<PRE>
Subject: Re: Xanadu is answer to what?
Date: Wed, 23 Aug 2000 10:14:24 -0500
From: Jeff Rush <jrush@taupro.com>
To: "David C. Kankiewicz" <kankie@thegrid.net>
</PRE>

Hi David,

"David C. Kankiewicz" wrote:

<TABLE BORDER=0 BGCOLOR=SALMON><TR><TD>
  I am very interested in digital information structure, connecting
  information, and pretty much anything related.  After receiving your
  email announcement (almost forgot about that list) and visiting your
  site, I got to wondering just what Xanadu was trying to solve.  In
  the same line of thought, I guess, why wasn't their more people
  trying to solve the problems Xanadu reportedly could solved.  The
  conclusion I came to, after re-reading its history and As We May
  Think (among other documents), was that its only half the answer.
</TD></TR></TABLE>

I believe Xanadu hasn't done a good job providing a concrete
explanation of their solution to the problem.  They tout that the
problem they are solving is nothing less than the elevation of human
thinking, bringing it into an improved virtual community of organized
discussion.

However, their materials I've seen focus so heavily on the back-end
that they haven't done much innovative on the front-end, although I'm
sure Ted Nelson would disagree, citing his zippered-lists and side by
side contrast viewer.  However, the front-end needs so much more.

You are correct that once Xanadu is initially built, it becomes
mostly a way for a small group to coordinate their documents.  I'd at
least like to see that happen, and I am so surprised/frustrated that
the Xanadu group didn't at least achieve that, and *then* work on
scaling it up to a global system.

<TABLE BORDER=0 BGCOLOR=SALMON><TR><TD>
 Its a solution, INHO, to single document or multiple document
 organization and building referencable structures.  I have to admit
 its a really interesting and fundamentally workable concept with
 many applications in day to day operations within *small*
 interconnected groups.  Within larger groups, the concepts, as far
 as I understand them, lose their appeal because of the inability
 to locate documents that should be referenced.  A chicken and egg
 problem, for sure; one has to know of the document to be able to
 reference it.
</TD></TR></TABLE>

There are three possible solutions to locating documents:

The Xanadu view is that "experts" will continually track those documents and make them available in reputation-based public directories, similar to how we trust or don't trust certain websites
today.  I agree that this is quite fragile and not likely to be
successful in the big view.  It also implicitly assumes there will
be one or a few central places to learn things.  Politically
dangerous and not scaleable.

The second view, still discredited in my opinion, is that AI programs
will scan the public tapestry, parse the english and create nice tidy
indices.  I think this is not feasible in the near future, as the
various search-engine R&D departments are showing.

The third view is I think where you and I see eye-to-eye.  I see an
open ended, structured discourse over time, where instead of setting
up fiefdoms on each topic, we weave a tapestry of
knowledge/point/counterpoint
where people can bring up topics, others can critique it and filters
can be set to highlight or mute those we believe are most effective
in making their points.  *All* the knowledge won't be in such a system
and the jungle of the current web will continue to thrive, but being
able to place a *layer* of discussion over the jungle, pointing to
the good stuff, rebutting the erroneous stuff and tying together
seemingly disjoint topics is, I believe, what Eric Drexler et al
are saying they want to manage global issues like nanatechnology.

On the Internet anyone can say anything, however wrong or dangerous.
The solution is not to appoint experts who only get to speak but to
allow the group-mind to filter and assign importance to such speech.
(but I ramble...)

<TABLE BORDER=0 BGCOLOR=SALMON><TR><TD>
  This all parallels the problem of searching and locating documents,
  or knowledge, information, data, etc..  While the Xanadu structure
  would probably work if everyone implemented it, which is a huge
  problem with market acceptance (maybe the reason it never got off
  the ground), people usually only change when required to...  And
  the indexing of information is still a massive problem, no
  automatic concept mapping yet, and we are left with the same
  problems we have today... Can you say, information overload?
</TD></TR></TABLE>

I do think though, that if we adopt a system of explicit
linking/tagging that software can then be written to traverse the
Xanadu space and report interesting congruencies, summarize
conclusions and watch for the appearance of selected, new information.

By analogy, instead of building a robot rover (the AI) that has to
run over wilderness terrain (the web today), we build roads thru the
wilderness (Xanadu++) and then build suitable vehicles that can
automatically take you to relevant national parks (sites of structured
discussion) and then let the human disembark and explore the
unstructured portions manually.  And as he does so, he can blaze new
trails for the next traveller, exactly as Vanaever Bush talked in his
*As We May Think* article.

<TABLE BORDER=0 BGCOLOR=SALMON><TR><TD>
  I'm wondering if the logical solution would be to implement a
  massive store of (single concept or tiny pieces of) information,
  data, etc. and use it along side the Xanadu system for indexing
  and general knowledge?  Well either way, I'm wondering what the
  requirements for interfacing such an archive from Xanadu would be
  and if your interested in making sure something like this could be
  interfaced as apart of your implementation???
</TD></TR></TABLE>

I'd love to.  I think it could be done with Xanadu's linking system
rather than a parallel system.  I'll add a [Philosophy of Linking]
link to my ZWiki and start putting some ideas there.

<TABLE BORDER=0 BGCOLOR=SALMON><TR><TD>
  A simple overview:  The structured archive would be built on a Node
  hierarchy, with hierarchy nodes (subjects, etc.), grouping (by type,
  etc.), virtual nodes (collections of Non-child/parent nodes, etc.),
  thought nodes (perspectives, etc.), two way linking (types:
  supports, part of, analogy, etc.), object specific data (plants
  have growing data, etc.), unique searching interfaces for every
  type of data, and all this is really extensible... I'm prototyping
  (still playing around) with postgreSQL database with a Zope
  interface, so its designed to change.  I even went as far as to
  create a nonprofit org for the overall goal of the project, only
  its really one person until something gets going or I feel people
  want to support it.  (A lot of questions about Intellectual
  Property, etc....)
</TD></TR></TABLE>

Ah, Zope.  Yes, I'm into Zope as well, and am scratching my head
figuring out how to integrate Zope and Xanadu.  Certainly Xanadu
documents could be written in DTML and XML, and the ability in Zope
to factor out common source into fragments 'acquired' from elsewhere
in the hierarchy is something we need to address in Xanadu.  However,
in Xanadu it may be addressed by defining a 'link' to the fragment
from inside the master document, and specifying that the 'link'
should follow-the-latest-version, to get the proper effect of a
header/footer modification rippling thru the site.

FYI, I'm also looking at the Freenet project as a place to store
documents, since it has some of the features Xanadu wants in a
back-end <-> back-end network.  It's also important that Xanadu
not be easily censorable, IMHO.

-Jeff Rush

P.S. May I have your permission to repost our discussions
on the ZWiki, in the hopes of encouraging others to join in?