So i was wondering about this : do professional use the AI between days during multiple days matches and to what extent can it influence the game ?
Well everything I say is a speculation on my part, but I kind of doubt it. Professional Go is still very traditional, and I think it would not be viewed fair to use computer help while the game is still ongoing.
Also in the event of multiple-day matches I think it is a common practice to seal the last move, in which case AI would not be that much help anyway. For the last player the move is already set, and for the other one it would probably not be as practical to try and analyze the game without knowing where the opponent has played. But who knows…
Well often in pro games there are 2 or 3 main possibilites so i don’t think your second argument really works. However go still being very traditional is a good point. Maybe it is dishonourable to do so.
The use of a “second,” or a team of seconds, to analyze an adjourned position has been common and legal in world-class chess for many decades. So the first question to ask is whether that is commonly done in go. As a newbie, I don’t know. Whether chess permits computers as seconds, I also don’t know, as I haven’t followed chess in a long time. The only logical objection would be that computers are so overwhelmingly superior that they are not the same as a team of seconds. Of course, since both sides could use a computer for adjournment analysis, the effect would theoretically be equitable. It might even make for a better game. On the other hand, I have always found the whole idea of seconds to be distasteful. Its legal status is probably a capitulation to practicality, since otherwise there would likely be many disputes and scandals.
To quote Wikipedia on the Game of the Century, between Go Seigen and Honinbo Shusai:
At the time of the match, the tradition dictated that the player holding white had the right to adjourn the game at any time, and there was no sealing of moves before adjournment. This meant that Shusai, being the nominally stronger player and thus holding white, could adjourn the match whenever it was his turn to move and continue deliberating at home before the match resumed. Shusai shamelessly abused this privilege by adjourning the game more than a dozen times, all at his turn to play. For instance, on the eighth day of the match, Shusai played first, and Go Seigen replied within two minutes, Shusai then thought for three and a half hours, only to adjourn the game. It was no secret that Shusai, during adjournments, discussed and studied the game with his students to come up with the best moves. Go Seigen was therefore put into an especially adverse position for having to take on the entire Honinbo establishment
A few years later Kitani Minoru defeated Shusai on a match where Kitani insisted on sealing the move. That match is subject of Kawabata’s Meijin or The Master of Go.
Interesting that they didn’t seal moves, probably the result of the “honor” tradition. Sealed moves have been done in chess at least since the early 20th century, possibly even in the late 19th century.
In chess they have now solved this by limiting the time control so that it fits in one day
If the team is big enough, do they just call it a minute?
I bump this topic because there are no real answer and I am curious and surprised if the organizers of all these big tournaments with sealed move didn’t take any statement.
Yeah but it’s only move in the sequence, I’m not sure it really stops you analysing the position. Imagining sealing an Atari or what would otherwise be a timesuji, and knowing your opponent will answer. You now have a long time to think and discuss about your real next move/sequence of moves.
Indeed, I doubt there’s been a single adjourned game in professional chess in the last ten years.
Right, but Chess is a fundamentally shorter game, right? If a Pro in Chess needs X minutes reasonable play, we could suspect that a pro in Go needs at least that much (bigger state space for them to explore) so Go games are fundamentally longer. If sensibly paced Pro Chess games “just” fit into a day, then surely Go ones will not…
But what is a sensible pace? A correspondence player will say a day per move; a blitz player ten minutes for the whole game; and, currently, a tournament professional would say one – five hours for each player, with overtime. But the first professional time control, back near the start of last century, was sixteen hours / player, absolute time! Clearly this was, at the time, thought sensible. And if we look back further in time to the classical era, it was known for games to sometimes last over a week – compared to this, even a 32-hour game can seem stingy.
The point I’m trying to make is that the line in the sand which separates a time control which is “too fast” from one that isn’t is (whilst not arbitrary) fuzzy. I would personally consider that a tournament game of the top level can be played adequately within eight hours, ie. a working day, and would not need to be adjourned.
I would estimate it based on game length. I’m not sure on statistics, but maybe 40 moves for a chess game and 200 for a go game are comparable? That would make go take about 2 1/2 times as many moves to complete as chess, so should arguable have about 2 1/2 as much thinking time.
In chess, 40 moves actually means 40 pairs of moves (one by White and one by Black), which would be 80 moves in Go.
All taken into account with my 2 1/2 times figure. I use the chess convention when speaking of chess moves, and the go convention when speaking of go moves.