Comments on https://lichess.org/@/hgabor/blog/effective-training-methods-doing-tactics/5tyH0JJg
Let's start saying that i like your posts quite a lot, however i think that this one is kinda biased. You are stating your opinion like it is the truth, while it's not. You like the way you train, or at least it works for you, so you belittle different trainig methods...
if you want to be serious about it, you should try conducting some sort of experiment, like having a few students with similar ratings, train some in your way, and some in other ways (same time spent obviously) and track IF your method is more effective.
-
That will be a very interesting read!.
I totally disagree. more difficult puzzles will not improve your ability to calculate better.
It is exactly the opposite: you need to do a lot of easy puzzles, which do have clear patterns.
small patterns are the building blocks of more difficult puzzles, which consist of mixed patterns, where 2-3 patterns are intervined together.
The biggest bang for time invested for me was actually doing attack/defense/check training, which you can find in the GUI if chessbase. The GUI is the "graphical unit interface". The chess program is the "engine", but the menue is the GUI.
If you have a commercial chessprogram like Fritz, Rybka, Houdini, maybe komodo - then they are usually shipped with this chessbase GUI. It does not need to be a new version, the attack/defense/check training was already shipped more than a decade ago.
the attack trainig presents you with a random position, and you need to click on all pieces that you can attack (caputre). Even "stupid" captures, like queen takes pawn (which would lose the queen, though).
And this for both sides. When you found the last possibility of an attack, a new random position is presented.
At first I saw around 12 attacks per minute, but after 30 minutes of training I was already at more than 20 attacks per minute.
I kept going and my total training time was maybe 3-4 hours, I reached 50++ attacks per minute.
And it directly translated into my chessgames: I was much more aware what my pieces can do, the pressure they excert the squares they cover.
If you can only see 12 possible attacks per minute - no wonder you can not be great and fast in tactics.
Defense training: find all "hanging" (undefended) pieces in a given position. Same story, I started with 10 undefended pieces and grew to more than 50 per minute.
These are not even real tactics, or mini-mini-tactics. Very easy, and super effective.
Then I trained 200-300 fork puzzles. Only forks, and for me easy range. I repeated this set with spaced repetition, and guess what? I saw many more forks in my games, I could feel the effect of my training.
Overall, I started training when I was 37 years old, was rated 1812.
3-4 years later, I had a performance of 2200 several times in various tournaments.
But strange as it is: my bullet and blitz did not improve as much like my otb. Message me privatly, I can show you my real name and rating and you can see proof.
I trained at chesstempo as a gold member, which enables you to filter for certain themes (patterns) and rating ranges and you can do spaced repetition with your created sets.
Fazit: not harder puzzles, but instead, simplify/ break down complexity in chess, so you can learn is more easily. You should not learn as difficult as possible, but as easy as possible. Like you do with EVERYTHING else in learning, too?
How you learn a language? you start from simple and do difficult in the final chapters.
You only get fluent in a foreign language if you repeat to learn high frequency words. You wont get far if you start learning "desoxyrybonucleic acid", no matter if you repeat to learn it 100 times. Its not that useful in everyday conversation, thats why.
I like to mix easy and difficult tactics. Fast and easy tactics to train pattern recognition and hard tactics to train calculation.
In general I like your advice on doing difficult puzzles and what you suggested in that section and the pitfalls with guessing or optimizing for quick solving time. Also what you said about using the wrong metrics for difficult tactics. Targeting one's tactical weaknesses specifically sounds very useful as well. I like all of that.
At the same time I also strongly disagree with two points in the blog post.
1) Solving difficult tactics (difficult relative to one's level) is essential for calculation training, agreed. Yet training pattern recognition is incredibly important as well and for this I find it very helpful to do simple tactics over and over again. That doesn't train the calculation process (for that it might be even harmful), but I believe it helps with an important aspect of it.
I may be wrong, but I think it would be a bad approach to just learn the patterns (without drilling them until you know them in your sleep) and from there just do complex calculation training.
Pattern recognition is very important and to recognize a simple tactical pattern in 10 seconds instead of one minute is a valuable improvement, same from ten seconds to five or from five seconds to one or less.
It is helpful for the calculation process as well because it makes it much easier to calculate forcing lines (if you miss a forcing line in your candidate move tree it could be fatal for the decision, even if your overall process was perfect with perfect visualization).
Quick solving times are not important and can even be harmful for difficult puzzles in calculation training, but for pattern recognition training they are essential.
2) Timing difficult puzzles I believe can be very useful for every tournament player. There is a certain time you usually have available during tournament games when you do a deep think in a critical position given the time control. I think it's helpful to get used to solve tactics within that time, even if you haven't reached the final truth of the position but still need to come to a decision.
I feel that is very useful in order to train yourself to come to a final decision in that relative short duration. This helps with the process during real tournament games.
That's not to say that it cannot be useful to take more time or even just do untimed puzzles or solve studies over days or to try and really get to the bottom of a position or a complicated tactical situation. I think that can be very useful and even kind of enjoyable if you arrive at a result after hard work.
@Munich said in #3:
We indeed disagree on that one.
I think you are mixing up studying and training a little bit. Also, the things you describe as your training is more beginner or early improver level stuff. Yes, it is really useful at that stage, but there is a point when you need to move on. Where I find my students completely stuck is at a somewhat higher level, and most of the times they have been doing the very easy puzzles you describe.
By the way, when you start training the first time in your life, almost any kind of training feels very effective. Even suboptimal methods can bring impressive results, compared to doing nothing. Later you need to be a bit more selective about your methods.
you have to create a course think so
@svensp said in #5:
"Yet training pattern recognition is incredibly important as well and for this I find it very helpful to do simple tactics over and over again."
I agree, this is incredibly important - for beginners and early improvers. This is how they learn the basics. Later they need to move on, otherwise they get stuck.
"Timing difficult puzzles I believe can be very useful for every tournament player."
"I feel that is very useful in order to train yourself to come to a final decision in that relative short duration. This helps with the process during real tournament games."
Yes, you absolutely need to make educated guesses in a tournament game. Especially when you are in time trouble. My problem is that people who do that regularly with puzzles can't really do anything else. They try to guess their way through a tournament game, because they haven't learnt how to calculate lines properly. They have a completely broken thinking process. They pretend to calculate for a while, and then they just make a guess. (Typically it is their very first idea.)
My guitar teacher used to say that if you play a piece with the wrong technique, you should slow it down, learn to play it properly, and then try to speed it up again. Most amateurs speed up the wrong thinking process with timing.
No, I am not really beginner level. with 1800 rating you are rather one of the stronger club players. With 2000++ you are expert player.
Just last season, meanwhile 51 years old, I still managed 2219 performance.
Before training I was just at the brink of 1800, so between B-class and A-class.
However, tactics on its own - I actually did not improve THAT much. I guess 150 rating points is due to this, the other 200 points or so I gained by better openings (and middle game systems), better time management, better endgame skill,
The non-tactic stuff:
openings: I looked into the chesstempo-DB and looked which moves are statistically promissing. I found some opening patterns, which helps me to find my way in openings even if I am out of book.
Example: 1.Nf3 is the least amount to learn, yet it is one of the strongest white moves you can play - statistically.
if black plays at some point ...e6 then fianchetto g2-g3 is almost always the top promising opening choice.
As white you do well if it is a rather symmetrical position, while asymetry is statistically better for black. so no wonder the sicilian (1.e4 c5) is stronger than (1.e4 e5). And indeed the petrov has one of the worst statistics for black.
I am not saying petrov isnt playable. Sure it is. But Statistics do work in your favour, somehow like a miracle. A computer program only needs to find the best move, and it doesnt matter that it always need to find the "only" move - it does, no problem.
Thus, you can not see if a position is difficult to play or not. Contrary, with the reti, you can play quite a few crap moves, and you still stand comparably well. So if you mess it up, you still have a playable position, thus the stats are good for white.
The rat defense is quite nice: 1.d4 d6 2.c4 e5 - and if white captures 3.dxe5 dxe5 4.Qxd8 Kxd8 - black has an easy life most of the time, with pet like set-up, saves time, and statistcs (If I remember correctly) are more wins for black than for white, something you would not expect. Didnt black just lose his right to castle?
Statistics do count, and I started looked at stats everywhere. Example: which are the most common endgames? Rook endgames, 50% of all endgames.
Tactics: most common motiv? Forks, by far!
Since chesstempo takes all tactics from real master games: what is the most common tactic rating that master players blundered in there games? Lo' and behold! it is just a mere 1450 Blitz rating, so quite simple stuff, often only 3 moves deep.
I meanwhile sense the tactics ahead, what kind of tactics could emerge, way before they happen. its all somehow connected, but the bottom line is:
chess is a complex game. It is a difficult game, no matter how good you are, you still blunder. Even Magnus loses games occasionally.
In order to learn, we need to break down complexity, not increase complexity. We need to extract the missing part that we do not know.
When I trained in the Chesstempo Blitz rating range (Blitz is misleading, it means timed mode, taking solving time into consideration) with self-created sets in spaced repetition mode - I learned a lot of patterns.
I isolated the sets into their motives: forks, pins, check-mate-in-2, overloading, skewers.... there are maybe 30 or more named motives, alone for non-check mates, I think.
I discovered that there are not only patterns, but some patterns belong to typical rating ranges. A chimney check mate is rather higher rated, a typical smother check mate rather below 1400 chesstempot blitz rating.
After I trained quite a bit in the low range, I later climbed a bit up, exploring the 1600-1800 ranges, but not too early, I really got a good understanding of the lower rated ranges first.
But... I dont think me as a 1800 starter is not "exactly" a beginner level.
One last example of patterns in a total different context:
Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.
Why can we read this? In chess its not that different: we see certain piece constellations, the rest of the set up is less important.
Note that you would not be able to read a very difficult text that contains words like "desoxyrybonucleic acid" or "adenosin triphosphate".
Agree with you.
In general what percentage should one dedicate to tactics vs more positional stuff?