Date: Thu, 01 Jun 2000 10:32:33 -0400
Mr. Rod Welch
The Welch Company
440 Davis Court #1602
San Francisco, CA 94111 2496
Welchco site comments and proto-proposal
I must say that every time I poke around the Welchco site I am more and
I especially like the better sense I get of informal communications
related to the Engelbart Colloquium.
I just noticed you comments on mine from:
I have comments on some of the
issues you raise (5330 -),
time enough right now to address them. I guess this also points to
issues of having multiple ways of storing and communicating -- I was not
aware of your comments but would have been very pleased to have seen
them near the time of that post.
Most of my replies would have to do with elaborating on topics in
William Kent's book "Data & Reality", and as I mentioned in that email
the list was in large part inspired by looking at that book's table of
I also appreciate your comments on Pointrel:
Having poked around Welchco some more I have a comment and a
I agree with the sentiment that Welchco mentioned on Welchco
(related To Eric's and others comments) is hard to navigate. However I
don't think this is a failing of the Welchco site directly. The issue is
more that Welchco packs so much information and links together that it
demonstrates the weakness of HTML-based hypertext for handling complex
data. In short, HTML is designed for one view of data -- but with
complex data you often want to view the same data but in different ways
depending on your search motivation or level of understanding (e.g.
introduction, overview, tutorial, thread following, subject based).
For example, right now I have about twenty Netscape windows open from
trying to follow various topical threads via links on Welchco. There is
so much interesting content I lose track of why I opened each window, or
which window links back to a previous topic. Browsers also provide
limited context of just exactly what on the page is linked to.
What I would like is:
Many of these could be construed as limitations of the HTMl browser I am
using, not Welchco, Communication Metrics, or SDS.
Of course, other features like enabling "privacy" or "email-hypertext
integration/replication" would require more direct changes to
I've been thinking of using Pointrel to archive mailing lists. But your
site suggests another possibility.
I call it "proto" because I probably don't have the time
to do it right now and I have reservations related to squabbles over IP.
So it's open to negotiation and abandonment. But here is a thought of a
[Note: I could easily decide to substitute "Squeak" for "Python" in the
- I download all the contact on your site using a site grabber program in
- I create a Python program that parses the HTML and puts is into a
Pointrel data repository, complete with links modeling relationships.
- I create a browser in Python that supports search, meta-information,
limited (easily defeatable) privacy, and pivoting over that
- [Optional] I create Python tools to allow adding new materials and
exporting repository materials to HTML.
- [Optional] I integrate this with a mail reader and sender.
- [Optional] I modify Zope to serve Welchco content from a repository
[not static HTML]
- The content remains yours, and is either returned to you in a Pointrel
repository or better as a set of Python tools you can use to build your
own repository from your site. [Optional: Maybe you let me keep a copy
for additional local testing, not to be redistributed without your
- The resulting Python code is put under the same license as Pointrel is
now (X/MIT) [or maybe instead GPL or LGPL.].
- So, you share your experience doing Knowledge Management with hypertext,
especially the great example you have produced with Welchco, and I share
my experience with data repositories and GUIs -- and at the end there is
an open source result.
Some reasons you might not want to do this:
Some reasons I might not want to do this:
- It undermines your proprietary interest in the structure of your site
and related tools.
- It creates more copyright & licensing issues & liability in your
redistributing the web content.
- It would take some of your time.
A reason we both might not want to do this:
- It's a lot of work (mainly for me) and I'm working full time on another
project right now.
- It might be better to start from an email-based system and build up to
a HTML generation capability.
- Your Welchco site may not be consistent or easy to parse.
- The initial Welchco site parsing code is probably a throw-away as far
of being of interest to others.
- I may decide to abandon anything hinting of the Bootstrap Institute /
Engelbart Colloquium if it goes commercial in a silly way.
A reason to do this:
- Collaborations can easily go sour.
Anyway, just some things to think about. Feel free to say this doesn't
interest you at all, or suggest something that you think makes sense.
Other possibilities could include you do the work in Python or Squeak
and I advise you (as long as the result is open source).
- It would be nice to see some open source KM code, even if it isn't
everything an OHS could be.
- You might have some better open-source tools for doing your
- I would have a good example of using Pointrel to show people.
Developers of custom software and educational simulations
Creators of the Garden with Insight(TM) garden simulator