itself, but producing those single
> threads of thought collectively. Something somewhat akin, perhaps, to
> how two authors can collaborate on a piece of fiction -- refining and
> hammering each paragraph, each chapter, the overall structure until
> neither can even clearly point to "which bits" they wrote. But with
> more people.

In programming, merging two pieces of source code can be done automatically
without any intelligence in a lot of cases. Intelligence is needed to make a
decision when there are overlapping differences, rather than simple addition
or deletions. If the source code is well organized due to good software
architecture, most text merges will be of a simple nature.

In a collaborative system, we are using ideas instead of text to merge
together. But unlike linear text that is pre-designed, tested and
scrutinized before attempting to merge, ideas are very non-linear and the
resulting natural language text is very complex, so that the only simple
automated merges are additions and deletions.

I myself have had visions of an email client that could manage threads of
conversation at the sub-message level. I despise traditional methods of
quoting and re-quoting, line-wrapping, inconsistent markup that various
individuals prefer. The next step from having a dumb system that can follow
threads of conversation where the textual merging consists of additions, is
to do what you are asking and use Natural Language Processing to consolidate
the linear thread of a conversation, removing redundancies, and making
intelligent decisions between two texts that represent the same idea.

Your authoring example also has the requirement that everyone is speaking
the same language. Ideas should be parsed and merged correctly, regardless
of their natural language text encoding.

> I imagine anonymity as helpful in eliminating ego
> effects; in the right context with the right people it may not be
> absolutely necessary but I'd still opt for it because ego has a way of
> sneaking up on you.

This is a social problem. You need to figure out the idea parsing first.
Personally, I prefer people to be identifiable and accountable for what they
say, egotistically or not.

> Two things that come to mind as threads to explore: mind mapping
> software (googleable; I've seen better examples than at
> http://www.visual-mind.com/introd.htm but it'll do for a start),

The trademarked Mind Maps were developed by Tony Buzan, and is designed very
specifically for humans to work and think in a more visual mode.
http://members.ozemail.com.au/~caveman/Creative/Mindmap/index.html
Creating a discussion board that uses graphical artifacts will be difficult
to sell to people because they are still used to thinking linearly, in a
"train of thought" as you put it. The introduction of HTTP hyperlinks in
email is commonplace because it is encoded as simple text; you just type it
in, no buttons to press or commands to remember. Not very many people
re-read what they just wrote in an email, let alone grammar-check or simply
spell-check it!

> the "glass bead game" (also googleable, see e.g.
> http://www.corewave.com/core/ and check the tour and main map).

It "looks" interesting because of the visual component of the map. I will
read more of this site when I have time. The moves, mapping to progression
of idea discussion, seem to take a very long time on the order of months. I
would not find a serious discussion very productive on that time scale.

cheers,
    Simon



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT