Quote:
I'm very excited for Seki Kotaro and happy to have different Japanese players holding titles.
I'll take a chance and presume that you don't just want different players, but also care the variety that each one brings to the party. Seki certainly seems to bring something special. For one thing he won his major title at the first attempt against a player (Ichiriki) who's been in form all year. For one thing, he just seems nerveless.
He's been around so little time that we have not yet heard much about him, but one thing stuck in my mind.
When he won one of the Tengen games, there was a comment about a particular move he played. It was prefaced by some remark stating that AI go theory was beginning to shine through in his play, but the killer comment (can't remember if he said direct, or if an answer of his was just being summarised) was that the move in question was notable because he was following AI guidance to make sure his groups "had a base, were settled, and could not be attacked."
It wasn't just the belt, braces, truss, chastity belt and suit of armour way of putting it that caught my eye.
What really hit home was a wider insight into the AI debate. As most people here know, I produce books of commentaries in which I collect as many pro commentaries as I can find. That can be a lot. For one game in Kamakura (which I just finished tonight) I had over 50 commentaries. When I splice all these comments together, I pay great attention to the terminology. In many cases I even produce a corpus, so that I can have some statistics to work with.
This has provided many unusual insights. For example, Shusai produced a very large number of commentaries for Kido in its early days. He was commenting on weaker professionals, of course. What is noticeable is that he harps on and on (justifiably, as Meijin) about these players missing probes and tsumes (checking extensions). In general, other very strong players will mention probes, but no-one gets near Shusai when it comes to tsumes. But if you stand back for moment and say, "What's so special about tsumes?" (a question I'd venture next to no-one on L19 has ever asked themselves before), the answer is quite disconcerting.
Let's analyse together. It's an extension for you that also stops an extension by him. But because it's usually a narrow extension, it hugely strengthens your position. And because your position is now strong, you are now threatening safely to invade his position next door. So there you have it? A move with
three or four functions? No! There's more. Because it abuts the opponent's position you can also regard your tsume as a probe. Because it is on the third line it makes some territory. For the same reason, it offers a subsequent safe jump out into the centre, which also would start sketching out a moyo. It also occupies the base area of a Go Seigen group. This is a real multi-tasker of a move.
So why don't people play tsumes more often? I can only guess, but in the case of amateurs I'd be pretty sure of the reason - a reason that quite possibly also applies to weaker pros: it is gote. But actually even that's a plus, because it's really a case of gote no sente. And guess what - that concept was introduced to the go world (from kendo) by the first Honinbo to follow Shusai: Sekiyama Riichi.
Now another of the many things I noticed when analysing go terminology was the massive emphasis on "settling" moves. In Kamakura alone, there were 38 mentions of settling groups in just 10 games. But if you instead create (as you should) a nexus of related words, such as bases, hontes, call & response moves, forcing moves that are reserved to settle groups and so on, you are getting into at least 70 mentions. If you add in also closely related concepts such as prophylaxis, you are quickly well over the 100 mark. In other words, a very strong pro typically can talk about settling in all its forms over 10 times in one game, and that's just the moves he deigns to comment on. And this is typical; not Kamakura-specific.
Let's be clear what this means. These comments derive from very strong players commenting on moves that, on the whole, could have been improved upon by following their advice for groups to have a base, be settled, and not be attackable. In other words, they are doing, and have been doing for decades, exactly what AI is purporting to tell us.
The obvious question is why did the players in the games (no weakies these: Go and Kitani) not play that way in the first place. Again it seems to be a problem of psychology. Base making and honte and mamori and so on are gote. Just as many people start with the premise that it's easier to ask forgiveness than permission, and to sort the mess out later, many go players like to be the most active one - bull in the china shop - and to sort the mess out afterwards. The lure of sente, sente, sente exists in pros just as in amateurs, even if not at the crudest level of atari, atari, atari. I have linguistic data on this, so it's not entirely speculation on my part.
Under normal circumstances before AI came along, young players like Seki (and Shibano Toramaru before him), would never get much of a chance to hear a very strong player berate them over and over again about the need to settle, make a base and build a castle. Indeed, it's a Catch 22 situation for most of them: you need to get to a title match to hear lots of commentators talk about your game, but you can't get to a title match until you have heard lots of such good advice.
That is really what AI has changed. Young players can do their go PhD at age 14 or so by using Katago as their very strong player. I'm not at all sure that it's of much help to amateurs because I think you need the knowledge database and the electron-microscope kind of view that a pro has, and not the mere magnifying glass and tsumego paperback that most amateurs possess. But for both sets of players, I think we can now safely say that settling moves should now be very firmly on the curriculum.
This thread started with a retrospective on the top stories in 2021. I would suggest students of go should start, for 2022, already looking to the future, a future in which their top theme could usefully be osamarigatachi.