Go Where You Are Afraid

In the GameOfGo, you often know which moves are good and which are bad. But you also often see moves that are just damned scary, where you don't know if they're good or bad. Play there - you might just learn something.

That's good advice. The reason I got interested in the GameOfGo in the first place was that the decision problems involved in it seem not to be amenable to computation (on the current state of the art): current Go playing programs are notoriously weak, compare with the best Chess programs. In fact, it's this hard: pondering Go end-game problems led JohnHortonConway to invent a new kind of number, SurrealNumbers, which were then the subject of a nice little book by DonaldKnuth.

Are you sure about your history here? Nothing I've read about SurrealNumbers indicates that Conway was led to them particularly by thinking about Go endgames. (His theory has certainly been applied to Go endgames, though.) --GarethMcCaughan

Not 100%. I do seem to recall an interview with JHC a couple of years back where he mentioned that he'd never got any further with Go that pulling out the recursive definitions that lead to SurrealNumbers. My sub-concious could be making it up, though.

OK, I just got surer, see [http://www.maa.org/mathland/mathland_3_18.html]. I quote:

John H. Conway, ...trying to understand how to play Go,
[...]applied the same logic to other games of strategy,
[...]and he came to the conclusion that certain types of
games appear to behave like numbers with distinctive properties.

See also: WorstThingsFirst, IntoTheTrulyTerrifying?


View edit of June 2, 2002 or FindPage with title or text search