Re: computer-go: Good Play (was FPGA)

[Robert Jasiek <jasiek@xxxxxxxx>]

Matt wrote:
> Is Go inherently more difficult than Chess?

On a human strategic level you mean? I'd say yes: In chess after
the joseki the number of moves until the game end is considerably
smaller. So in Go planning must be deeper even on the strategic
whole board level.

> If so, is there a way to quantify the degree of difficulty?

Take standard joseki books for either game, determine the average
joseki move numbers and the average game length move numbers
(e.g. in go until we reach a (0|0)). The difference of middle
(and end) game move numbers may be a degree of greater difficulty.

> Do expert Go players really play near perfectly?

No.

Apart from comparing the constant game tree node numbers of a 
human evaluation and of the complete game tree:), having studied
abundant ko situations also on only small boards and their
surprising, easily overlooked variations tells us that experts
do not play nearly perfectly, which may be measured as the
maximal error value for a move where the error value is the
score value difference between perfect play and chosen play
followed by perfect play.

--
robert jasiek