Knotwilg's practice

[xela]

...I want to highlight some moves that came intuitive to me but weren't on KataGo's radar at all or the other way around.

Thanks, I find this style of commentary helpful. It's especially interesting to see the occasions where KataGo wants to leave the shape "unfinished".

[Knotwilg]

It will take a lot to keep it but today I bumped up my KGS rank and OGS rank to 3d and 2d. Might be inflation again but I might have grown a little again too. Anyway it's motivating to see consistent playing and reviewing pay off.

[Knotwilg]

I recently saw this video by Veritasium on expertise. It claims, supported by scientific studies cited, that there are 4 prerequisites to achieve (high levels of) expertise. Two of them are about the domain and two about yourself

The domain needs
1. to be valid, i.e. expertise in it should be verifiable and repeatable and not just attributed to luck. The trivial example is roulette, the less trivial one is the stockmarket
2. to allow for fast feedback, i.e. you should see the immediate impact of your actions so as to gauge if any modification to previous action is potentially successful

You need to
3. put in the hours
4. but also push yourself in trying out new things, gradually attempting a higher level

Go is obviously skill based so for those putting in the hours and pushing their limits, expertise should be looming ahead. The 2nd condition is the interesting one. Although the end result of a single game is a clear outcome of either success or failure, it is not so easy to attribute any move to that win or loss, let alone learning from the experience in the next game, because games are very different, especially when losing moves are played.

This is why tsumego, replaying pro games (trying to predict every move, as per John's recent advice) but also review of your own games with AI. Tsumego are isolated situations, removing most of the variety of the whole board situation, leading to insight about vital points or living/killing techniques. Predicting a pro game has instant feedback every move of the game. Reviewing your game with AI too, although the lessons can only be drawn in bulk after the game.

[John Fairbairn]

Predicting a pro game has instant feedback every move of the game. Reviewing your game with AI too, although the lessons can only be drawn in bulk after the game.

Move prediction through transcribing certainly does give feedback on every move, and this type of learning has a deeper value, too - one usually overlooked in other fields. Namely, for each move you are getting data about the exact frequency in which each type of move occurs, and by extension data on the exact frequency of contexts in which each move occurs (and all of this is being done in a way that suits the way your brain's neurons collect data).

As an example of this, one thing I noticed years ago (and wrote an article about it somewhere) was that the type of move most likely to trigger a resignation was a monkey jump. I confirmed this by writing a program to show on which line resignation-inducing moves occurred and finding examples with Kombilo. But because I was intrigued and sensitised, I started noticing (from transcribing) that monkey jumps were often ignored. So I became aware of context. Although I had some fairly reliable data to show that at least considering ignoring a monkey jump was fine, in in terms of what we might call "go knowledge" I didn't really understand this, and so would never ignore the damn simian. But I did find the explanation eventually, in O Meien's celebrated endgame book. So my intuition ("can ignore") was right, and I was wrong to doubt it.

But there was an interesting additional take-away from that episode, for me. Being a words guy, I started wondering whether having a name for each kind of move was a way of optimising how to learn by this process. In other words, if a type of move didn't have a name, would I notice/learn it no matter how frequently it occurred? My gut reaction is to say yes, but with the reservation that I might not learn it as thoroughly as when it did have a name. But I still have niggling doubts. It may simply be that when a move type has a name, you can see it talked about in books or you talk about it with other people, and all this talk has the effect of strengthening the associations already existing within the brain. But it has nothing to do directly with the actual frequencies noticed (subliminally) when transcribing. This certainly feels true of the monkey jump.

This seems to be potentially important in the case of studying with AI. If you want to play like an AI, the correct procedure seems not to review your own games with it, or to study pro games with it, but to transcribe moves from a single diagram of games in which AI plays AI. But AI plays moves we find strange, and because they are strange, we don't have a name for them. It seems our instinct is then to give names to them, and the prime example has to be Direct 3-3. But other moves have defied naming so far. Does that matter? Again, my gut reaction is to say yes, but with reservations.

One reservation became clear to me when looking at the book chess book "Game Changer" that dfan mentioned somewhere. He suggested, I think, that this proved that AI can already offer specific ways to improve in chess, and by extension go. My impression, after just an early reading of the book, is that the authors (Sadler and Regan, the latter being a British go player BTW) don't claim that in the manner that we want to take it. In other words, it won't work that way for oldies like most people on this site. It will work, however, but only eventually, for youngsters who come into the game with clean slates of a brain and who learn by putting in the hours replaying AI games instead of the human pro games we grew up with.

I further get the impression that Sadler and Regan believe that this process for the newcomers can be enhanced by identifying now (i.e. naming) the most distinctive and important chess concepts that they think they can detect in AI play (and do note they take as their basis AI versus AI games, not AI versus humans). As with the French Revolution, it is too early to say they are right. But I find their approach interesting, with interesting implications for go.

What S & R do, as dfan mentioned, is to identify core concepts (rather than move types) such as opposite-side castling, or outposts connected with piece mobility, or advancing rook pawns.

This made me aware first, that chess and go differ markedly in how they use names. Move types are not named very often in chess, whereas just about every move type can be named in go (this may contribute to the over-obsession with shape that many go players have). Even when a move type does seem to have a parallel in both games, the associations are quite different. A knight's move in chess is mainly just to do with explaining the rules, but in go it has to do with attacking, waist-cuts, or even (for some of us) to do with disagreements between Shusai and Kitani. Moves types in general have very rich associations in go. In chess the associations seem to derive more from wider concepts.

But I am wondering now, in the AI age, if we should be following chess. For example, we still talk of shoulder hits even though the AI use of them tends to baffle us (and pros). Maybe we should be inventing concept phrases such as (for the old shoulder hit) "creating an early outpost in the centre". In a way, this is what Direct 3-3 does: it stresses the time element rather than the shape element. I would also argue, incidentally, that the ancient Chinese got there first with their "call & response" concept (zhao ying), and I would further posit that amashi is a Japanese equivalent (i.e. an example of a concept rather than a shape, though in this case with the caveat that Go Seigen thought the Chinese genius Huang Longshi was the first master of amashi).

This urge to name does seem universal. I mentioned elsewhere that shogi players learnt to favour early pushes of the side pawns (the rook pawns in chess), something that horrified generations of chess masters but has now become common. The S & R book told me that a British chess master (Simon somebody) has given the pushed king's rook pawn the name Harry (from being on the h file). I think this is a great example of our naming urge, an urge we probably should extend to go, but with an emphasis, like chess, on concepts rather than shapes.

Going back to my intro, what I meant by repetition having a deeper value that is often overlooked in other fields can be illustrated from my own experience in languages. I remember from early schooldays having to learn the names of flowers in French. One that stuck in my mind was glycine = wisteria. I had no idea what a wisteria was (no internet in those days, no bookshop in my home town), and I never found out until I saw Japanese hanafuda playing cards. Since I later discovered that it was named in honour of Caspar Wistar, and so should really be wistaria, it has a firm, but useless, space in my memory.

But when I was reaching Japanese at university, my students were all naval architects (who only wanted to read Japanese, not speak it). The approach I took, therefore, was to handle the grammar with special look-up tables and for the vocabulary they simply translated technical articles in their own specialised fields. This meant that they only ever encountered vocabulary that they wanted in, each term in the exact frequency it occurred in real life. This approach (using translation as an equivalent of transcribing in go) was so efficient that the course took only 49 hours. At the end they could read technical texts quickly and accurately, though with the back-up of a dictionary (Katago?) and the grammar tables.

Would that we could get good results so quickly in go - but maybe we do, already? Is that - playing over pro games - why the likes of Sumire and Fujita Reo got so good so young? After all, when they were five so they could hardly read books or understand pro talk. I remember being with five-year-old Liao Xingwen. He couldn't read much text but he was a whizz with go diagrams, and kept books full of them under his pillow.

[dfan]

(Hopefully Knotwilg won't mind that this discussion is continuing here.) Naming things is incredibly powerful. It's both helpful for memory and also provides a shortcut and automatic generalization in discussion (if I say "shortage of liberties" in discussing a position, a player will both immediately know what I'm talking about and also instantly "page in" their knowledge of how shortage-of-liberties techniques tend to work).

A striking example from another field: all music historians know how mid-18th-century classical music works, of course, and know what sounds idiomatic and what doesn't, but many of the patterns that occur in it were not thoroughly identified and named until quite recently when Robert Gjerdinden published Music in the Galant Style in 2007; he made a big list of typical patterns, often giving them names for the first time, and demonstrated that he could name and classify pretty much any phrase in a 1750-era piece of music. Now "schemata theory" is the standard way to look at music of that period.

In chess, one recent book I enjoyed was Bronznik's Techniques of Positional Play, which takes 45 concepts that strong players know but haven't generally codified, giving them names and providing examples.

One danger with naming concepts is that it is the first step on the dangerous road to thinking about games in English (or Japanese or whatever) rather than go-ese or chess-ese. This is the main topic of Willy Hendriks' interesting chess book Move First, Think Later, which I know John has read because he's written at length about it.

The naming of move types is indeed much more rare in chess than go. There's castling, of course, which is unique enough both in form and function to be treated specially. The closest thing I can think of to "hane"-type names is fianchettoing a bishop. Other names tend to be more about tactics (fork, skewer, etc.).

[xela]

Chess just spends its names in a different place. There are literally thousands of named openings.

[Knotwilg]

I certainly don't mind

I've started replaying pro games while predicting the next move - done that before - and while it's fun, instructive and the kind of activity you can practice intensively with the prospect of becoming better at it, the main question I keep having is what impact such "deliberate practice" has on one's quality of play.

The same question can be asked about tsumego and reviewing your own games (with AI or without or hybrid). There's traditional advice and anecdotal evidence that all of these practices do lead to better playing ability but in fact we still don't really know the extent to which each of those does. I don't know of any scientific study that takes a group of players of various strength, let them perform any of those deliberate practices over a certain period of time, and compare their progress to a control group of players of similar strength distribution.

So we're stuck with our hypotheses, beliefs and some expert advice. One element that I keep coming back on and which is often neglected in expert advice, is the underlying assumption that you take games seriously. Pros do that, because their playing and training environment offers such "seriousness" naturally. We, amateurs, have mostly swapped the original club environment, where games were played physically but opponents of similar strength were scarce, for online play where the latter abound but we tend to play faster games because of the inherent unreliability of anonyous opponents or our network connections, discouraging an honest effort and investment of our precious time. We may play our games when half asleep, or drinking alcohol, or while watching TV ... all but serious match conditions.

I have often talked about my major progress, 2k to 2d in a few months in the early 2000s, and have attributed that to serious game review. But the underlying feature was serious play, on a physical board, without time constraints. Stupid errors way below my level were unlikely because I played every move thoughtfully. Hence the review was more about unknown unknowns than "if it weren't for that self atari".

Hence I still hold the belief that forcing oneself to play serious games, even online, is the key: be fully concentrated, awake, no alcohol, no distractions; major time limits (or none), no blitz; think about every move; don't resign too early and develop some fighting spirit ... Then again, this too is anecdotal evidence, part of my story, not tested in objective research.

[xela]

I'm curious about the "internet generation" of chess players. The prevailing culture on the servers is for very fast games. 3-minute blitz is normal, and 1-minute "bullet chess" is fairly common too. That's 1 minute absolute time for the whole game. Some people will happily play 10- or 15-minute games, but it's next to impossible to get a game at anything like a tournament time control.

Anecdotally, there's a bunch of people who've reached international master strength (similar to 6d amateur in go?) simply from tens of thousands of blitz or bullet games, with no other form of study.

Rather than deliberate practice, it's about throwing patterns at your brain, with a fast feedback loop, to see what sticks. And if I understand correctly, this is more or less what John is advocating when he speaks of playing through pro games quickly.

[John Fairbairn]

this is more or less what John is advocating when he speaks of playing through pro games quickly.

To be pedantic, it's not really me who's advocating this - it's a lot of pros. I don't study go to become strong so I have no experience to fall back on, except with the version of the technique that is replicated by transcribing games (which does, however, seem to confirm this pro view).

There is one common aspect that has not been mentioned yet. Quite a few pros mention playing through all the games of Player X several times (usually the likes of Jowa or Shuei, but Shusaku seems rare, as does Go Seigen, which I find odd). Perhaps, by repetition, they are further reinforcing what they think has been good fertiliser to feed their garden of intuition.

Over the years, I have also been struck by the number of very strong European players I have met who say they have played over the complete works of a favourite player, sometimes more than once. In this case, the name of Go Seigen has cropped up fairly often. Quite a few players can remember lots of specific positions (but not whole games), and I can do this, too. I didn't set out to do - it just happened (after transcribing, but more often after writing about a game, so perhaps that doesn't count).

[Knotwilg]

this is more or less what John is advocating when he speaks of playing through pro games quickly.

To be pedantic, it's not really me who's advocating this - it's a lot of pros. I don't study go to become strong so I have no experience to fall back on, except with the version of the technique that is replicated by transcribing games (which does, however, seem to confirm this pro view).

There is one common aspect that has not been mentioned yet. Quite a few pros mention playing through all the games of Player X several times (usually the likes of Jowa or Shuei, but Shusaku seems rare, as does Go Seigen, which I find odd). Perhaps, by repetition, they are further reinforcing what they think has been good fertiliser to feed their garden of intuition.

Over the years, I have also been struck by the number of very strong European players I have met who say they have played over the complete works of a favourite player, sometimes more than once. In this case, the name of Go Seigen has cropped up fairly often. Quite a few players can remember lots of specific positions (but not whole games), and I can do this, too. I didn't set out to do - it just happened (after transcribing, but more often after writing about a game, so perhaps that doesn't count).

To be ignorant, could you repeat to me what exactly is meant by "transcribing"? Is it taking a paper kifu and digitizing it?

On the other note, I'm also somewhat surprised by the overwhelming practice of Blitz games. I still see that as a sign of the times - not wanting to invest a lot of time in a casual encounters. Yes, it accounts for fast feedback, but surely to understand what's going on beyond the surface, one has to do some studying alongside.

[John Fairbairn]

To be ignorant, could you repeat to me what exactly is meant by "transcribing"? Is it taking a paper kifu and digitizing it?

Do you mean a kifu or a kifu?

Transcribing here is taking a game record on paper and rendering it into sgf form, using a comnputer. It is possible to do this by OCR or other manual means (such as scanning from left to right and filling in an sgf move template). In terms of creating a database, it is possible (indeed, a blessed relief!) to use multiple source diagrams (1-100, 100-200 etc), but in terms of pro advice to play over games to become stronger, they are just referring to recreating the game on a real board, normally using a single paper diagram as a source. The thinking behind using a single diagram is presumably to make the task harder. It's too easy to peep ahead when you are using multiple diagrams.

To go off at a bit of a tangent, you see or hear so often a basic error of thinking among western amateurs which is encompassed in phrases like "to become strong at go you need to do X", to which someone else will respond sternly with "no, to become strong at go you need to do Y". Indeed, what I have noticed over the years is that very many western amateurs will become very stroppy when given advice, seeing this as an insult to the "logical" ways of thinking to which they cleave. We have seen a good example of that in a concurrent thread here. I regard that as a mistaken mindset, too, although I can see that one normally wants to be cautious when taking advice from a fellow amateur than from a pro.

Japanese amateurs are obviously just as prone to wrong mindsets as we are, and it might seem pros are, too, when we see books touted as (in English) "How to get strong at by doing Z". But the mindsets , and so the mistakes, can differ. Japanese does not have a distinct comparative or superlative form for adjectives (e.g. fast, faster, fastest). As a result, the plain form (fast) in Japanese often overlaps with the comparative form (faster), and can even overlap with the superlative. Therefore, a book translated into English as "become strong at go, do Z" really means, to a Japanese, "Become strongER at go..."

The significance of this is that pros are not recommending a single method (as western amateurs tend to do), though of course they will think that one method is better than another. It is my experience that replaying games by past masters is seen a, at least, one of the best methods, to build up intuition. And in essence this is what AI does.

Pros seem to stress speed in doing this. In the Kitani school, children were expected to rise at 6am and play over a game or games from the past until 7am, when they did radio calisthenics. Breakfast was at 7.30. So, replaying was the first task of the day, and took priority over breakfast. Kato used to cheat and look for short games, but he was the Kitani pupil who found it hardest to win titles, so maybe there was a connection there!

The only other regular go study was from 6-9 in the evening and was playing games and studying with each other. It was very rare to play with Kitani himself. He spent a lot of time with the children, but on daily tasks (such as tending the goats), and no doubt offered many words of advice. But one thing he did do, most days, was to set tsumego problems when he came home in the evenings. Speed was stressed here, too. There was a race to get the right answer. Any visitors were roped in too, and in the nature of things they were usually very strong players. But they always ended up being embarrassed by the children. The tactic was to write the answer and show it to Kitani, who secretly marked it right or wrong, but everyone had to continue till they got it right. The duffers might have to do the washing up, or such.

Other go activities included charity visits to hospitals and the like. Little Otake Hideo was famous for his behaviour at these. He played instantly and usually read comics while waiting for his opponent to play. The old men were nonplussed by a diminutive lad who replied before they had released their fingers from the stone (while looking at the game next door). If they thought too long, he would soon get bored and go and look out of the window, or even wander off outside.

This speed of play was never criticised. Otake was gregarious and didn't like to study alone. His main study group was with Tozawa Akinobu and Kitani Reiko. Otake/Tozawa games took half an hour. But Reiko was notorious for very slow play and games with her could take 5 hours. Kitani used to take Otake and Tozawa on his charity trips, but apparently never Reiko!

I have heard pros say that they were taught to play fast because that skill was important when playing with amateurs, or in exhibition type games with other pros but for amateurs.

What I have inferred from all this (and much more, of course) is that pros recommended certain things (replaying games being a/the prime one) to become strongER, but to become just plain strong you need a varied diet. Replaying may be your staple rice, tsumego your meat or fish, and endgame may be for those who like dessert. But, whatever, you eat, a well-balanced diet is important.

But this is for pros. All amateurs, however, are opsimaths, and very often busy ones. It is understandable, therefore, if they choose to study tsumego because they can take a book on the commute to work, or play blitz games online before bed (as opposed when getting up with the lark). But being understandable doesn't make it right. At least if you want to become truly stronge as opposed to just a bit strongER.

[Knotwilg]

Well I might just take that last writing of yours as a good summary for times to remember. I love the linguistic tangent!

Replay professional games, fast, predictive. Play games and review them together. Don't mind the speed too much. Do tsumego that challenge you but try to do them fast.

[dust]

I'm curious about the "internet generation" of chess players. The prevailing culture on the servers is for very fast games. 3-minute blitz is normal, and 1-minute "bullet chess" is fairly common too. That's 1 minute absolute time for the whole game. Some people will happily play 10- or 15-minute games, but it's next to impossible to get a game at anything like a tournament time control.

Anecdotally, there's a bunch of people who've reached international master strength (similar to 6d amateur in go?) simply from tens of thousands of blitz or bullet games, with no other form of study.

Rather than deliberate practice, it's about throwing patterns at your brain, with a fast feedback loop, to see what sticks. And if I understand correctly, this is more or less what John is advocating when he speaks of playing through pro games quickly.

There's also a sizeable amount of people whose grade never goes up despite playing lots of bullet or blitz games.

I really have my doubts that it's possible to play consistently at a higher level much beyond one's own reading ability. In my own personal experience, I may get some nice ideas and shapes from pro games but it's amazing how it only takes a few moves from a player who is much stronger in reading to send me floundering.

[kvasir]

Since this discussion turned to chess, maybe I'll be excused to share one of my favorite videos on the topic but in context of chess.

I don't know why I find this guy entertaining, maybe it is the shared love of Perrier water, but his point is very good and blunt. I'll try to be more blunt, if you want to improve at anything then you need to pick up on what you are doing wrong and fix that. There isn't much more to it. If you make a mistake in a game then you need to stop making that mistake, if the opponent tricks you then you can use the same trick on someone else, and that is that, and if you trick yourself then you also need to stop doing that. Doing tsumego, replaying games from paper and playing actual games does nothing for anyone in terms of playing better if they don't resolve to not make the same mistakes again and again. It is not the activity, it is the attitude!

Also what he said about people not making mistakes because they don't know it is a mistake. Most of the time, in Go too, it is not that we don't know it is a mistake, most of the time it is that we don't realize it in the moment.

I also think it is easy to forget how hard one worked to get to where one is in terms of Go skill. It gets harder so it is natural to not make progress if one doesn't have the memory of working hard before, the lack of reference belies faulty expectations. Anyway, there is nothing wrong with just enjoying instead of being on a quest or doing one's own thing; I think it was Lee Chang-ho who played himself which I think is very silly.

[John Fairbairn]

There's also a sizeable amount of people whose grade never goes up despite playing lots of bullet or blitz games.

I'd go further and say it's the vast majority.

I think what's happening is that they are just filling their brains with amateur shapes and sequences. Garbage in, garbage out.

And next to nobody even records the game results, so there's not even a rudimentary feedback loop.