- 126comments
- 40comments
- 28comments
- 433comments
- 176comments
- —discuss
- 2comments
- 49comments
- 11comments
- 3comments
- 543comments
- 16comments
- 29comments
- 107comments
- 98comments
- 128comments
- 186comments
- 1comments
- 33comments
- 301comments
- 360comments
- 23comments
- 18comments
- 89comments
- 23comments
- 9comments
- 50comments
- 4comments
- 44comments
- 7comments
I got a security pop-up going there (Exploit: HTML/IFrameRef.gen).
Very interesting read. I can only imagine how exciting it must have been for him to utilize these concepts so effectively.
For anyone choosing not to read, it's about one man's invention of a program that is the closest thing to Artificial Intelligence that I've ever seen. See the last paragraph for the best example.
FWIW, Doug Lenat founded Cyc after doing Eurisko (http://www.cyc.com/). His first discovery system along these lines was AM ("automated mathematician"), which was math-only, and Eurisko was an attempt to make it domain independent, so it could target things like the famous gameplay-strategies example. He eventually grew frustrated at how much domain knowledge had to be hand-coded for a new domain, though, so hoped that by building one massive database of commonsense background knowledge, future systems could more easily be retargeted to new domains: the system would be able to assume most "normal" knowledge, and only the information unique to that domain would have to be added.
And that sort of turned into its own multi-decade project...
http://lesswrong.com/lw/10g/lets_reimplement_eurisko/
One recurring theme in these discussions is that no one has actually seen Eurisko's source code except Lenat, so it's very difficult to know how much directed guidance was required to get the results in question.
Is Cyc/OpenCyc dead?
http://blog.cyc.com/ - last post Nov 2009
http://www.cycfoundation.org/ - last post Nov 2008
http://cyc.com/cyc/company/press - last post Sept 2009
If Cycorp is dead, what is Doug Lenat up to now?
Quoting myself (from http://news.ycombinator.com/item?id=1859185 ):
I've been looking for material about Eurisko. Other than this championship, there appears to be no credible documentation of any of Eurisko's reported achievements.
It might be super-duper, but the fact that he wouldn't let anyone else use it, and that those amazing feats have not been reproduced by others despite huge advances - might just mean that there's more myth to truth in the stories about Eurisko.
(and ... quoting jonoff from same thread: http://news.ycombinator.com/item?id=1861890 )
Adding to the mystery, I read the newyorker article (2009) linked from wikipedia on Eurisko, and it seems Lenat forgot the winning strategy Eurisko developed, or the first article is wrong.
"Eurisko, however, had judged that defense was more important than offense, that many cheap, invulnerable ships would outlast fleets consisting of a few high-priced, sophisticated vessels. There were ninety-six ships in Eurisko’s fleet, most of which were slow and clumsy because of their heavy armor" - 1984 article
"astronomical number of small ships like P.T. boats, with powerful weapons but absolutely no defense and no mobility, Lenat said. They just sat there. Basically, if they were hit once they would sink." - 2009 article <
I don't really see the mystery (unless you're suggesting that he never did win the tournament, which seems very unlikely). It'd be surprising if he did remember the exact strategy 30 years after the fact.
The Eurisko/Traveller story has never made sense to me psychologically though. People who are hardcore enough to enter a tournament for a very math-heavy game have probably already spent ages hacking the system, twisting it into an unrecognizable shape. It should have been very unlikely for these loopholes to have first been found by a program years later.
https://web.archive.org/web/20050308172043/http://www.alicia...