How the book came about

My co-author Philip Hurtado was good enough to share his perspective on the origins of our book Think Like a Super-GM

“In my youth I came across a fascinating chess book, which I picked up at a small library in the south of England. That book was titled ‘Thought and Choice in Chess’. What was so captivating about the book, was that it contained the verbalized thoughts of various world-class chess players as they tried to solve a chess puzzle thinking aloud. In particular, it contained the thoughts of my all-time favourite player Alexander Alekhine, as well as other leading grandmasters of the time, including Max Euwe, Reuben Fine and Paul Keres, amongst others.

I found it fascinating to compare each player´s train of thought as they tackled a position for the first time.

The purpose of that book was to uncover the process by which chess players think, and choose their move in a game of chess. The author, chess master and psychologist Adriaan de Groot, discovered that chess players go through four distinct phases prior to making their move. The “orientation phase”, in which the player assesses the situation, and forms a general idea of what to do next; the “exploration phase”, in which the player looks into the various possible continuations; the “investigation phase”, in which the player chooses the best move, and finally, the “proof phase”, in which the player confirms that the results of his investigation are valid.

De Groot conducted his experiment in 1938, and at that time Stockfishwas unfortunately not available to him, which made it hard to assess the quality, in numerical terms, of some of the lines given by the player. The Elo rating was not invented either, so de Groot was unable to establish any form of mathematical correlation between a player´s strength and the quality of his moves and thoughts. Even if he had found some way of determining the quality of the moves, and strength of the player, de Groot would have not been able to demonstrate if the results were of any statistical significance, as the concept of statistical significance and hypothesis testing were only just being developed, and the formulas were far too cumbersome and complex to be carried out with pen and paper.  

Thanks to my working career in the car manufacturing industry, where I have worked both as a Process Improvement engineer and Continuous Improvement consultant, I have learned to use highly specialized software packages to detect even the smallest of improvements in a process, by simply conducting a statistical analysis on a small number of samples.  It was whilst working at Jaguar Land Rover on the new Range Rover Velar, that I came up with the idea of revisiting de Groot´s experiment, but this time, using advanced statistical software to mathematically determine if there were any significant differences in the way a Super-GM thinks compared to other players.

For the experiment, all I needed was a set of unpublished chess puzzles (to avoid any advantage from players already knowing the solution), a chess engine to numerically evaluate the quality of the moves chosen by the players, a stopwatch to record the time taken to solve the puzzles, specialized statistical software to analyze the data, and – of course – a super GM to compare how he solved the puzzles against the other players.

The concept was in essence very simple: I would show the same chess puzzles to a wide array of chess players of varying strengths, and record their thinking process in real time, and use specialized analytical software to determine if there were any statistically significant differences in the way the super GM solved the puzzles and the rest of the players.

Apart from the obvious “time to solve the puzzle”, and “right or wrong move”, I wanted to monitor how players of different strengths could evaluate positions, as I suspected there would be detectable step changes here, and also because it is an area rarely covered in chess literature. 

I calculated the number of puzzles needed as well as number of participants and Elo rating bands. I had everything, except for one thing, the Super-GM.

Bold as you please, I decided to contact Britain´s top rated player and super-GM Michael Adams. I got his email address from his website, and “cold” contacted him with my idea.

To my very pleasant surprise, Mickey replied positively and wanted to know more about the idea. I was really excited, as Mickey is a player of legendary strength who had won the British championship numerous times, and had not written a book for more than 20 years.  Mickey prompted me to arrange a phone conversation in which I could explain the project in more detail.

Prior to contacting Mickey I had already tested the puzzles with various amateurs and strong club players, including some players of IM strength. One IM, and good friend of mine, Juan Reyes, suggested that I try the puzzles with his compatriot Julio Granda. I asked Juan if he thought there would be much difference between the way he solved the puzzles and how GM Julio Granda would solve them. “You´ll see” he said.

I was curious. I respected Juan a lot for his precise calculation ability and chess skills, and had doubts whether even a top GM like Julio Granda could analyse that much better than him.  

For those of you who have never had the chance to witness a super-GM calculate a chess puzzle, I will tell you now that it can be a shocking experience. Because I had the calculations from Stockfish, and knew beforehand the solution to each puzzle it came as a total shock to listen to Julio give me the exact same lines as I had from the chess engine. I was dumbfounded by his speed of thought in some puzzles.

The exact same thing happened when I tested the puzzles with Mickey a few weeks later. His very first move suggestions were often exactly the same as those given by Stockfish, and within minutes the lines he calculated would converge to those listed on the top of the chess engine.

Even without analyzing the data, I knew then there were qualitative, and quantitative step differences between how top rated Elo players solved the puzzles, and how lower rated players solved them. And these differences had nothing to do with superior opening preparation or chess knowledge.

Once I gathered all the data I needed, I number crunched the data on the software, and found extremely interesting mathematical correlations between a player’s Elo rating and his puzzle solving ability. Recording player’s assessments of the resulting positions proved particularly enlightening, and hearing their thought process in real time also provided some unexpected results that could not have been foreseen prior to the experiment. Other insights regarding the time it takes a Super-GM to solve a puzzle, and how they use that thinking time were also quite eye opening. The data highlighted significant correlations between a player´s Elo rating, and the time it takes them to spot the best move, as well as relationships between a player´s Elo rating and calculation skills, evaluation skills, etc.

Interestingly, I witnessed the same 4 phases of thought a player goes through, as de Groot described in his book.  Furthermore, I immediately detected that the orientation phase happens at lighting speed with super GMs, in most cases it just takes them a few seconds to get at the heart of the position, whereas for low rated players it takes minutes. Conversely, and rather unexpectedly, the super-GM spends noticeably more time on the “proof phase”, whilst for very low rated players, this phase is practically non-existent. 

These mathematical correlations are so conclusive, that it enabled me to write an equation that predicts a player´s strength simply by looking at their performance solving the puzzles.

In fact, one of the features of the book is that you can estimate your Elo strength depending on how well you fared solving the 40 puzzles.

Finally, I would like to add a few words about my working relationship with Mickey, and how his input enabled the book to evolve from a simple research project into what IM Andrew Greet, our editor at Quality Chess, has dubbed as a “masterpiece”. Throughout the whole editing of the manuscript, Mickey and I collaborated to share and compare our views on the key differences between the way he solved the puzzles, and other participants. I would reveal my new mathematical discoveries, for example the number of moves a Super-GM can see, or the time it takes for a Super-GM to spot the best move, and contrast these findings with his personal experience as an elite chess player.

Furthermore, even as the co-author, I find today, as I recently received my first copy of the book, an immense pleasure reading again Mickey´s pertinent and sometimes witty “Insights” given at the end of each puzzle, as well as his extraordinary engaging section “Grandmaster Secrets”.

I suppose it is difficult to remain objective being in part a creator of the book, but I am very proud of it, and think it is one of the best in my own collection.”

~ Philip Hurtado