Life In 19x19

Although the term has been used in some of the discussions on improvement, I find that there is a lack of records of "deliberate practice" or perhaps a lack of such practice altogether. When I browse through the study journals, I will find a lot of

* game analysis through self review or requested review (which is great)
* tsumego (which is universally agreed to be necessary)
* pro game study (which I find not very efficient)
* technical discussions (mostly on opening)

What I seldom see is a game review request with "In the following 10 games I tried to get into the endgame with not too high a disadvantage and then apply good endgame techniques to clinch or reverse the game".

Or, "I want to improve in ko fights, so I set up some games to play a ko whenever an opportunity came up. Can you tell me (a) if I fought them well, and (b) if I should have started the kos in the first place"

Or ... you can invent your own deliberate practice.

Go players seem to think that studying tsumego or replaying pro games or acquiring some knowledge will magically integrate into their games, which it probably does. However, I think that studying one subject at a time and forcefully integrating it into one's game, to see the concept at work, is the fastest way to improve. The idea includes self review, which is the best study method even for games where trying to win.

Deliberate practice and integration is also a great antidote for fear of losing. If you disregard the end result and concentrate the feedback around the implementation of an acquired idea, then it will be much easier to play for improvement and not for victory.

Does anyone have experience with such learning? Or is go simply too complex to concentrate on one technique/concept at a time?

I agree, for now, I try to always play a sanrensei when I'm black... and to test what I read in "handicap go vol. 7" and "patterns of the sanrensei"
and sometimes, it even payoff... I just got my first victory ever against Igowin at 3k... (but It still win most of the time...)

Knotwilg wrote:Although the term has been used in some of the discussions on improvement, I find that there is a lack of records of "deliberate practice" or perhaps a lack of such practice altogether. When I browse through the study journals, I will find a lot of

* game analysis through self review or requested review (which is great)
* tsumego (which is universally agreed to be necessary)
* pro game study (which I find not very efficient)
* technical discussions (mostly on opening)

What I seldom see is a game review request with "In the following 10 games I tried to get into the endgame with not too high a disadvantage and then apply good endgame techniques to clinch or reverse the game".

Or, "I want to improve in ko fights, so I set up some games to play a ko whenever an opportunity came up. Can you tell me (a) if I fought them well, and (b) if I should have started the kos in the first place"

Or ... you can invent your own deliberate practice.

Go players seem to think that studying tsumego or replaying pro games or acquiring some knowledge will magically integrate into their games, which it probably does. However, I think that studying one subject at a time and forcefully integrating it into one's game, to see the concept at work, is the fastest way to improve. The idea includes self review, which is the best study method even for games where trying to win.

Deliberate practice and integration is also a great antidote for fear of losing. If you disregard the end result and concentrate the feedback around the implementation of an acquired idea, then it will be much easier to play for improvement and not for victory.

Does anyone have experience with such learning? Or is go simply too complex to concentrate on one technique/concept at a time?

I've done this on occasion, as well for the tsumego practice (like: I'll try not to fall for stupid corner kos.) In some sense, most of the stuff I'm pouring into Anki is a form of deliberate practice, since part of it is focused on mistakes I make on my games (as pointed by teachers.) So I'm trying to distill the cause and remove it from the root.

Maybe I got it wrong, but my understanding of deliberate practice is not just consciously focusing solely on one thing.

As far as I know, deliberate practice includes:

• focussing on your weaknesses rather than your strengths,
• aiming to perform at expert level and
• recieving immediate feedback of how well you did compared to the expert

There are multiple ways to do this in Go but it's not so easy to do them right.

You can play games and immediately get a review from an expert Go player to compare both your ideas and reasonings.
In my opinion, the problem here lies in the vastness of Go and the possibility to repeat the exact same positions to see if you learnt something.
Artificially created positions as in problems might help but in the end, you also have to understand the moves leading to and following the position.
A second problem might be the "best/only move"-issue. Rarely there is only one best move, so how can you aim to perform at expert level when experts play differently?
The best approach I can come up with is only using one opening for each colour and only replay/review pro games using this opening (to see how the expert reacted to different situations, although mostly the why will elude the average Go player) or better if you have an expert Go player at hand to let him/her immediately review the game. You are still facing the "best/only move"-issue but the better the teacher (the expert Go player) the better will he/she be able to explain or show different strategies.
And since you are limiting yourself to only one opening, the same positions will occur more frequentely.
It is also a good approach to play a ton of games (plus immediate review from the expert) to battle the vastness of Go : D

You can solve problems with only one answer (best play for both sides).
The easiest approach and arguably the best to do on your own.
Given that the problems are not ambiguously designed, so there's only one best play for both sides (equals the play of the expert Go player), you can both aim to solve the problem, including every relevant variation and check your times to see if you are getting faster.
I think you should take a big enough sample size to not get into the situation of just memorizing the problems, although I also believe that some basic shapes are worth memorizing after they are solvable.
The problem here is the - in my opinion - limitation of suitable problems. In the end you can solve Life-and-Death, Tesuji and endgame problems (localized problems), but e.g. opening problems may become victims to certain fashions or new inventions, the same might be true for other whole-board-problems. So, improvement is only possible in these three areas (and reading in general but that is hard to quantify), which falls a bit short to the - again - vastness of Go.
(By the way I really liked the approach Michael Redmond took concerning the endgame: He selected castle games because in those times the players had sufficient time to play out the endgame in the best way possible and tried to reach the same score starting some douzen moves short of the end of the game.)

You can replay/review/memorize professional games to learn how they coped with certain situations.
The problem, which immediately pops up, is the lack of immediate feedback and focus. Again Go is a vast field, so just studying everything and hoping it will fall in the right places seems a bit naive. It's also questionable how you can compare yourself to the expert.
Guessing the next move has the limitations of (arguably) not knowing how good or bad the not-played move is (or if it was a branching point with multiple sound candidate moves). It has also the problem of not knowing when the expert player played a subpar move (one player lost, du'h ; ) ). So in everything you do, you stay in the dark when it comes to comparing yourself to expert player performances.
One good approach - in my opinion - is focussing on one feature of the professional game. See above: the opening, combined with also limiting your play to this opening. You can argue that most pro games are not decided in the opening, so subpar moves might occur less frequently. Although branching points might still exist but here they should be manageable if your sample size of games with just one opening is large enough.

Off the top of my head these are the methods for deliberate practice I can come up with. Maybe there are more or maybe I understood the concept completely wrong =)

I couldn't agree more. I think Knotwilg is right in saying this is not the preferred way of studying and 'playing-to-improve' in most go communities. Which might be considered weird, as it is so prevalent in other, similar games. I believe it's one of the main practice tools for people playing RTS (real-time strategy) video games like e.g. StarCraft 2, which is where I spent a lot of time doing this. I remember grinding games where all I would focus on was one single aspect of the game, not even bothering with the rest. It had some remarkable results. And even when at first it didn't, I really liked the focus it provided for my practice. I knew what I was doing and where I wanted to go with my practice.

I can absolutely second that it works wonders for losing anxiety. This in turn helped me improve faster. As a nice by-product the one aspect you were focusing on would sometimes improve so rapidly, you could win games off just that. For go this would be to be so far ahead after playing a certain opening you've mastered, so that your opponent could not catch up for the rest of the game, or play so well in life and death situations that you can win even when horribly behind after the opening.

I think one thing a lot of go players do - which might easily turn into deliberate practice if they review the results afterwards - is playing the same opening as black for many games straight, then when they feel they have mastered it, move on to the next.

Thanks for the idea. I shall try digging up my practice notes from StarCraft and see if I can't craft a similar training schedule for go.

"Playing the same opening" seems to be the common implementation of the idea I described. I think it is a mistaken idea that recreating similar board positions be a necessary condition for focused learning. The examples I gave were:

- in the next game(s) I will create and manage ko fights
- (...) leave weak groups/surround opponent groups, and live/kill

There are more. Since people inevitably return to "playing the same opening", I'll list some ideas

- in the next game(s) I will only concentrate on keeping the score
- (...) counting liberties of (weak) groups
- (...) leave my groups disconnected, then find connection tesujis
- make all groups strong, except for one weak group, which I will manage
- play greedy and invade more than I would usually do
- play less greedy and reduce whenever I feel the urge to invade
....

In fact you can take almost any concept and make it a focal point. You could combine 2,3,4 such concepts, if you have not many opportunities to play. This was in fact Tamsin Jones' Compass idea.

Of course, after having trained one aspect of the game, it should stick for a long time, until it's time again to improve on it.

As to SoDeSuNe's reply, I concur with your definition. What I mostly challenge is the idea that Go would be so complex that it can't be done. When developing a complex skill, the situation (board position) will always be different. The variety of input is precisely what helps us developing the skill. The focus lies on exercising the same technique (ko, endgame, L&D) in different conditions (games), not on recreating similar conditions (openings).

Knotwilg wrote:"Playing the same opening" seems to be the common implementation of the idea I described. I think it is a mistaken idea that recreating similar board positions be a necessary condition for focused learning. The examples I gave were:

- in the next game(s) I will create and manage ko fights
- (...) leave weak groups/surround opponent groups, and live/kill

There are more. Since people inevitably return to "playing the same opening", I'll list some ideas

- in the next game(s) I will only concentrate on keeping the score
- (...) counting liberties of (weak) groups
- (...) leave my groups disconnected, then find connection tesujis
- make all groups strong, except for one weak group, which I will manage
- play greedy and invade more than I would usually do
- play less greedy and reduce whenever I feel the urge to invade
....

In fact you can take almost any concept and make it a focal point. You could combine 2,3,4 such concepts, if you have not many opportunities to play. This was in fact Tamsin Jones' Compass idea.

Of course, after having trained one aspect of the game, it should stick for a long time, until it's time again to improve on it.

As to SoDeSuNe's reply, I concur with your definition. What I mostly challenge is the idea that Go would be so complex that it can't be done. When developing a complex skill, the situation (board position) will always be different. The variety of input is precisely what helps us developing the skill. The focus lies on exercising the same technique (ko, endgame, L&D) in different conditions (games), not on recreating similar conditions (openings).

Hmmm I should read The Talent Code to be sure (have only read The Little Book of Talent, so the references are thinner,) but playing a full game of go is part of the "soft skills" (high flexibility, like a soccer player improvising a dribble or pass) and not of the hard skills. Hard skills are the ones benefitted most from deliberate practice. Daniel Coyle defines them succinctly as:

• Hard skills are about ABC: Always Be Consistent
• Soft skills are about RRR: Reading, Recognizing, Reacting

Of course, in very complex endeavours both get thoroughly mixed (no matter how good you are at spotting a pass if you can't execute it perfectly) but as such, for deliberate practice to work it should focus in hard skills, not soft skills. According to Coyle (who, again, researched quite a bit the literature and interviewed relevant people, like the same K.A.E.) for soft skills, you should (copying the section name from TLBoT) "Play like a skateboarder." In go, this can be accomplished in several ways, actually I have pending trying one: playing the first N moves of the game at random (not on first line, though) and then playing from there.

I think that deliberate practice sounds like a great idea, but there are a number of factors that work against sticking with a single goal for any length of time. Most of the factors that work against it aren't unique to playing go, though. They just reflect the difficulty of incorporating intentional practice into a leisure activity.

Here are some of the excuses that keep me from more frequent deliberate practice:

• As a mid-kyu player, it can be difficult to choose meaningful goals without the input of a stronger player.
• When a proper goal is chosen, it is hard (though not impossible) to measure progress without the same stronger player providing consistent input.
• It's difficult to stick with a single goal when so many weaknesses abound! (And I keep finding that most of those weaknesses reduce to a deficiency in reading ability rather than particular strategic lapses, so I keep being driven back to tsumego.)
• It can also be difficult to manage the flow of the game well enough to find an opportunity to meaningfully work on a certain goal.
• Finally, it's difficult to enjoy the game when you get constantly obliterated.

Still, I have found some meaningful sub-goals to incorporate into my games. I don't know that I've dedicated myself to one with enough intentionality to deserve the label deliberate practice, but here are a few items for kyu player consideration.

• Focus on making intentional choices for each of your opening moves rather than just playing a few corner enclosures out of habit. Compare your choices to known joseki/fuseki for evaluation.
• Commit to playing a pincer (or other type of move) during joseki whenever it's appropriate. Again, known joseki/fuseki can be used to evaluate your choice of moves.
• Intentionally try to settle groups in as few moves as possible and then use sente to make a big move elsewhere. Warning: Until your reading skills improve, this will lead to the loss of large groups when you play one move too few. (This is my current goal... I think it's good because it provides motivation for improving my reading while also paying attention to the larger flow of the game, but it can definitely be discouraging when you get it wrong.)
• You already mentioned this, but attempting an invasion where I might normally let my opponent get a moyo is a healthy practice for those of us when can tend to play passively.
• Focus on making endgame moves in the proper order and NOT making unnecessary defensive moves inside your own territory. This suddenly became important when I lost a game by half a point.

Perhaps the most difficult piece in all of this is determining whether we were successful at a particular goal. I guess this is where you would expect to see a directed request for analysis, but unless you're getting input from the same person each time it can be hard to use the feedback effectively. It's also seems easier to get general feedback than specific feedback when your whole game is weak. I think this is probably one of the strongest arguments for getting a teacher: they can help you choose proper goals and evaluate them well.

Hi Jeromie,

I agree with your remarks. It is indeed challenging to find positive feedback for your deliberate practice in absence of a teacher. Most players make poor choices for their focal concept in that respect. They will study and practice a particular opening, while the impact on the result of that focus is zero. It can be great for learning purposes but there is no emotional reward.

This is why we should think about which concepts have the most dramatic impact on winning percentage. The answer may be as trivial as it is surprising. Here are 3 concepts one can implement immediately with great potential impact on winning percentage:

1. don't resign
2. don't lose on time (play any move before your flag falls, keep some time in reserve for the big moments)
3. play in good conditions: not too blitz, not too drunk, not too deprived from sleep, no 20 games in a row

http://senseis.xmp.net/?DieterVerhofsta ... mesmanship

To anyone who wants to boost their game in absence of a teacher, I'd recommend trying this for the first 10 games and stick with that habit afterwards. N° 1 may sound like a lack of sportsmanship but I think it is good to develop fighting spirit first and become a gentleman later. N° 3 is all about improving concentration. We all know it's the blunders that destroy our games, not the inappropriate dealing with the micro-chinese fuseki. N°2 is a no brainer but it may require a good internet connection.

You mention "reading" as a main issue in your games and indeed, this would be my next major focus. But tsumego is too often treated as a magical cure. There are more methodical ways to learn how to read.

http://senseis.xmp.net/?HowToRead
http://senseis.xmp.net/?CountingLiberti ... uringRaces

Finally, your "Intentionally try to settle groups in as few moves as possible and then use sente to make a big move elsewhere." sounds like a great "intermediate level" goal. Good luck!

Dieter

I certainly do this, though slightly less structured. I haven't set out in a series of games to create as many ko's as I can, but I have set out intending to read out who is ahead in ko threats before starting a ko, or to make sure to look for ko and seki solutions in life and death on the board.

I find it just a natural way to play. I'll have a sizeable lead I'll squander in an unnecessary fight, and then for the next week as soon as I pull ahead in a game my mind will start warning me not to get involved in fights I don't have to. Or at one point I lost multiple groups in a short span of time to a connection on the first line, and for a week or two I was looking very actively for those, and now it's more or less automatic.

The idea of choosing what to focus on, and forcing boards into relevant positions sounds powerful. But personally, I find I do well attacking whatever mistake is losing me games lately. You mention using the approach to avoid a fear of losing, but that's actually roughly the emotion that drives me to improve