The go-playing literature offers a bit of advice: when ahead, make moves which
simplify the game and preserve your advantage. When behind, take some risks to
grab more than you are "entitled" to - but not too many. Computer programs seem
bizarre in this regard, they tend to play quite unsatisfactory moves when
behind. It would be more attractive to see the computer preserve its advantage,
the better to spring when one's opponent lapses.
A complex seki was recently posted. Magnus explained that his program does not
understand the seki per se, but finds the correct move through several
intricately-meshed rules which prune bad moves. He's not entirely sure of the
correctness of those rules - that is, whether they apply to all cases, or
merely most cases. At the single-digit-kyu level, inducing sekis is not
unusual; it is basic for dan-level players.
Some high-dan players prefer to give a large negative komi instead of handicap
stones; the game mechanics are closer to those of an even game. The stronger
player gains incrementally with every suboptimal move by the weaker;
eventually, the advantage of the negative komi dissipates.
When can we train our programs, instructing them "don't do that, it will only
hurt you?"
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
