Last 1st of December I had the impromptu decision (well, the idea had been brewing for a while but I just went to execute it, more details at the end) of starting to memorise pro games (the first ~100 moves only), of course, I chose Go Seigen for obvious reasons on that day (having 3 books on analysed games and his complete game collection also weights in.) To do so, I wrote a small script taking advantage of my "sgfrender" utility which can add labels, so I know who played against who. It splits a game into 10-move chunks, which I then use to create Anki cards like this:

Front:



Back:



The code, in case you are interested (and have a compiled copy of sgfrender, which you probably don't)



I selected all Go Seigen games in my computer databse (which, I have to confess, I don't know where I got from, so I find it hard to know how they got sorted afterwards,) sorted the files by name (because, why not?) and started creating cards.

I start properly memorising at move 10 (since the first few moves are more or less "meaningless" and too hard to place without context, now I'm finding it hard starting at 10, too.) 10 cards, comprising moves 10-110. I memorise them 2 at a time (so, I set 2 new cards/day and "custom-study" 2 cards until I'm done with a game,) I've found this the easiest way: any more than 2 cards at a time is incredibly time consuming. On first seeing the card I just go mentally through the moves I see on the front, and then just press to show the back of the card. I imagine the moves appearing on the front card from the back card, then mark the card as "wrong" to see it again. Once I have seen the card once, I try to visualise the next 10 moves (which should be on the back of the card.) Once I'm confident on the next 10 moves (or I'm sure I don't know where they go) I press to check. If it is a card I've seen many times, I try to visualise more than 10 moves to sharpen the skill. Also good is trying to answer where a previous (imagined) move was respect the current imagined move,

I also go through these games with the "guess" function of SmartGo Kifu (usually I score 90%+ with a 10% near,) to also keep a sense for the flow of the whole game in one chunk.

At the current point I have memorised (well, I have probably forgotten a little, but Anki takes care of this) 11 games, so, 1100 moves already. The first days I started I did 1-game per day, but when the 5th day came my mind was pretty mushy and I stopped for 3 days. So I decided aiming for around 4 games a week should be fine. I don't have a specific goal of how many games I want to memorise or when I will stop. For now I'm happy with how I'm doing.

Why did I start doing this?

Well, I was looking for some go-related task which was low-effort (something I could do while unwinding) & low-yield (which well, won't make me a super-strong dan, but will improve a little my go knowledge). In an ideal world, I'd look for a low-effort & high yield, but well, that seems more like dreaming. Just replaying pro games seems pretty close to low-effort, low yield, but I thought memorising the games would result in a higher yield than just mindlessly pressing "next." After these 1100 moves, though, I've found that memorising pro games, even if it can be easily done while watching something on TV, it is quite effortful. And the jury is still out on whether it is low yield or zero yield, but we'll see in the future.

If there is some interest in joining, I could share the Anki deck .

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Yes, please.

_________________
be immersed

Tadaa

So far all added games are from around 1967 (1968-09-00b, 1968-07-00b, 1968-05-00b, 1968-05-00a, 1968-01-00a, 1967-12-07a, 1967-09-20a, 1967-05-29a, 1967-02-17a, 1967-01-06b), except the game with Sato Sunao, which is from 1945 (1945-01-13a). Make sure when you add the deck to set it to "show cards in the order they were added," or use the desktop version to suspend all cards and then un-suspend one game (10 cards) every time you want to learn a new game.

Also, sgfrender (I don't know if inherited from sgf2png) has a bug with ko captures, and recapturing leaves a stone in the ko place. This can make some games (specially given Go's love for ko's!) slightly tricky. Another reason to give the games some SmartGo Kifu or GoGoD playthroughs to feel the real rhythm of the game.

I'm not sure if when I add cards to the deck it gets shared automatically, I assume it does. If so, expect more or less a new game every 1-2 days (I aim for 4 a week as I said, this is roughly 1 every two days, depending on when I take the break)

Oh, shared decks need to be updated manually. Then I'll do it once a week or every two weeks, not sure.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

When you memorize long sequences, the middle is the hardest to recall. IOW, there are primacy and recency effects. Go is peculiar in that stones on the board give clues to previous play. So maybe memorizing the game sequences in this order would be good: Last, First, Middle.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Yes, I remember your post about "starting at the end." Indeed, once you know a "posterior" determining the moves of a sequence is easier (since you have mental hints,) but somehow adding the cards in that order would break my own order. It's surprisingly easy to skip a card without realising, even if just counting from 1-10

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Well, eventually you want to show the cards in random order, right?

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Yes, the problem is not skipping one!

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Awesome, thanks for sharing the deck! Keep us apprised of how it goes/any results you see.

I usually play the game on the board from a book and then replay the game from memory as far as I remember.

I find this the easiest, as well as the most useful.
and I don't care that I've forgotten the games from previous weeks. One at a time, remember it for a while, then let it go.

Downloaded the deck and so far it seems great. I'd been thinking about doing something like this for a while but always got stuck on how to make the cards.



To me this isn't really about effective study. I Like go a lot but I go long periods where I feel like I just want to dabble without committing serious time to it, and it's frustrating to have every game/book/game record remind me how out of practice I am. An anki deck provides me with enjoyable bitesize go tasks and quantifiable progress. When I do commit to studying a game I usually take the same approach as you, but I don't think this is intended to replace that.

First, I'm not convinced that memorising pro games helps with anything other than memorising pro games. I'd have thought more value is to be had from either memorising pro fusekis (which will fortuitously include some josekis) or, for the middle games, memorising next-move positions (and the next move of course).

Also, a more useful task would seem to be to practise looking at a full-board position and then hiding it and recreating it from memory (this may help with 'chunking', e.g. exposing those parts of the position which you don't see as chunks and should).

But if you feel a whole game is what you really want, e.g. as motivation, isn't the 1-10/1-100 method inherently flawed? It's as if you are learning a poem in weird chunks such as "I wandered lonely as a cloud when" "all at once I a saw a crowd, a host of" then you omit "golden daffodils" because you have reached your self-imposed letter limit. At the very least you have to create cards of variable numbers of moves which each form a connected, flowing unit.

Another tip in that connection is to say the names of the moves as you play them. Apart from the memory-reinforcing effect, this will throw up many common sequences, some of which you may not be fully aware of - there are rather more than just hane, tsugi and atari, connect.

Also, if you concentrate on one player, you might find it beneficial to limit yourself for a while to fusekis of the same colour by that player. This may make it easier to get a feel for his style.

Or even recreating it without hiding it is still incredibly hard to do (at least at my current level). I want to practice that some more, as I think it will greatly help with my game recording/broadcasting duties.

I'm not sure either, but wanted to try and see what benefits I may or may not get. Next move positions are hard to pinpoint, though: was "that" tenuki a "best next move" kind of play or did he just decide to switch sides "just because"?

So far (I did it for ~30 days and then stopped because I kept messing the games up... just a couple of days ago I paused all games except one so I can get back on track) I've seen a subjective improvement in my visualisation (of course, if each day I did visualisation practice of chunks of 10 moves, it would be weird if it didn't improve some) but not much more.


I agree that this can be very useful. But keep in mind this "experiment" tried to do without a board or anything similar (just my brain and my iPad, so I could do it easily on commutes or anywhere), so it's a task requiring other tools (a board, or at least, two go apps at the same time, one with the original position and the other with my recreated board).

It is. But I didn't want to handle all idiosyncrasies of the games and do it by hand, but wanted something that could be done fast and efficient (generating images like this is not that straightforward anyway.) Thou still unravished bri... what?

Hm, I may be missing something I wrote some time ago.


Yup, that's right. On the other hand, it makes "similar looking games" really, really hard to differentiate.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

I had a bout of sort-of memorizing pro games as I was improving from 4-dan.

At the time I was studying games by playing them out on the board and studying the interesting bits. Sometimes this took more than one day, so I'd end up having to rebuild the game position a few times. As I did so, I found that some of the moves that confused me yesterday became much more clear because I had seen ahead in the game--a strange move was actually a probe that left good aji, he didn't defend reduce that moyo because he already had a tesuji to get in, etc.

Finding this was useful I ended up playing the game through several times to get a sense of the game flow before sitting down for real study. It was very useful and had the effect of committing the moves to memory for a good chunk of the game.

As an aside, though, if you're looking to start memorizing games from a particular player you might find someone a little easier than Go Seigen. His games were sharp and, to me, confusing. Takagawa comes to mind as someone who plays a more serene game, and I'm sure John might have a few other ideas.

yes, that's how I tend to study as well... I'll play through the whole game a few times, essentially committing the game to memory, before going back to read the commentary in the book. It makes reading the commentary much easier to understand, and easier to follow, and also easier to follow the alternate variations. And often for the very same reasons... you already know where and how the game actually played out.