440 Davis Court #1602
San Francisco, CA 94111-2496
415 781 5700

Date: Tue, 08 Feb 2000 02:02:20 -0800

04 00067 61 00020801

Eric Armstrong
Colloquium at Stanford

Subject:   Knowledge Space
DKR for Open Source: Core

Dear Eric,

Powerful question in your letter today.

What do we want the system to do?

One approach is to say we want to augment the driving force of human capabilities: intelligence (set out in the record on January 20, reviewing your cogent query on January 17 about how to design for knowledge management). If we can build and use a tool that helps generate useful "intelligence," it will be the proverbial tide that lifts all boats, helping people solve problems, large and small. How is human "intelligence" augmented now? The alphabet is a solution that has existed for thousands of years without any significant advance, as reported on November 11, 1999. It is the little engine that drives civilization.

How does it work?

Alphabet technology gives external form to the internal thought process that grows new knowledge from daily experience through a simple cycle of plan, perform, report. Making objects out of mushy subjective thoughts permits objective improvements. The current application in books, memos, articles, email, commonly called "documents" creates information. Bill's wordprocessing tools provide greater power in arranging and editing, making things bold and colorful, basically pretty. But documents still only convey information, not intelligence. So, it is not a really big improvement. We need a new technology that moves up a notch on the cognitive scale from information to knowledge. Intelligence capability is the bridge.

This objective requires a definition of "knowledge" that can be effectively aided by technology.

Suppose we define knowledge as chronologies of cause and effect segmented into chunks according to objectives. At the most basic level "objectives" are human needs, but pretty quickly we get up the scale to asking how do we organize a DKR project, what are the objectives for fixing the car, going to the dentist, mowing the lawn, building a bridge? The human mind does all this constantly on the fly, and it is all organic, as explained on January 20. Integrating automated MBO with chronologies of cause and effect, that supports calling up stuff under a wide range of different subjects, where any body of information can appear in a variety of different contexts, emulates a large part of human "intelligence," because humans reason by experience. So essentially we need an automated experience machine, something that manages sequence.

Let's not get too greedy. Let's not look for a machine that does our thinking for us under the AI model, as set out in the record on 000120....

Let's follow Doug's lead, and aim for augmenting human thinking. This is the computer aided thinking model described in one of Roy Roebuck's many excellent references on 000125.

This concept strives to step beyond the notion of "documents" and start thinking about Knowledge Space, as a continuous information stream, and second, how to solve meaning drift that causes those big projects to fail, which Dick Karpinski cited on January 25. Meaning drift also causes the high cost of medical mistakes, reviewed on September 24, 1999. it causes the space probe to crash on Mars, instead of go into orbit, it causes endless argument in meetings where people waste 70% of the day, and it causes small, petty frustrations among each of us, all day long that hinder collaboration. All of the linking tools you folks are discussing in XML and so on, can help solve meaning drift, once we figure out how to capture meaning. What does meaning "mean"?

You observed the other day after coming forward with some preliminary work, that using this new environment would "boggle the mind." That is an understatement. Even a crude version of this capability, while it provides significant new power to generate useful intelligence, the work product can be a Pandora's Box for some, because it exposes a lot of complexity the human mind is designed to suppress, so that people can function. Most human thinking, the "intelligence," occurs in the subconscious. When we look at a tree, a pencil, an apple, the eyes do not take in the whole thing, they take in some outline features and our memory immediately fills in the rest from its reservoir of experience. We do the same thing in listening to dialog. Building a powerful Knowledge Space may not be as difficult as providing a means for people to use it without being overwhelmed.

I think it will likely turn out to be just another thing to get use to, like traveling at 40 miles per hour was once a big deal, then the telegraph, the light bulb, and so on, presented in Johnna's Neuman's compelling book Lights, Camera, War, reviewed on March 11, 1996. That may be the biggest contribution of the DKR: getting people used to the idea of adding "intelligence" to management. Many feel it is impossible, off-the-wall, funny and alien.

Anyway, just throwing in a few comments from "front". You are asking the right question about what the DKR should do? As with the alphabet, human capabilities can be augmented by an order of magnitude with some very simple tools. The trick is getting onto the right design path. Then, as you say, a lot of very bright people can contribute effectively in moving civilization forward.



Rod Welch