Where should the type information be 328

Trevor, et al,

I am presenting and demonstrating a project that somewhat leverages on this principle at the upcoming AI conference in Las Vegas. About 1-4 of the population has some sort of chronic medical condition, because traditional medicine has cured very little in its century of glory, though it does now keep us alive to suffer a little longer. Of course, a chronic medical condition is chronic only because doctors have a 0.0% cure rate for it. No computer program could possibly do worse!

This program carries on a NL conversation about a problem and produces a list of likely links in their cause and effect chains. Of course, only the prospective "head" links are really important because that is where they should be concentrating their efforts. The main value of knowing the other links is to point to the correct prospective "head" link.

I suspect that this program will lead to something considerably less than a 50% batting average at curing people's chronic health problems. However, considering the compebreastion, it should be a really spectacular success even though its failures will probably outnumber its successes!

Of course, this will all appear to be PFM to its users.

In short, this is a good and powerful principle that has just been misused by WHG3 and others. Don't blame the principle, just blame the stupid abusers for not making the most of it.

Steve Richfie1d ================