Date: Thu, 13 Dec 2001 16:08:49 -0800

Mr. Rod Welch
rowelch@attglobal.net
The Welch Company
440 Davis Court #1602
San Francisco, CA 94111 2496

Subject:

Meeting Wed Dec 19 10a

At 02:40 PM 12/13/2001 -0800, Rod Welch wrote:

Fine. Directions to my office are in....

At 02:40 PM 12/13/2001 -0800, Rod Welch wrote:

Browsing a bit more through SDS entries (I assume that's what your cryptic URLs are to) I've gotten more of a sense of what it is. It seems very much like the original NLS system, with hyperlinked outline-structured entries used for a wide variety of purposes. So I'd be interested in what Doug sees as the differences.

A few observations:

1. From an HCI point of view I find it maddening. The screen is littered with non-informative numbers and cryptic link names. Those are great for internal identifiers but there is no reason the user should have them as distractions all the time. You can either have informative anchor text, or just a simple link symbol.

2. It isn't clear what if any access control it has. I was reading about various meetings with Pat Lincoln, responses to business plans, etc. Presumably anyone on the web could as well. I know that the original NLS took the attitude that it was for internal use in a mutually-trusting group who wanted opennessand didn't care to hide things. Is that the same?

3. The web version, at least, didn't make any search tools available. I assume that there are other interfaces that provide more in that direction.

4. From a number of the comments you have received, it appears to fall into the problem that a great many systems do when they try to get people to perform explicit recording and linking so that later information access will be enhanced.

   As with things like design rationale, few people are willing to take the time to do this unless it is so automatic as to be invisible. A few people (either because they love the system or have certain personalities) will be compulsive about making sure things are properly entered, but since they are in the minority, the overall web of entries can't be trusted to be complete, up to date, etc.

   The typical way around this is to drop the structuring and let people have open-ended search capabilities through documents produced naturally (email to people both in and out of the systems, documents to be printed, etc.) so that if information has made it through a keyboard in any form it can be found, even if it is not properly linked. Does SDS integrate with these other forms?

5. The "information overload" feel is partly the interface (see above) and partly because of the overall hyperlinnked structure rather than a clear hierarchical structure where you can do a depth-first traversal of main points without having to wade through detail. Again this may be a function of my using the web interface instead of better ones you have.

6. My own experience working with Flores was that in educating people it can be useful to have strong abstract theories of management, but that in building useful tools the devil is in the details, not in the good theoretical intentions. When I'm looking at systems, I'm not so interested in what theoretical principles the designers had in mind, but in what experience is actually created for the users.

This is one of those information overload places where I started trying to see what you meant and the forest totally got lost for the trees. I'm not a great PowerPoint fan, but I long for a few slides with 5-plus-or-minus-2 bullet points in short phrases.

At 02:40 PM 12/13/2001 -0800, Rod Welch wrote:

Based on the above, you can decide.

At 02:40 PM 12/13/2001 -0800, Rod Welch wrote:

See the book and various papers I wrote or contributed to in

Sincerely,

--t

Terry Winograd
winograd@cs.stanford.edu
Stanford University

Copy to:

1. Engelbart, Douglas C., doug@bootstrap.org
2. Park, Jack, jackpark@thinkalong.com