Fllecha wrote:
Finally feel free to comment my daily schedule to emprove (I am a math teacher, so I can't play full time go)
1) play a serious game against strong computer no handicap.
2) do at least 5 tsumego
3) Study 1 professional game trying to understand moves.
1) Not necessarily the best way to use a computer opponent. Especially with a program using a MCTS evaluator because you can't really determine that you are getting better, making fewer mistakes, just because the margin of loss decreases.
Ideally you want the difference to be such that you win one out of every four or five games. In other words, set the program as strong as you can and take the number of handicap stones that result in that, and then when winning one in every three or so, take a stone less. If you are bothered about being too unlike an even game, you can choose a weaker level for the computer so you are taking just 3 stones (the range 2-4 stones is a much more open board than high handicap games and at 2-3 stones you are going to learn about joseki).
The reason you need to have the effective difference such that you usually lose but sometimes win is that you are trying to learn the difference, trying to learn not to make a particular mistake and trying to unlearn bad habits. You can't do this by trying to reduce the margin of loss since a MCTS evaluator is going to play the same way a match (two boat) sailboat race is sailed -- the boat in the lead aims to stay between the boat behind and the next mark to guarantee staying ahead, not necessarily best to maintain the margin of the lead. In other words, just because you are consistently losing against a program of this sort by ten points does not mean you will win by taking another handicap stone. You might still find yourself losing by almost ten points and again the same taking even another stone.
Look, you say you are a math teacher? (long ago I was too) How do you set problems for your students? << what difficulty of problems to the current state of their knowledge leads to the fastest learning? >>
2) A good idea. Best if you have a program serving you up the problems that is keeping track of which you got right and which wrong (and serving the latter up to you again, etc.). Best if you have a program with various levels of problems available so you can use a level where you can fairly quickly get about 75% correct (
not just a correct first move, since knowing that it is a problem certain principles might let you know "if there is a solution, this must be the first move").
But I suggest you actually work with two levels of problems. One level easy enough to get right 90% of the time where the problem question is "what is the status?" ( in other words, not "black to live" but "can black live or get ko?"). That's the level where learning tsumego can transfer to play because in a game nobody is telling you "this is a tsumego".
3) I think this depends entirely on your current level. I would say looking at well commented games that are between players maybe a half dozen or so ranks above you would be more useful since those are the mistakes and bad habits you are trying to learn to avoid.
is not very common "ranked choice 21" and is more ofen approached at "a"
) seems to be a nice choice here as you allread have the marked black stone.