After 2 weeks of solving only mate in 2 problems i discovered 2 things. For one I'm getting faster in solving , still not at fast as I want but for now after thousands of problems I 'see' about half of the problems. this proves that pattern recognision 'works'. I blindsighted a 1800+ player with a stunning mate combi. Kinda cool but also there was a dark cloud on my chess horizon...
Well all the tactical problems are on the farside of the board it's always 'you win' but never 'your opponent will win' well I get it it's depressing to loose but also very very nessesary because pattern recognision doesn't discern between friend or foe.
As it is right now i'm tactical blind for all the beautiful combinations my opponent can mate. I mean I can calculate the danger but I just don't see the danger because I have never trained any positions/patterns in which my opponent would make me. Well I'm repeating myself but that's the great blindspot in learning tactics I think. And after loosing many games with simple mate's my 1800+ opponent had proven to me that Tactics simply don't address the issue. Well I'm not alone because everyone on Chess.com has the same problem!
With that in mind I created a database with positions and started to learn/recognize the 'my opponent will mate me' puzzles. A simple concept but why do people not see it ? Well I do and I would advise everyone to do the same. For now it's too soon to tell if it has any effect. But I'm sure that the next week it will kick in ! Because pattern recognision when learned with many many puzzles is way faster than expected (at least for checkmate). I'm still in the very beginning of my 'GM training' but I'm already sure that it will be much faster than the 'old' method.
Checkmate patterns how many ?
Chess is a very complex game but the number of mate in 2 problems is finite. When generalization kicks in , and humans are way better in that than computers (that's why captcha's can only be read by humans). The number of possibility's in mate patterns are far less than expected. What this means is that generalization is the key ! Generalization is connected to language processing and that's why I started to make many associations on chess positions. Part of the thinking process is 1 'learn' patterns by seeing them very often. 2 if patterns are learnt give them word associations. 3 try NOT to remember the patterns completely but partially remember the patterns but rereading the word associations. In that way word associations are a kind of hash, a trigger that will bring processes into shortterm memory that will solve the problem.
Well there are many checkmates as said and now I'm working on a program that will find them for a limited set. If a pattern is found the set will mutated and stockfish will verify if it's still mate , if so the pattern will be shown... basically it's a checkmate generator nothing more or less... with the possibility to translate mutate and mirror positions.
another thing I learned is that it's very good to learn almost mate patterns. That is patterns in which there is only one move for the opponent/or myself. Almost alway an escape mate will lead to a mate after 1 or more ply
The same is true for stale positions. We all know mate patterns , but stale positions are at least as important.
On the average I can do a mate in 2 very quick. But I also often fail. I analysed why and came to the conclusion that the complexity of the mate is the factor that predicts the chance that I will fail/succeed. Let me explain. After every move there is a mutation on the number of fields and moves on the board. If that number is high , and also the number of penned pieces at deeper ply I have a great chance in failing. Short term memory makes more mistakes and that is also the reason why it's more difficult to mate in the center of the board. I'm working on an algorithm that will sort the mates based on complexity and will drill them.
more to come....