I think it could be very valuable to use a language model to iterate over thousands of problems and identify the most common data structures and algorithms in order of how common they are.
For what it's worth, I read it when it came out and loved it. I lent it to a friend who never gave it back, which is probably another point in favour. I also enjoyed the follow-up "On the origin of good moves".
The author is Dutch International Master Willy Hendriks (IM is the rank below Grandmaster; there are ~4000 IMs and ~2000 GMs). He takes aim at chess instructors like IM Jeremy Silman, whose popular books The Amateur’s Mind and How to Reassess Your Chess teach students a structure for thinking about chess positions:
Imbalances include a space advantage on one side of the board, or two knights traded for two bishops, or an unsafe opposing king. Silman lists imbalances like these and tells you to use them to formulate plans which only then suggest individual moves.
Hendriks rejects this definitively: “No chess player thinks like this, no one has learned to play chess by thinking like this and even trainers and authors of chess books don’t think like this.” Instead, the verbal descriptions are mostly retroactive and follow the initial step of identifying good candidate moves from pattern recognition alone. He gives you some reasons to think this:
The book isn’t worth actually reading. Several of the chapters are taken from Hendriks’s old magazine columns or something and so have basically no relation to the thesis. He repeats himself, rambles, and the text is awkward (translated from Dutch?)
A typical bizarre tangent:
Chapter 3 is called “My Most Beautiful Move,” has nothing to do with the book’s thesis, and opens with this paragraph:
Some sections barely relate to chess.
Another reason not to read the book is that for all its outward presentation—encouraged by the author—as iconoclastic, the audience has no enthusiasm in arguing. The book won the English Chess Federation book award in 2012 and boasts praise from former US, British, and world champions. Silman himself responds by basically conceding the major points:
He defends his explicit pattern-descriptions as being more efficient and adds a layer of defense by saying his method is at least more pleasant and therefore easier to stick to (emphasis mine):
IM John Watson, another writer facing Hendriks’s fire, makes a stronger concession that descriptions merely the “medicine” easier to swallow:
The references to liveliness bring to mind the psychological principle of desirable difficulty (warning: PDF link): methods like rereading material create a momentary feeling that learning is taking place but are actually much worse than testing, which induces more learning but creates immediate feelings of lower performance.
Also interesting is to look for other cases of this broader distinction between methodical/formal study and just gobbling down patterns as fast as you can. Sports stories often feature scrappy players who have ingrained thousands of patterns just kicking the ball around compared to their well-heeled competitors. The most obvious parallel is language learning: Stephen Krashen’s input hypothesis argues that explicit instruction doesn’t affect learning, which flows entirely from the quantity of “comprehensible input” (which is understandable while containing a marginal increase in language-knowledge). Anecdotally a lot of people attribute most of their English fluency to watching a hundred hours of Friends while most language students describe retaining very little.
On the other hand, while Silman and others recommend skimming hundreds of high-level chess games to quickly absorb patterns, I’ve never heard of anyone quickly skimming hundreds of mathematics problems or leetcode problems. People are instead strongly advised to try as hard as they can to solve problems before giving up.
I asked the nearest IMO medalist to explain. He said that unlike peeking at the best move in a chess position after briefly considering what move you’d play yourself in the situation, reading the solution to a hard math problem doesn’t give you much of what you need to solve similar problems. Individual moves are the final atomic unit of a chess game, but math solutions are divided into many smaller “moves” and there isn’t any resource to quickly skim many math problems at various stages of solution and peek at the best “next move.” In the extreme case, reading just the numerical solutions to math problems would obviously not help you at all. But specific steps could be trained via pattern-avalanche—e.g., for problems that ask you to “prove statement X or provide a counterexample” you might be able to absorb the patterns that give you the intuition for whether you should start searching for a proof or a counterexample.
Reading thousands of leetcode problems while briefly chunking the solution into parts still feels promising. My recent refreshing of some technical chops is what reminded me of Hendriks’s book again in the first place (I’d first skimmed it while in a book store set up at the US Junior Open in 2013). The book isn’t very good but lends some evidence to the general question above and confirms my biases about “spinning your wheels reading” vs “grinding out reps.”