IMHO - this statement represents a fundamental mis-understanding regarding how computing power was able to beat humans at Go, as well as some of the differences between Go and Chess.
Computers were able to beat humans at Chess because Chess is vulnerable to brute-force attack. In other words, a computer could actually read out ALL of the possible board positions and determine the one with the best winrate.
If we were waiting for computers to become fast enough to beat humans at Go using this brute-force approach we would be waiting until the heat death of our universe. The number of possible positions is too great, and evaluating each one to see which of them was better than another would take several more universe-life-times.
Instead, the people who create the software those computers use had to take a completely different approach - neural network learning (as opposed to algorithmic programming). This process involves training a neural network over time to mimic some of the analysis, prioritizing, and decision-making that a human mind would do, as well as providing the means by which it can investigate and find new ways of combining that information to generate novelty.
So yeah, the reason AlphaGo / Master / AlphaGo Zero are able to beat top pros isn’t because they’re “faster” - it’s because they do something completely different than old computer programs used to do.
Also, I will never count an AI “win” as a “win”, unless the AI becomes so “I” that it enjoys it. Until then, it’s a program made to do something. I’m glad excel exists and I can do better things than calculations half my work day.
Tell it to lose, it will lose. Tell a human player to lose when they could win against another, and their neck muscles will freak you out.
I’m not so sure if the same could be said if we scale chess up to giant chess, played on e.g. a 20x20 grid, with more pieces doing more things, for example with large variants of shogi the game also becomes extremely complex for computers to win using the same methods applied in '96.
No, it’s not just computers that became faster, although that is also part of the reason. One big reason is that modern AI uses deep learning methods that wouldn’t have been feasible in '96, since there wasn’t enough computing power. But AlphaGo, and especially AlphaZero, is not just the '96 method “scaled up”, it’s really a different way of building artificial intelligence, and the AlphaGo papers were genuinely a breakthrough in AI research.
I’m very sorry to hear how chess has fallen into such a generally unfriendly state, which I have heard from others as well. It wasn’t always so. Forty and 50 years ago, there was a tremendously convivial atmosphere at chess tournaments. The large majority of players would review the game with their opponent in the friendliest manner, revealing what they were thinking along the way. It spilled into the halls, like a party, where some top young players would flop down on the floor and analyze with a small group of friends (most memorably I saw Joel Benjamin and Seirwan do this as young teenagers). I loved that because it was such a contrast to the bridge world, where most of the players were unpleasant. The crucial difference, I think, is that individual competition breeds personal responsibility.Chess and go have that in common. I found the same friendly atmosphere in competitive distance running (my other youthful passion). Competitions were deadly serious, but characterized by intense respect for one’s opponent (the occasional disrespectful one was a “hot dog”). Afterwards, competitors were so affable, talking shop, giving advice about injuries (runners had little sports medicine back then), and even partying together. Everyone recognized and admired the effort of others, at all levels, because we all knew how hard it was. It would have been sheer meanness to do otherwise.
And now I find again that wonderful, we-are-all-in-this-together atmosphere in go—a little less pervasive, perhaps, due to the demoralizing nature of online psychology, but nevertheless heartwarming. If the reports are true, that is go’s main edge over chess.
Just remember, in order to win the argument with your friend, you only need to convince him that Go is as good as chess, not necessarily better. Much like in Go, you only need to win by half a point.
I’m not sure that rescaling is a good argument. I find it’s correct to keep each game at the size it is played commonly or we are then going to debate if a 19x19 chess would be playable and enjoyable by humans… And then why not a 25x25 goban…
Well I dunno how interesting it is for the sake of the argument, we don’t play these games whatever. When they applied the new technology to the normal size, they did put much less time to get a dominating AI for chess which is again a proof in terms of combinatoric.
More interesting is to underline that complexity is not just introduced by pure combinatoric but by variety of pieces and movement (and resurrection in Shogi). Complexity for human of course, not for AI.
Chess players in my own experience make quickly interesting partners in weiqi because they have a taste for reading which is not always shared by others. Same time it can be a big obstacle to introduce our game to them because it’s hard to accept to not base your approach on this reading and let ideas flow without backup.
I don’t see that topic as much as evil if we can keep it under control and if it doesn’t hurt susceptibilities too much, and this is difficult because on many points the games are opposing themself: building/destroying emptyness/fullness simplicity/variety in the rule…
At least when arguing, if we would like to introduce weiqi to them, let’s not forget the cultural side: it’s not easy to propose a new board game to someone who enjoyed chess with his papa* and classmates and keep a good fun out of it. Respect.
I sort of forgot about this, but indeed the Netherlands was dominating go in Europe in the 80s. But in the 60s and 70s it seems that Germany was dominating (https://en.wikipedia.org/wiki/European_Go_Championship) and the Netherlands was #2 in the 70s.
After the collapse of the Soviet Union, there was an influx of strong players from Eastern Europe and Ronald Schlemper quit playing go. But the Netherlands kept a place at or near the top in the 90s with strong players like Rob van Zeijst, Guo Juan and Frank Janssen and the Netherlands got the EGCC.
But in the 00s, Russia and Romania took over the top spots and other Eastern European countries are following since the 10s. And the EGCC just closed its doors.
During the 00s and 10s the Netherlands dropped many places down in a ranking list of European countries. A major reason is that most of our top players are still the same players as 30 years ago. They are now in their (late) 50s. This reflects what is happening with our membership in general. The average age is over 50. Go has become an old men’s game here. We have far fewer young strong players than many other go countries. This stagnation is a result of neglecting youth go for decades.
This makes it seem like it suffers from the same thing as other smaller associations and clubs in general these days: the people taking care of it are essentially the same who founded it or joined when it was fresh and new, and the association grows old with them.
For me it’s linked with the general crisis ambiance from decades, where the commitment for associations, future generations and so on dropped down to give place for survival and individualism.
Those may be causes for stagnation of the go community in the Netherlands, but it doesn’t explain why chess in the Netherlands doesn’t seem to suffer as much and why go communities in some other Western European countries are doing just fine. For example, France and Germany seem to have a vibrant and growing go community with lots of young strong players.
So I feel there are some specific things that the Dutch go community did not do as well as communities in other countries. I strongly suspect that is attracting and keeping new players, youth players in particular.
That’s a really interesting distinction I had not thought of before, at least in those terms.
So how would one categorize games where all past positions still exist in a very relevant sense, and each new position is created whole-cloth?
Zendo (aka Xendo) would be the prime example of this.
And I suppose Bao la Kiswahili is extremely mutative, far more so than chess, whereas Arimaa, while closer to Chess than Go on this scale, is somewhat more iterative than Chess by design.
I had to look up all three of those games, proving myself completely uncultured – I’m going to enjoy reading about these.
I should note that there are examples of completely iterative games (which as said, Go isn’t quite), including gomoku / renju (and other connect-N games), Carcasonne, Trax, and most obviously dominoes.
There are also games where one aspect is iterative but another one is mutative. An example is Quoridor, where players can either place walls (iterative) or move their avatars (mutative).
Edit: Just read about Zendo, what a great original idea! So simple!
For what it’s worth, I found chess clubs to be a bit of a mixed bag. There was one in my old city that took chess very seriously, wasn’t much of a learning club, and met in a library. The one I went to met at a bar, took no issues with players kibitzing active games while eating burgers, and had a lot more fun. I played tournament chess in high school (~15 years ago, now,) and it was likewise a relaxed and enjoyable environment. No beer, though
I imagine Go clubs enjoy a similar range in areas with lots of Go clubs. Unfortunately, my current city has none, but I intend to try to get one started once COVID is behind us.
