Proponents of the de la Maza/Woodpecker method often refer to the successful application of spaced repetition in foreign language learning. There are two big differences though between chess tactics and foreign languages:
1. To become conversational in a foreign language, one needs to learn about 2000-3000 words (and a few relatively well defined grammatical rules how to convert these words to their different forms). If I know "love", I can quite easily create "loves", "loved", "loving", "lovingly", etc. And the cost of making a mistake (e.g. using "love" instead of "loves", or "loving" instead "lovingly") is negligible.
On the other hand chess tactics is potentially an infinite set of positions. Infinity is much more than 2000-3000. Even if you have two very similar chess positions, a small nuance may decide if a Greek Gift Sacrifice works, or not, in a specific position. And unlike in language learning there are no rules how to convert a tactical position into another already known tactical position. One simply needs to calculate it for each concrete position. The cost of ignoring a nuance may decide the game.
However, spaced repetition might work very well for opening study: given that a typical repertoire on Chessable has 100-500 lines, and line would have around 12 moves at average, it means 1200-6000 moves. If we deduct moves that are shared between two opening lines, we could really reduce it to 500-2000 moves to be memorised. This is even less than the number of words to become conversational. This is something one can actually memorise. There is a potential trap though: as soon as one reaches a position that is similar but not same as another that you learnt, one needs to switch on thinking again.
2. But my far bigger concern with the Woodpecker method is a very different one: in spaced repetition applied to language learning, the intervals between repeating a specific word are increasing, at least if you can still remember the word. You are expected to see a repetition of the learnt word when you are close to forgetting it. I believe it is similarly implemented in Chessable: you first get to repeat a learnt line just after a few hours after learning it, but then the intervals are increasing.
However the de la Maza/Woodpecker method proposes something completely reverse! Decreasing intervals! Intuitively this makes no sense to me at all. I would be curious if there is any scientific research backing this idea? (There have been 40 years of research backing the idea of increasing intervals for language learning.)
Personally I did try the de la Maza/Woodpecker method several times over the last 14 years, the last time for 5 months between September 2023 and January 2024. I have spent between 16 and 23 hours each of these months. It did not improve my tactical skills, neither as measured by chesstempo or lichess, nor in my real OTB games. Before, in the previous years, 3 months long (or longer streaks) of solving tactics (but not in the Woodpecker way) used to boost my rating temporarily by 50-100 points. Of course, this is only a single data point. It would be great if someone organised a proper scientific study of effectiveness of spending XX hours using the Woodpecker method over Y months versus spending the same XX hours over the same Y months but without repetitions...
The comparison with SRS is very interesting. In addition to the increasing/decreasing intervals you mentioned, in SRS you typically repeat missed problems earlier/more often. In woodpecker you just repeat the whole set indiscriminately.
I've always been a little skeptical of this method as well, but I wonder if it's actually similar to learning opening theory. When I study openings, I could pick new variations each time I study (because I might encounter them someday), but I'll end up not learning any of the variations well at all. Is learning tactics similar? I sometimes encounter the same puzzle on Chess Tempo, often a decade or so after I solved it the first time. In the vast majority of cases I get the same problems right a second time, or I miss them a second time - it's much more unusual that they flip from missed to solved or solved to missed.
The hypothesis I'm coming to here is that my tactical practice on Chess Tempo is more about developing stamina and practicing what it's like to play tournament chess. The Woodpecker Method might involve more learning of new tactical patterns. I should probably give it a try.
Of course. But since there's no way to memorize everything, part of learning an opening is about familiarizing yourself with patterns and trends: if black seems to play Nf5 in each variation that you're studying, you're probably going to look at lines involving Nf5 when confronted with something new.
That's how I imagine the Woodpecker Method works - it's helping you understand the context in which a type of tactics works, so that you can apply it to comparable situations.
I am ambivalent on this. As an adult who did not really play chess seriously when I was younger, I definitely see that I don’t retain much information after seeing something once. The game I reviewed yesterday is mostly forgotten and I often re-fail puzzles. I felt like going over and over the Dan Heisman tactic book on Chessable helped me a lot to really absorb a lot of patterns… however, I would say that slowly going over the Step method book has done just as good a job because the tactics are greatly curated to show lots of the same patterns on slightly different positions. I felt like the big polgar books was also a lot like that. So the question remains, is doing the same set over and over better or it’s because the people who got better put so many hours into tactics ? It seems logical that the individuals willing to spend a day at some point doing 1000 puzzles would have more chance to be in the group of massive improvers haha.
I haven’t done the woodpecker method but have worked a lot on tactics on various online platforms. Decided to start the Steps method and so far feel I’m learning more through that method. Still doing online tactics through chessable and chesstempo mainly but the paper method with steps is beneficial I feel. As an older chess player still new to the game.
I'm working through the steps too. I also have found the paper based solving valuable. I'm thinking of working through the current book I'm on (step one, plus) and then maybe trying to repeat it in half the time.
I tend to « guess » the move more online vs on paper, even if I have gotten better at it. The steps method is not miles beater than everything else, but to me it’s a good set of tactics gradually improving from 1-2 moves to currently mostly 5 moves tactics at Steps 4. Good enough for me.
I haven't tried Woodpecker, but I suspect the key aspect is increasing the speed of each cycle. This pressure to get through the puzzles more quickly forces the solver to more rapidly scan the salient features of the position and generally become more efficient. You might say that you could just solve normal puzzles(non-Woodpecker) more quickly to get the same benefit, but the problem with that is you're probably going to get a ton wrong. Woodpecker gets around this by letting you "cheat" because you've re-solving puzzles.
So Woodpecker could be seen as similar to training wheels on a bike, or the pull-up machine at the gym. It trains you to operate at the next level - yes you're getting a little help but you're exercising exactly the right muscles/skills.
The interesting thing about De La Maza is that once he got to his expert rating he stopped. Heisman once told me that he simply said that he couldn’t learn what the other players knew beyond that point. Not sure why
What I suspect is that repeating puzzles may simply just build up your knowledge. Call it one way, not THE way. I imagine it’s like anything, some people will find the protocol very useful and some tedious. But anything that gets you to do puzzles will work.
It’s also interesting to me that to this day GM Smith will do Woodpecker to prep for a tournament .
Thanks for the shoutout. Well written post of course.
For me the program is challenging for sure. I urge anyone to give it a try and really commit. It’s no easy feat to accomplish. Secondly doing the Woodpecker or 7 Circles does have the definite end as you so eloquently started. So for me the idea is to do the process again with a new puzzle set. This way the puzzle set next gets stale. Memorization doesn’t really come into play rather, I found the concepts and motifs became easier to spot. Sure some of the puzzles would become automatic and easy to solve over time but with the plethora of tactics sets out there this process may never get old.
I will state that following either of these programs will have a law or diminishing returns once you reach certainly levels I would assume. After Master I expect you’ve solved your tactical blindness but what’s to say you couldn’t cycle sets or courses suitable for master level training? Endgames. Calculation courses. So on and so forth. Always working on that weakest part of your game?
I think the Woodpecker method also relies on several other 'modern learning'' ideas.
The first is an attempt to counter the Forgetting Curve (eg https://www.mindtools.com/a9wjrjw/ebbinghauss-forgetting-curve). This is the discovery/realization that new learning decays quickly. By tackling a large body of -things- repeatedly you are doing a sort of 'hyper'-review of those things. And spaced repetition is also a side effect. And it -should be- a way to counter Andy Lee's observation that haphazard discovery of old problems has the same result as the original.
Surprisingly, your idea that problem sets should be composed of problems with identical or similar themes runs up against another modern idea. In my math teacher career, many students (and some administrators) like the idea of doing a problem set of similar problems. When they are done, they feel like they have learned something. In fact, a week or two later students still feel like they've learned - there's a cognitive bias that suggests that confidence is build upon increased familiarity and not increased understanding. The modern idea to counter this is interleaved repetition - presenting multiple ideas actually fosters deeper understanding of the underlying ideas (in math, and I'm pretty sure it works similarly in chess, although that is more faith than my actual experience).
The tricky part about all this is that when you are learning how to solve math problems, you have to do "enough" similar problems to get an idea of what you are actually doing (although 'enough' frustratingly varies from person to person). Then moving on to something else is important, so that when you come back to this you learn to rely on the deeper patterns (or perhaps in chess the calculation ....).
I agree that there is a chance that the student will just be memorizing the solutions without building an understanding - but many of us strive for the understanding and "won't" fall victim to that.
At least that's what I very much wish to do! Thanks for the article!
Proponents of the de la Maza/Woodpecker method often refer to the successful application of spaced repetition in foreign language learning. There are two big differences though between chess tactics and foreign languages:
1. To become conversational in a foreign language, one needs to learn about 2000-3000 words (and a few relatively well defined grammatical rules how to convert these words to their different forms). If I know "love", I can quite easily create "loves", "loved", "loving", "lovingly", etc. And the cost of making a mistake (e.g. using "love" instead of "loves", or "loving" instead "lovingly") is negligible.
On the other hand chess tactics is potentially an infinite set of positions. Infinity is much more than 2000-3000. Even if you have two very similar chess positions, a small nuance may decide if a Greek Gift Sacrifice works, or not, in a specific position. And unlike in language learning there are no rules how to convert a tactical position into another already known tactical position. One simply needs to calculate it for each concrete position. The cost of ignoring a nuance may decide the game.
However, spaced repetition might work very well for opening study: given that a typical repertoire on Chessable has 100-500 lines, and line would have around 12 moves at average, it means 1200-6000 moves. If we deduct moves that are shared between two opening lines, we could really reduce it to 500-2000 moves to be memorised. This is even less than the number of words to become conversational. This is something one can actually memorise. There is a potential trap though: as soon as one reaches a position that is similar but not same as another that you learnt, one needs to switch on thinking again.
2. But my far bigger concern with the Woodpecker method is a very different one: in spaced repetition applied to language learning, the intervals between repeating a specific word are increasing, at least if you can still remember the word. You are expected to see a repetition of the learnt word when you are close to forgetting it. I believe it is similarly implemented in Chessable: you first get to repeat a learnt line just after a few hours after learning it, but then the intervals are increasing.
However the de la Maza/Woodpecker method proposes something completely reverse! Decreasing intervals! Intuitively this makes no sense to me at all. I would be curious if there is any scientific research backing this idea? (There have been 40 years of research backing the idea of increasing intervals for language learning.)
Personally I did try the de la Maza/Woodpecker method several times over the last 14 years, the last time for 5 months between September 2023 and January 2024. I have spent between 16 and 23 hours each of these months. It did not improve my tactical skills, neither as measured by chesstempo or lichess, nor in my real OTB games. Before, in the previous years, 3 months long (or longer streaks) of solving tactics (but not in the Woodpecker way) used to boost my rating temporarily by 50-100 points. Of course, this is only a single data point. It would be great if someone organised a proper scientific study of effectiveness of spending XX hours using the Woodpecker method over Y months versus spending the same XX hours over the same Y months but without repetitions...
The comparison with SRS is very interesting. In addition to the increasing/decreasing intervals you mentioned, in SRS you typically repeat missed problems earlier/more often. In woodpecker you just repeat the whole set indiscriminately.
I've always been a little skeptical of this method as well, but I wonder if it's actually similar to learning opening theory. When I study openings, I could pick new variations each time I study (because I might encounter them someday), but I'll end up not learning any of the variations well at all. Is learning tactics similar? I sometimes encounter the same puzzle on Chess Tempo, often a decade or so after I solved it the first time. In the vast majority of cases I get the same problems right a second time, or I miss them a second time - it's much more unusual that they flip from missed to solved or solved to missed.
The hypothesis I'm coming to here is that my tactical practice on Chess Tempo is more about developing stamina and practicing what it's like to play tournament chess. The Woodpecker Method might involve more learning of new tactical patterns. I should probably give it a try.
The difference is that with openings you actually want to memorize specific lines.
Of course. But since there's no way to memorize everything, part of learning an opening is about familiarizing yourself with patterns and trends: if black seems to play Nf5 in each variation that you're studying, you're probably going to look at lines involving Nf5 when confronted with something new.
That's how I imagine the Woodpecker Method works - it's helping you understand the context in which a type of tactics works, so that you can apply it to comparable situations.
I am ambivalent on this. As an adult who did not really play chess seriously when I was younger, I definitely see that I don’t retain much information after seeing something once. The game I reviewed yesterday is mostly forgotten and I often re-fail puzzles. I felt like going over and over the Dan Heisman tactic book on Chessable helped me a lot to really absorb a lot of patterns… however, I would say that slowly going over the Step method book has done just as good a job because the tactics are greatly curated to show lots of the same patterns on slightly different positions. I felt like the big polgar books was also a lot like that. So the question remains, is doing the same set over and over better or it’s because the people who got better put so many hours into tactics ? It seems logical that the individuals willing to spend a day at some point doing 1000 puzzles would have more chance to be in the group of massive improvers haha.
I haven’t done the woodpecker method but have worked a lot on tactics on various online platforms. Decided to start the Steps method and so far feel I’m learning more through that method. Still doing online tactics through chessable and chesstempo mainly but the paper method with steps is beneficial I feel. As an older chess player still new to the game.
I'm working through the steps too. I also have found the paper based solving valuable. I'm thinking of working through the current book I'm on (step one, plus) and then maybe trying to repeat it in half the time.
I tend to « guess » the move more online vs on paper, even if I have gotten better at it. The steps method is not miles beater than everything else, but to me it’s a good set of tactics gradually improving from 1-2 moves to currently mostly 5 moves tactics at Steps 4. Good enough for me.
I haven't tried Woodpecker, but I suspect the key aspect is increasing the speed of each cycle. This pressure to get through the puzzles more quickly forces the solver to more rapidly scan the salient features of the position and generally become more efficient. You might say that you could just solve normal puzzles(non-Woodpecker) more quickly to get the same benefit, but the problem with that is you're probably going to get a ton wrong. Woodpecker gets around this by letting you "cheat" because you've re-solving puzzles.
So Woodpecker could be seen as similar to training wheels on a bike, or the pull-up machine at the gym. It trains you to operate at the next level - yes you're getting a little help but you're exercising exactly the right muscles/skills.
The interesting thing about De La Maza is that once he got to his expert rating he stopped. Heisman once told me that he simply said that he couldn’t learn what the other players knew beyond that point. Not sure why
What I suspect is that repeating puzzles may simply just build up your knowledge. Call it one way, not THE way. I imagine it’s like anything, some people will find the protocol very useful and some tedious. But anything that gets you to do puzzles will work.
It’s also interesting to me that to this day GM Smith will do Woodpecker to prep for a tournament .
Thanks for the shoutout. Well written post of course.
For me the program is challenging for sure. I urge anyone to give it a try and really commit. It’s no easy feat to accomplish. Secondly doing the Woodpecker or 7 Circles does have the definite end as you so eloquently started. So for me the idea is to do the process again with a new puzzle set. This way the puzzle set next gets stale. Memorization doesn’t really come into play rather, I found the concepts and motifs became easier to spot. Sure some of the puzzles would become automatic and easy to solve over time but with the plethora of tactics sets out there this process may never get old.
I will state that following either of these programs will have a law or diminishing returns once you reach certainly levels I would assume. After Master I expect you’ve solved your tactical blindness but what’s to say you couldn’t cycle sets or courses suitable for master level training? Endgames. Calculation courses. So on and so forth. Always working on that weakest part of your game?
I think the Woodpecker method also relies on several other 'modern learning'' ideas.
The first is an attempt to counter the Forgetting Curve (eg https://www.mindtools.com/a9wjrjw/ebbinghauss-forgetting-curve). This is the discovery/realization that new learning decays quickly. By tackling a large body of -things- repeatedly you are doing a sort of 'hyper'-review of those things. And spaced repetition is also a side effect. And it -should be- a way to counter Andy Lee's observation that haphazard discovery of old problems has the same result as the original.
Surprisingly, your idea that problem sets should be composed of problems with identical or similar themes runs up against another modern idea. In my math teacher career, many students (and some administrators) like the idea of doing a problem set of similar problems. When they are done, they feel like they have learned something. In fact, a week or two later students still feel like they've learned - there's a cognitive bias that suggests that confidence is build upon increased familiarity and not increased understanding. The modern idea to counter this is interleaved repetition - presenting multiple ideas actually fosters deeper understanding of the underlying ideas (in math, and I'm pretty sure it works similarly in chess, although that is more faith than my actual experience).
The tricky part about all this is that when you are learning how to solve math problems, you have to do "enough" similar problems to get an idea of what you are actually doing (although 'enough' frustratingly varies from person to person). Then moving on to something else is important, so that when you come back to this you learn to rely on the deeper patterns (or perhaps in chess the calculation ....).
I agree that there is a chance that the student will just be memorizing the solutions without building an understanding - but many of us strive for the understanding and "won't" fall victim to that.
At least that's what I very much wish to do! Thanks for the article!
I appreciate Nate’s insightful chess commentary. There is a lot to be said for perseverance and consistency in chess improvement plans.