Anti-AI opening?

[John Fairbairn]

I was intrigued by this fuseki. So Yokoku, a high-ranking 9-dan, was pitted against one of 2020's new 1-dans. It might be argued that he was extracting the urine as a protest at such an obvious mismatch, but I don't think that is remotely likely, from either the context - a sort of reward game for qualifying as a new pro - or past experience. (A high-ranking amateur in Japan played a pass for his first move in what he thought was a mismatch, but his opponent appealed and was given the game.)

My own suspicion is that it was an anti-AI strategy by a player who assumed his young opponent would know nothing but AI openings and probably says "Dosaku who?"



So's plan worked - he won. But the way it worked looked especially interesting. He reduced the game to a five-groups-versus-five-groups format, as here, applied a touch as amashi seasoning, and at the end of the game all White had left on his plate was four kipper-bone territories and one poffle of around 15 points in the lower right. Nothing dead (apart from the kippers). And note that in the final fuseki position shown, Black seems somehow to have more stones, though with still just five groups.

A masterpiece of human thought, in its way, I thought. And fantastic instruction for young Kondo.

[xela]

Interesting stuff! I wonder how many people will post ironic AI analysis of the anti-AI opening,

[Bill Spight]

OK, I'll bite.

I do think that this is a case of the veteran taking the new player into unfamiliar territory. And since the popular openings are all analyzed with AI, that perforce means making plays that are not the bots' top picks. However, particularly in the opening, we cannot say that a play that is not on the bots' collective radar is a mistake, so there are many playable moves that the bots do not choose.

and are traditional plays that, while not the bot's top choices, cannot be considered mistakes. may be anti-AI, but it's anti-traditional, too. Even if it is a mistake in the sense of reducing Black's chances of winning, it still looks playable. Black continues to play large knight's approaches, which are neither AI nor traditional. However, and look like AlphaGo choices to me. AlphaGo chose such keimas over pincers, which humans were wont to play before the AI era.

I don't know, but looks to me like a side extension that the bots may not like. In any event, Black's unfamiliar opening has forced White to rely upon his own judgement, which is not as well developed as that of a veteran.

Did Black choose an unfamiliar opening? Sure. Anti-AI? I don't think so, given the AlphaGo keimas.

[Ferran]

O-kay... sorry for ressurecting an old thread. I don't know what's the appropriate time limit, here, but I think this is both recent enough and a short enough thread to allow for it.

I was thinking (yeah, sure, run for the hills...), not so long ago (say AlphaGo-Lee game 1) people played against computers using moves that were not in their "database", were they'd have been inputted by a person. Now we are evaluating moves on the likelyhood a person has learned them from an AI.

I find it a weird mix of interesting, puzzling, worrying and exciting.

Take care.