I have solved puzzles around 2100~2300 for a long time. This range of puzzle focuses on tactics/trading that helps you win a piece or threatening to checkmate. But as I tried to go upward I was completely stuck. Often times the moves are very subtle and confusing, and even with the analyzer I was not able to understand why. What is something that I'm missing out?
Playing real games is what you're missing out.
Well, that's why they're hard, I suppose. A good idea is to try doing set tasks (ie, mates in 3) so you have a clearer picture of what you're aiming at.
That's how I went from stable ~2200 to ~2500:
- 1 month of "easy" puzzles forced me to work perfectly the answer, or lose 35-40 points. It then becomes a habit or, at least, an easier task to fully calculate the moves (especially opponent's moves!). I'll look for the forum fellows that recommended this to me. Great advice.
- Spotting my weaker areas (pawns endgames and checkmate in 3-5 moves, many puzzles are really about threatening checkmate) and working a lot on these, including some books on pawns endgames (I think these are painful to learn just from the puzzles).
- Usually (not lately, but I was on vacation) I try just the daily puzzle, 12 mix and 12 checkmate, no time limit and no room to recover from a bad streak.
- Carefully reading solved puzzles and trying to provide a similar explanation for my own puzzles (especially the missed ones). I borrowed "The Smyslov Workbook" by Hansen and Lakdawala, which is way above my level, but has nice explanations.
Hope that helps, if you try it let me know how it went!
- 1 month of "easy" puzzles forced me to work perfectly the answer, or lose 35-40 points. It then becomes a habit or, at least, an easier task to fully calculate the moves (especially opponent's moves!). I'll look for the forum fellows that recommended this to me. Great advice.
- Spotting my weaker areas (pawns endgames and checkmate in 3-5 moves, many puzzles are really about threatening checkmate) and working a lot on these, including some books on pawns endgames (I think these are painful to learn just from the puzzles).
- Usually (not lately, but I was on vacation) I try just the daily puzzle, 12 mix and 12 checkmate, no time limit and no room to recover from a bad streak.
- Carefully reading solved puzzles and trying to provide a similar explanation for my own puzzles (especially the missed ones). I borrowed "The Smyslov Workbook" by Hansen and Lakdawala, which is way above my level, but has nice explanations.
Hope that helps, if you try it let me know how it went!
Did you try to use the radar plots to work on detected skill set decomposed into themes.
you need to play a bunch under normal (drop down menu), in random mode.. for the spectrum decomposition into theme performance ratings to become relevant. daily flukes get cancelled with more puzzles over many days.. limit is 90 days unfortunately.. so 3 months view of tactical skill decompositions. Post-puzzle and failed puzzle revisit.
And the above mate known as solution stopping point, allows to separate the tactical plan goal setting problem (now mate), from the methods search/discovery/learning/recognition problem. Toward game generalization, I think mates are never wrong plan targets.. The problem with non-mate is double unknown. the depth of planning and the nature of goal that is considered a solution to the position setup presented. 3 problems. And programmed SF based criteria which might not be very transposable to human whole game scope of play.
drop down difficulty menu from easiest to hardest (probably that is how you aim at puzzle ratings).
puzzle offer on normal in random theme, or populated enough themes (in puzzle IDS), should be a symmetric band around your instantaneous/current rating. The drop down menu, rating deltas are the shift amount of that band center, AFAIK.
I assume you don't care about your own rating. caring about it, would kind of waste a lot of things one can do with the puzzles as a learning opportunity.. and conscious project of chess study.
very hard puzzles, even failed, can be very motivating for studying the tactical motifs (the exercise of matching theme definition to board information might help understand why one is missing the subtleties). They do not help for weakness detection disambiguation. The system is slanted toward being good at telling your strengths, I would suggest that one should assume that all the other themes are weakness (the motif ones, not the score ones) until they make it to your strengths.
Don't assume other people experience made chess theory made theme definition by lichess, can't help you, and that only more puzzle, new puzzles all the time is the only way to gain from lichess puzzle system..
you need to play a bunch under normal (drop down menu), in random mode.. for the spectrum decomposition into theme performance ratings to become relevant. daily flukes get cancelled with more puzzles over many days.. limit is 90 days unfortunately.. so 3 months view of tactical skill decompositions. Post-puzzle and failed puzzle revisit.
And the above mate known as solution stopping point, allows to separate the tactical plan goal setting problem (now mate), from the methods search/discovery/learning/recognition problem. Toward game generalization, I think mates are never wrong plan targets.. The problem with non-mate is double unknown. the depth of planning and the nature of goal that is considered a solution to the position setup presented. 3 problems. And programmed SF based criteria which might not be very transposable to human whole game scope of play.
drop down difficulty menu from easiest to hardest (probably that is how you aim at puzzle ratings).
puzzle offer on normal in random theme, or populated enough themes (in puzzle IDS), should be a symmetric band around your instantaneous/current rating. The drop down menu, rating deltas are the shift amount of that band center, AFAIK.
I assume you don't care about your own rating. caring about it, would kind of waste a lot of things one can do with the puzzles as a learning opportunity.. and conscious project of chess study.
very hard puzzles, even failed, can be very motivating for studying the tactical motifs (the exercise of matching theme definition to board information might help understand why one is missing the subtleties). They do not help for weakness detection disambiguation. The system is slanted toward being good at telling your strengths, I would suggest that one should assume that all the other themes are weakness (the motif ones, not the score ones) until they make it to your strengths.
Don't assume other people experience made chess theory made theme definition by lichess, can't help you, and that only more puzzle, new puzzles all the time is the only way to gain from lichess puzzle system..
"I was not able to understand why"
My guess is that you are over reliant on calculation rather than detecting the puzzles' themes and juggling them around to find the solution whilst not over-looking at the opposing colour's counter play (engines/solutions do not show counter play!).
The advice above to drill in terms of themes is probably the way to go. Also I wouldn't take the puzzle elo rating too seriously as I managed to get to 2700+ just doing mates-in-2s! Currently I am on mates-in-3s and holding at around 2500 after dropping to 2300+ when I first started the mates-in-3s drill process. I think the best use of the Lichess puzzles database is to pick a suitable puzzle elo rating and drill a theme up to it, then switch to another theme. Expect a saw-tooth pattern as far as puzzle rating goes.
There is a very good community blog recently posted that will give you some good advice on which themes to target for your playing level - has a graph of variance of puzzle themes against puzzle rating. For example, doing the back-rank theme would be a bit pointless if your rating is 2000+ but the pin theme will always be valuable to master.
My guess is that you are over reliant on calculation rather than detecting the puzzles' themes and juggling them around to find the solution whilst not over-looking at the opposing colour's counter play (engines/solutions do not show counter play!).
The advice above to drill in terms of themes is probably the way to go. Also I wouldn't take the puzzle elo rating too seriously as I managed to get to 2700+ just doing mates-in-2s! Currently I am on mates-in-3s and holding at around 2500 after dropping to 2300+ when I first started the mates-in-3s drill process. I think the best use of the Lichess puzzles database is to pick a suitable puzzle elo rating and drill a theme up to it, then switch to another theme. Expect a saw-tooth pattern as far as puzzle rating goes.
There is a very good community blog recently posted that will give you some good advice on which themes to target for your playing level - has a graph of variance of puzzle themes against puzzle rating. For example, doing the back-rank theme would be a bit pointless if your rating is 2000+ but the pin theme will always be valuable to master.
I remember the first time I looked up "Motifs" in Chess, a bunch of new layers of analysis were unlocked. Some of them far more advanced than others, right? This is what happens as you get past the 2400ish Puzzle Barrier, you're reaching new motifs that you aren't familiar with given you skill level. Personally, as someone who is floating between 2300-2500 puzzles, I notice it's WAY MORE IMPORTANT to consider your opponents best move as seriously as possible, there are quieter setups for tactics, also very threatening opponent plans to counter, instead of forced offensive combinations.... there are positional ideas where we have to say "okay if I go here... how does he stop it? ... ohhh, is there a different approach? Wait a second, my opponent can do a devastating attack, how do I stop that while also advancing my own plan... Oh look there's exactly one square I can achieve this, lets start calculating from here..." type ideas. This line of thinking just doesn't really exist until this rating of puzzles, from what I've noticed.
TLDR - Around 2400 puzzle rating, I had to start taking my opponents position much more seriously and calculate their best moves much better.
TLDR - Around 2400 puzzle rating, I had to start taking my opponents position much more seriously and calculate their best moves much better.
@facevisi10 said in #1:
> I have solved puzzles around 2100~2300 for a long time. This range of puzzle focuses on tactics/trading that helps you win a piece or threatening to checkmate. But as I tried to go upward I was completely stuck. Often times the moves are very subtle and confusing, and even with the analyzer I was not able to understand why. What is something that I'm missing out?
Computer generated puzzles from database blitz and rapid/bullet games are not the best way to train in tactics, due to some of the illogical and convoluted exercises that exist, many with positions that tend to confuse humans and which NEVER have multiple solutions to an advantage (too often, the second line is 0.00 or disadvantage). They're only good if you have nothing available on you except a phone. You also get very random motifs which are also VERY poor for actually training your brain in segments and themes (that's why many book trainers start on a theme and slowly build on it). It's not much fun at all going from a position with some obscure Queen check moves to end up with a double attack five moves later, to a very difficult endgame in the very next problem. You're much better off doing a book written by a strong IM or GM. I highly recommend John Nunn's puzzle book for anyone who has already gone through M.Blokh's Combinational Motifs/Combinative Motifs OR his CT-Art 6.0+ software (which is based on mostly problems from his book, but designed as a trainer)--if you have NOT read the Blokh book however--you should.
> I have solved puzzles around 2100~2300 for a long time. This range of puzzle focuses on tactics/trading that helps you win a piece or threatening to checkmate. But as I tried to go upward I was completely stuck. Often times the moves are very subtle and confusing, and even with the analyzer I was not able to understand why. What is something that I'm missing out?
Computer generated puzzles from database blitz and rapid/bullet games are not the best way to train in tactics, due to some of the illogical and convoluted exercises that exist, many with positions that tend to confuse humans and which NEVER have multiple solutions to an advantage (too often, the second line is 0.00 or disadvantage). They're only good if you have nothing available on you except a phone. You also get very random motifs which are also VERY poor for actually training your brain in segments and themes (that's why many book trainers start on a theme and slowly build on it). It's not much fun at all going from a position with some obscure Queen check moves to end up with a double attack five moves later, to a very difficult endgame in the very next problem. You're much better off doing a book written by a strong IM or GM. I highly recommend John Nunn's puzzle book for anyone who has already gone through M.Blokh's Combinational Motifs/Combinative Motifs OR his CT-Art 6.0+ software (which is based on mostly problems from his book, but designed as a trainer)--if you have NOT read the Blokh book however--you should.
@Falkentyne said in #8:
> training your brain in segments and themes
Well you do verbalize very well that idea. I will try to borrow how expressive that is. I have been finding this so obvious for so long, and not really sensing it being part of common sense in theories of learning underlying what I have been reading on chess websites, mostly here in the forum, and blogs.
However, once we know the typical restrictions that lichess automatically generated puzzles have with respect to real game same segment length ranges, but with any sliding starting point along the game, one might become aware of the lacks, and then compensate with the more designed or composed puzzles if those are really aiming at the pattern and how to generalize them in more concrete situations. But I think what is missing from the spectacular PV profiles we agree is there, as position challenge experience set, I think would not be tactics books, but compositions which target the ability to use those as building blocks, perhaps, but that would stimulate beyond tactical skills: I mean plan level imagination, desirable future features to seek as stepping stones to further goals, , and how to chop a long solution into things one can manage within their personal calculation breadth. which is about teaching how to read a position potential beyond the material count system (including it perhaps but not limited to positions with that kind of signal only).
So, that the complementary study time window (a limited daily or weekly resource for a lot of us) (also a type of segment or time chunk of optimal attention resource available) can focus on complementary experience territory within the well said pedagogical framework of game segments (continuation segments from positions).
Thank you for using the word segment. I thought it was missing in chess language.. I did not understand why main line was used profusely, but not segment. or segmentation. One can chop a line into segment, while accepting that visual analogy of line why stop there.
> training your brain in segments and themes
Well you do verbalize very well that idea. I will try to borrow how expressive that is. I have been finding this so obvious for so long, and not really sensing it being part of common sense in theories of learning underlying what I have been reading on chess websites, mostly here in the forum, and blogs.
However, once we know the typical restrictions that lichess automatically generated puzzles have with respect to real game same segment length ranges, but with any sliding starting point along the game, one might become aware of the lacks, and then compensate with the more designed or composed puzzles if those are really aiming at the pattern and how to generalize them in more concrete situations. But I think what is missing from the spectacular PV profiles we agree is there, as position challenge experience set, I think would not be tactics books, but compositions which target the ability to use those as building blocks, perhaps, but that would stimulate beyond tactical skills: I mean plan level imagination, desirable future features to seek as stepping stones to further goals, , and how to chop a long solution into things one can manage within their personal calculation breadth. which is about teaching how to read a position potential beyond the material count system (including it perhaps but not limited to positions with that kind of signal only).
So, that the complementary study time window (a limited daily or weekly resource for a lot of us) (also a type of segment or time chunk of optimal attention resource available) can focus on complementary experience territory within the well said pedagogical framework of game segments (continuation segments from positions).
Thank you for using the word segment. I thought it was missing in chess language.. I did not understand why main line was used profusely, but not segment. or segmentation. One can chop a line into segment, while accepting that visual analogy of line why stop there.
Would agree mostly with NM Falkentyne's comments, however I think some qualification is needed.
For the last few months I have also been doing the CT-Art 4.0 system over on Chess King. As a c.1500 OTB player and doing only the 'easy' puzzles I have found the going very tough, with some puzzles taking 10 to 15mins. These are clearly superior puzzles compared to LiChess' offerings, and far more realistic for long duration OTB play. The Chess King system gives me a puzzle rating of 1900+ and I love the way it works. Due to the difficulty I am less inclined to recommended CT-Art to people below a gameplay level of say c1400 - I believe it puts the own lower cap at c. 1200 and higher at c. 2400. LiChess puzzles, despite the biases mentioned, should be just fine for most on-line players, imo.
Chess King has other puzzle courses which I suspect would be more suitable for lower level players - like myself. Note, I am very time sensitive and I suspect few players of my ability or lower would find spending 10 to 15mins on a single puzzle efficient when Blitz games are so easily available. In my case I tried thousands of Blitz games over on chessdotcom and found this pointless in terms of improvement, but took a long time to realise this!
I also have Nunn's 1001 puzzle book but have made little use of it - no reason - and can report that it is very well formatted for use on the Kindle - puzzle on one page, solution on the next. It's initial difficulty slope seems much kinder, imo.
One question though, what do you think of drilling the opening themed puzzles? I have so far never tried do this but may soon.
For the last few months I have also been doing the CT-Art 4.0 system over on Chess King. As a c.1500 OTB player and doing only the 'easy' puzzles I have found the going very tough, with some puzzles taking 10 to 15mins. These are clearly superior puzzles compared to LiChess' offerings, and far more realistic for long duration OTB play. The Chess King system gives me a puzzle rating of 1900+ and I love the way it works. Due to the difficulty I am less inclined to recommended CT-Art to people below a gameplay level of say c1400 - I believe it puts the own lower cap at c. 1200 and higher at c. 2400. LiChess puzzles, despite the biases mentioned, should be just fine for most on-line players, imo.
Chess King has other puzzle courses which I suspect would be more suitable for lower level players - like myself. Note, I am very time sensitive and I suspect few players of my ability or lower would find spending 10 to 15mins on a single puzzle efficient when Blitz games are so easily available. In my case I tried thousands of Blitz games over on chessdotcom and found this pointless in terms of improvement, but took a long time to realise this!
I also have Nunn's 1001 puzzle book but have made little use of it - no reason - and can report that it is very well formatted for use on the Kindle - puzzle on one page, solution on the next. It's initial difficulty slope seems much kinder, imo.
One question though, what do you think of drilling the opening themed puzzles? I have so far never tried do this but may soon.