Thought Processor
[ May 21, 2002 10:23 AM ]

Word processors have been around for years, they were one of the first useful applications for personal computers. Word processors help you to write, but they don't make you a better writer, unless your benchmark is correct spelling and pretty formatting (and even then they fail).

What surprises me is that no one has ever come up with a "thought processor" – a software application that helps you process your thoughts and ideas; the equivalent of a word processor for thoughts. Now perhaps, this actually might make you a better thinker.

How would it work?

I see the thought processor as relying on three key features:

• The ability to classify data in a hierarchical tree or graph topology;
• The ability to suggest possible new sets of relations between data classifications;
• The flexibility to visualize and manipulate data interactively.

All these features are technologically possible and currently available in various research and commercial software programs. The secret is in combining them in a way that permits a computer assisted, interactive exploration of ideas in a "desktop" application that's as simple to use as a good word processor.

It might work by allowing you to enter in all sorts of "data" in many different formats – text, drawings, images, mathematical formula. All this information resides in a "pool", a big space that holds everything you've entered, plus maybe some standard built-in things. Now, the tricky bit would be to do some kind of information processing on all the diverse data in its many formats to discover relations and classifications that you (the operator) may not have noticed or found difficult to express. There might be sets of built-in recognizer types or "concepts" such as people, names, formulae, size and dimension relationships, space, time,...

Each person's pool could be linked (via the Internet of course) so you could look at inferences from other people's pools. It might seem difficult to express a common relation between so many different forms of input, but with the computer we already have a common representation for all data – bits. Of course, this says nothing about the quality or reliability of that information, so I imagine that classification would need some kind of confidence measure about how reliable the data is. This could be based on how useful it has been in the past and also given a weighting according to its specificity (how generally or specifically it can be applied), plus perhaps the reliability of the "source" of the information, or the number of "experts" who support its validity.

My clunky version of the thought processor is actually a number of different programs which include a text editor, web search engine, mathematical manipulation and visualization program, data visualizer and information classifier. It doesn't work very well because they are all separate programs with their own funny way of doing things, plus a very limited ability to exchange data in any serious way (program exchange is only as good as the operating system that supports it).

That might be why no one has yet written my perfect thought processor — there has yet to be an operating system that could support it...

[created: October 11, 2001 5:00 PM, last modified: May 21, 2002 10:23 AM ]