Robert Jasiek
2016-03-17 12:19:24 UTC
Ugh, just wanted to post this on L19 in reply to a new thread, but
apparently a moderator is fighting a spammer and, in this case, blocks
discussion of a valid question. So let discussion take place here.
***
A match does not prove anything about complexity. For theoretical
measures of complexity of go, see
https://en.wikipedia.org/wiki/Game_complexity
https://en.wikipedia.org/wiki/Go_and_mathematics
http://home.snafu.de/jasiek/rulesfaq.txt
For practical measures, these two parameters give a good hint: average
game length (number of turns) L, average number of a player's legal
next moves M. Then L^M is a very rough estimate for practical
purposes.
The theoretical or practical measures put the complexity of Go above
that of Chess.
A different view can measure how long it takes to master the game. For
Chess and Go, "a lifetime" is a reasonable answer. Therefore, from the
POV of skill required to become a world champion, both games are
essentially easily demanding.
Yet another view describes the relative impact of strategy, tactics,
positional judgement, psychology and time management. Go seems to
emphasise strategy and tactics equally; Chess emphasises tactics over
strategy. However, neither is more complex than the other. We can
simply say roughly which fraction of thinking must concentrate on
which aspect.
Computer - human matches rather say something about the complexity of
designing particular kinds of AI. More complicated AI is more complex
than simpler AI from the POV of the researchers. Or we might compare
the "design complexity" of an AI versus the human brain.
apparently a moderator is fighting a spammer and, in this case, blocks
discussion of a valid question. So let discussion take place here.
***
A match does not prove anything about complexity. For theoretical
measures of complexity of go, see
https://en.wikipedia.org/wiki/Game_complexity
https://en.wikipedia.org/wiki/Go_and_mathematics
http://home.snafu.de/jasiek/rulesfaq.txt
For practical measures, these two parameters give a good hint: average
game length (number of turns) L, average number of a player's legal
next moves M. Then L^M is a very rough estimate for practical
purposes.
The theoretical or practical measures put the complexity of Go above
that of Chess.
A different view can measure how long it takes to master the game. For
Chess and Go, "a lifetime" is a reasonable answer. Therefore, from the
POV of skill required to become a world champion, both games are
essentially easily demanding.
Yet another view describes the relative impact of strategy, tactics,
positional judgement, psychology and time management. Go seems to
emphasise strategy and tactics equally; Chess emphasises tactics over
strategy. However, neither is more complex than the other. We can
simply say roughly which fraction of thinking must concentrate on
which aspect.
Computer - human matches rather say something about the complexity of
designing particular kinds of AI. More complicated AI is more complex
than simpler AI from the POV of the researchers. Or we might compare
the "design complexity" of an AI versus the human brain.