Think Like a Super-GM.pdf

Contents Title Page Key to Symbols used & Acknowledgements 4 Publisher’s Foreword 5 Part I Introduction 1.1 Introduction by Philip Hurtado 10 1.2 Introduction by Michael Adams 15 1.3 Methodology & Data Analysis 17 1.4 What is a Super-GM? 21 1.5 How to Use this Book 23 1.6 Tips for Solving the Puz...

Contents Title Page Key to Symbols used & Acknowledgements 4 Publisher’s Foreword 5 Part I Introduction 1.1 Introduction by Philip Hurtado 10 1.2 Introduction by Michael Adams 15 1.3 Methodology & Data Analysis 17 1.4 What is a Super-GM? 21 1.5 How to Use this Book 23 1.6 Tips for Solving the Puzzles 27 Part II Puzzle Section 2.1 Solver Biographies 30 2.2 40 Positions and Solutions 36 Positions 1-4 37 Positions 5-8 63 Positions 9-12 89 Positions 13-16 117 Positions 17-20 149 Positions 21-24 175 Positions 25-28 203 Positions 29-32 235 Positions 33-36 264 Positions 37-40 289 2.3 Calculate your Performance 315 Part III Bonus Puzzles 3.1 8 Bonus Puzzles and Solutions 318 Positions 41-44 Positions 45-48 Part IV Conclusions from the Puzzles 4.1 The Candidate Move 372 4.2 The Art of Falsifying 374 4.3 Why Can’t I Play Like a Super-GM? 379 2 4.4 Pearson’s Correlation 392 4.5 How Many Moves Ahead Can a Super-GM See? 395 4.6 Grandmaster Secrets 401 Part V Eyetracker Experiment 5.1 Introduction by Michael Adams 418 5.2 Introduction by Philip Hurtado 419 5.3 Solver Biographies 420 5.4-5 Eyetracker Puzzles and Solutions 421 5.6 Michael’s Analysis of his Heat Maps 448 5.7 Eyetracker Conclusions 453 3  Think Like a Super-GM By Michael Adams Philip Hurtado Quality Chess http://www.qualitychess.co.uk First edition 2022 by Quality Chess UK Ltd Copyright © 2022 Michael Adams & Philip Hurtado All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, electrostatic, magnetic tape, photocopying, recording or otherwise, without prior permission of the publisher. Paperback ISBN 978-1-78483-167-7 Hardcover ISBN 978-1-78483-168-4 All sales or enquiries should be directed to Quality Chess UK Ltd, Suite 247, Central Chambers, 11 Bothwell Street, Glasgow G2 6LY, United Kingdom Phone +44 141 204 2073 e-mail: [email protected] website: www.qualitychess.co.uk Distributed in North and South America by National Book Network Distributed in Rest of the World by Quality Chess UK Ltd through 4 Sunrise Handicrafts, ul. Szarugi 59, 21-002 Marysin, Poland Typeset by Jacob Aagaard Proofreading by Jeremy Hart Edited by Andrew Greet Cover design by Kallia Kleisarchaki Photos on cover and pages 36 & 447 by Tara MacGowran 5  Key to symbols used ² White is slightly better ³ Black is slightly better ± White is better µ Black is better +– White has a decisive advantage –+ Black has a decisive advantage = equality © with compensation „ with counterplay ƒ with an initiative ÷ unclear ? a weak move ?? a blunder ! a good move !! an excellent move !? a move worth considering ?! a move of doubtful value ™ only move # mate We shall define this to mean that Black has an advantage of magnitude 2.5 or more. In other >–2.5 words this means Black is certainly winning. We shall define this to mean that White has an advantage of magnitude 2.5 or more. In other >+2.5 words this means White is certainly winning. Acknowledgements Firstly, the authors wish to say a big “Thank you” to all the players who participated in the puzzlesolving sessions, as well as the eyetracking experiment. Thanks to Francois Labelle for helping to supply some statistical information, and to Charlie Storey for providing the game score and some comments about Puzzle 33. We would like to give special thanks to Jon Ward, Director of Tobii Pro UK Limited, for providing us with the software licences to run the heat maps and videos, and to Dr. Ben Marshall, for helping to set up and oversee the eyetracker experiment. 6 Publisher’s Foreword This is a chess book unlike any other. Even the authors have had a hard time summarizing the content, producing no less than six mini-introductions between them! You will ‘meet’ the two co- authors soon enough, as each author has written his own introduction. Michael (Mickey) Adams obviously needs no introduction – there are not many individuals in the world who can rival his credentials as a chess player. The name of Philip (Phil) Hurtado will be less recognizable to most readers, but Phil’s creative vision and scientific knowhow kickstarted this project and remained instrumental throughout. The main purpose of this foreword is to offer the reader a clear summary of what you can expect from this book. If I had to summarize this book in just a few words, I would say: “Puzzle book meets The Master Game meets Science.” (By the way, neither Phil nor Mickey has ever mentioned The Master Game as a source of inspiration for the project – but it was the first point of reference that came to my mind when I heard about the format of the book.) For those who don’t know, The Master Game was a BBC production of televised tournaments involving grandmasters and other strong players, which ran from 1976-1983. What made the production uniquely compelling was that, immediately after each game had ended, the producers made audio recordings of the players as they reproduced their thought processes out loud. Thus, in the final production, the television audience would watch the games unfold while listening to the thoughts of the players, as if in real time. The series was also turned into books, which I had on my shelf as a kid – and although I was too young to watch the shows when they were originally broadcast, I remember seeing a few of them on VHS tape when I was older. Whether in video or book format, I always found it fascinating to follow the thoughts of the players. So how do these three elements of puzzles, recorded thought processes and science come together? Essentially, Phil designed a grand science experiment to analyse the performance of different chess players of varying abilities when solving a selection of puzzles, and successfully persuaded Mickey to partner up in the project. Phil recorded players’ thoughts and, with the players’ permission of course, included a selection of them under the solutions for each puzzle, giving the reader a fascinating glimpse into the mental processes of each solver. Phil also meticulously recorded the moves chosen, time taken and much more for each solver, with a view to analysing the mass of data and using the scientific method to draw conclusions about what exactly separates the strongest chess players from everyone else. Along the way, Mickey’s involvement increased and the project evolved into something even more special than was originally envisaged. Here is a breakdown explaining what you can expect in each of the main parts of the book. 7  Puzzle Section The largest section of the book consists of forty puzzles of varying difficulty. You are invited to solve each puzzle, writing down your next move plus any supporting variations, as well as your evaluation of the position. So far, this sounds much like any other puzzle book. However, this book does not merely contain a standard solution. Instead, under the solution section for each puzzle you will find: 1) The detailed thought processes of several players of varying playing strengths – from near-novice players, through to club players, IMs and GMs, and finally Michael Adams himself During breaks in editing, I solved several of the puzzles myself – partly for my own training as a player, and also to be able to gain a better experience of this book from the reader’s perspective. Later, as I edited the corresponding solution sections, I found it fascinating to compare my thinking to that of the featured solvers. Every reader’s experience will be different, but mine revealed the following insights: * At my best, I was able to solve some of the most difficult positions – perhaps not with quite the same efficiency and precision as Mickey and the other top GMs, but still in a way that confirmed I am capable of performing well in certain types of position. * For other puzzles, by some combination of intuition and calculation, I was able to select the best move, but comparing the thought process of Michael and other top players highlighted gaps in my analysis and general chess understanding. I may have chosen the right move, but the process by which I got there fell far short of Mickey’s level. The ability to follow the exact thought process of Mickey and other strong GMs was a real eye-opener in terms of the kinds of details I can improve on. * At my worst, there were a few puzzles where I concentrated hard but completely failed to hit upon the right idea – giving me a clear idea of the types of position where my play needs to be improved. Again, it was illuminating to see how quickly Mickey and other leading GMs were able to get right to the crux of certain positions which I struggled with. In short, the ability to compare my own thought process with that of weaker, equal-strength and stronger players was something I found incredibly illuminating, and has significantly increased my awareness of my own strengths and weaknesses as a player. I have no doubt that the readers will find this process to be just as interesting and beneficial as I did. 2) A “Deeper Analysis” section by Michael Adams Once the thought processes of the various players have been presented, Michael then provides a comprehensive solution to the puzzle. Each solution contains the obligatory engine analysis of course – but more importantly, Mickey’s personal insights, including his reflections on the few puzzles where his own solution was not the best. 8  3) Post-Solution section Following the analysis section by Michael, you can find the scoring system for that puzzle, followed by a few final reflections about the puzzle from both of the authors. What made the position challenging? What were the most common errors in thinking among the panel of solvers? What does the Super-GM regard as the defining features of the position? And what were the key attributes which enabled the strongest players to find the best solution where others failed? *** By now you can see why there are only forty puzzles! The solutions are massively expanded by the inclusion of the thought processes of numerous players. This is of tremendous value to the reader in my opinion – and few (if any) other puzzle books have offered such a feature. This much alone would have made the book a valuable addition to chess literature – but there’s more to follow. Bonus Puzzle Section Essentially this is more of the same. These eight additional puzzles are positions which, in the authors’ opinion, were ultimately not quite suitable for the main puzzle section, but which nevertheless contain a great deal of instructive content. They are presented in the same way as the main puzzles, with solver commentaries and Michael’s detailed analysis included in every solution. Conclusions from the Puzzles The book’s title Think Like a Super-GM was not just a sales pitch or ‘clickbait’ – the authors really have left no stone unturned in trying to unravel the mystery of what separates the thinking of elite players from the rest of us. I think Phil was really in his element in this section, channelling what I will cheekily refer to as his ‘Mad Scientist’ persona to crunch the numbers and draw on the data to shed light on matters such as: * The importance of the candidate move * Why do the strongest players devote most of their thinking time to checking the consequences of the best move after they have already found it? * “How many moves can you think ahead?” is a question that you will probably roll your eyes at – but Phil has nevertheless drawn on the experimental data in an attempt to answer it for players all the way up to Mickey’s level. Once Phil has finished shining the scientific spotlight on these and other questions which the experimental data helps to answer, it is Michael’s turn to offer his conclusions in a section entitled 9 Grandmaster Secrets. Here, the Super-GM offers what I found to be some highly instructive insights into what he sees as the key attributes which enabled him and the other grandmasters to consistently outperform lesser players in the puzzle test. Once again, having every player’s thought process available proved invaluable here, as Mickey was able to illustrate his points by revisiting several puzzles and recapping the exact reasons players gave for choosing or rejecting certain moves. *** We would certainly have been proud to publish the book if it ended here, but the authors had one more ace up their sleeve. The Eyetracker Experiment The idea for this actually came from Mickey after he watched a sports documentary involving laboratory testing of an athlete’s eye moments while performing. Mickey mentioned the idea to Phil and, in not much more than the blink of an eye, the experiment was arranged in a suitably equipped testing facility. Simply put, the eyetracker experiment used sensors to follow the eye movements of Michael and other players as they solved chess puzzles in real time. The technology is so precise that it was able to pinpoint the exact squares on which players’ eyes were focused at all times. Once the player had finished solving each puzzle, at the touch of a button the lab technicians produced a ‘heat map’ for that puzzle, offering a vivid, visual representation of the areas of the board that received the most attention from that solver. Comparing Mickey’s eyetracking and heat maps to those of the other players offered yet another means of looking inside the mind of a Super-GM and comparing Mickey’s thinking process with players at lower rating levels. *** Summing up, this is a remarkable book with many unique features. With Mickey’s superlative chess ability alongside Phil’s scientific and data-analysing acumen, we could hardly have wished for two more suitably qualified authors for this project. The ability to think like a Super-GM is something only a select few players will ever acquire, but this book does offer a number of unique insights which help to uncover the mystery of how such players perform as well as they do. I hope you will enjoy solving the puzzles and comparing your thoughts to those of Mickey and the other solvers. With the further insights gained from Phil’s data analysis and Mickey’s expert conclusions, I am quite sure you will find yourself thinking about chess in completely different ways from before – taking you, if not all the way to Super-GM level, then at least a healthy step in that direction. Andrew Greet Editor, Think Like a Super-GM December 2021 10  That Day in the Library “We must be careful not to believe things, simply because we want them to be true.” Richard Feynman, Winner of The Nobel Prize in Physics 1965 Let me introduce myself: I am Philip Hurtado. I studied Aeronautical Engineering at the University of Hertfordshire, followed by a Master’s degree in Physics at the University of St Andrews. Although I had short stints in both aerospace and nuclear fusion research, I ended up developing a career in car manufacturing. Here is the interesting part: whilst working in automotive plants, I discovered a technique for improving construction processes called Six Sigma. This method uses advanced statistical analysis, to help engineers increase the performance of a production line, by decreasing process variation. This technique leads to a reduction in defects and a dramatic improvement in product quality. You can see an example of how effective the Six Sigma method is, when you look at the beautifully flushed doors of a new Range Rover Velar and compare them to older models. Anyway, why am I telling you all this? 11 Having spent the past five years working as a Lean Six Sigma Black Belt consultant, I thought to myself, “Why not apply these same continuous improvement techniques to my chess?” After all, I have access to both the statistical software and the chess engines that would enable me to measure and monitor my progress. For six months, I designed some experiments to figure out which changes had the biggest effect on improving my chess. I tried all sorts of things: I studied the effect of diet, exercise, time of play, drinking energy drinks, studying openings, studying endgames and so on. The funny thing I found is that a good night’s sleep, combined with being on holiday, with a happy financial situation and general peace of mind, gave better results than all of the other factors combined. Still, the improvement was only small. Maybe 100 Elo points, from 1900 to 2000, which still fell well within the natural variation of my playing strength. So a second idea occurred to me: why not devise an experiment that would uncover how a top grandmaster makes decisions? I could then compare it to the way I think about my own moves, and learn from the differences. Again, I would have the aid of a chess engine to help measure the quality of the moves, and I would have the statistical software to determine how significant these differences are. But before we begin, let me share a bit of my personal chess background. Thirty-two years ago, at the age of nineteen, I came across a fascinating chess book, which I picked up at Hove Library on Church Road, close to where I used to live with my grandmother. That book was called Thought and Choice in Chess. It was written by a Dutch psychologist who conducted a spellbinding experiment, where he asked famous chess masters and amateurs to think aloud, whilst trying to solve chess puzzles. I quickly asked the lady at the library for a piece of paper, so I could cover the solutions and set out trying to find out the best moves myself. Apart from the sheer enjoyment of trying to solve the puzzles, I was completely mesmerized by reading the train of thoughts of some of the masters, in particular those by former World Champion Alexander Alekhine, my all-time favourite chess player. 12  Chess master and psychologist Adriaan de Groot gave this position to Alekhine, Keres, Euwe, and other leading players at the 1938 AVRO tournament, as well as to a number of weaker, but still expert, players. With his experiment, de Groot was trying to unravel the mysteries behind thinking and decision-making in chess, and determine the differences between strong masters and weaker players. You may wish to consider the position yourself, before comparing your thoughts to those of Alekhine, as shown below. Here are Alexander Alekhine’s comments to the puzzle shown to him by de Groot. I am sure you will love them as much I did when I first read them thirty-two years ago, in a quiet library in the south of England. Alekhine spent nine minutes on the above position, on 28 November 1938. The original commentary has been changed to algebraic notation. Comments by Alexander Alekhine At first sight there is a dark memory of a tournament game between Botvinnik and Vidmar (Nottingham). There is certainly a resemblance: The same queen position on d3. Opening type: Queen’s Gambit Accepted. In half an hour I should be able to logically reconstruct the moves up to this position. Is the pawn on b2 really attacked? Let’s see what counter possibilities White has. Which moves are worth considering? 1.Bxd5 has to be calculated. 1...Bxd5 is not good; 1...Nxd5 is no good. So, 1...exd5. And now the knight on f6 is getting a little weak. 13  2.Ng4 or 2.Rfe1 might follow. Or also 2.Nxc6 followed by 3.Re1. Or possibly 2.Qf3. Well, in any case, 1.Bxd5 is one enticing move. But now let’s have a look at other possibilities: 1.Ng4 deserves consideration. But doesn’t look so good. 1.Nxd5 maybe. 1...Nxd5 2.Bxd5 Bxg5, with many recapturing possibilities. But at first sight, none convincing. 1.Nxc6. But after 1...bxc6, the d5-square is well controlled. That’s a pity. In any case, White has by far the best of it. I would be very happy if I had such a position in a tournament game. Are there any other moves apart from 1.Bxd5 that are decisive? 1.Bxd5 is very strong, that much is certain. 1.Nxc6 bxc6, there is pressure on c6, White has the pair of bishops. Also very agreeable. But the position promises more. Let’s check over other sorts of things again... Well, in case of time pressure, I would play 1.Bxd5. Analysis with additional variations added by Michael Adams Alekhine begins by calculating: 1.Bxd5! Alekhine describes as inferior three other moves: 1.Ng4? he mentions as “deserves attention, but doesn’t look so good.” The second part of this statement is more accurate, as 1...Nxg4 2.Nxd5 Bxd5 3.Bxe7 Bxa2 wins for Black. 14 1.Nxd5? Nxd5 he understandably says is unconvincing. Alekhine likes White after 1.Nxc6 bxc6. I would wish him good luck convincing Stockfish of that, who steadfastly gives 0.00, but that was less of a problem in 1938! 1...exd5 Alekhine correctly assesses this position as very good for White, but doesn’t actually commit to which move he thinks is best, proposing several possible ideas. Alekhine mentions 1...Bxd5? as being “no good”, without giving details. No doubt he had seen the simple variation 2.Bxf6 Bxf6 3.Nxd5 exd5 4.Nd7, winning the exchange. 1...Nxd5? is also accurately described as “no good”; indeed, 2.Nxd5 Bxd5 3.Bxe7 wins a piece for White. 2.Rfe1! This natural move is the most convincing path to a decisive advantage. 2.Ng4? is a mistake due to: 2...Nxg4! 3.Bxe7 Qc7! White can avoid the loss of a piece with 4.Qh3, but 4...Qxe7 5.Qxg4 Rce8 is only equal. 2.Qf3 Qd8 3.Rfe1 is an alternative winning continuation though, one important point being that 3...Ne4 4.Bh6 wins the exchange. Alekhine also mentioned 2.Nxc6 bxc6 3.Rfe1 but this seems less impressive. True, White retains 15 some initiative, but it feels too early to trade off Black’s passive bishop. 2...Bd8 2...Qxb2 3.Qf3 Kg7 4.Nd1 Qxd4 5.Nxc6 Rxc6 6.Rxe7 wins a piece. 3.b4! White keeps increasing the pressure. 3...a6 3...Kg7 4.b5 Be8 5.Qh3 threatens Bh6† in addition to the rook on c8. After the text move White has several promising options, but one good way to force matters is: 4.Qh3 Ra8 The rook has to give ground, as the natural 4...Rc7 runs into 5.Qh4!, creating serious problems due to the fact that the bishop on d8 is undefended. 4...Qc7 5.Bxf6 Bxf6 6.Nxd5 also wins. 5.Qf3 Kg7 Now there is a rather flashy combination available: 16 6.Bxf6† Bxf6 7.Qxf6†! Kxf6 8.Nd7† 8.Nxd5† Bxd5 9.Nd7† Kg7 10.Nxb6 Rad8 is less impressive. 8...Bxd7 9.Nxd5† Kg7 10.Nxb6 Rad8 11.d5 White should win the endgame. When de Groot conducted his famous research back in the 1940s, chess engines did not exist and the Elo rating system had not yet been developed. Thus it was impossible for de Groot to derive any mathematical relationship between a player’s strength and many of the other variables he measured during the puzzle-solving experiments. More than eighty years after de Groot’s first experiments, I have revisited his methodology, aided by Stockfish, a chess engine of superhuman ability, and Minitab, an advanced statistical software tool. Using new, previously unpublished chess puzzles, I have tried to finally uncover the exact differences between players of varying strengths. As I write this introduction now, in my apartment in Birmingham, having already conducted the full statistical analysis on the results from all the puzzles, I am excited to announce that the information I am about to disclose in this book will be of groundbreaking significance. It may forever change the way you think about top Grandmasters, and the way they choose their next move. Never before in the history of chess has anybody been able to unravel the mysteries behind the thinking process of a Super-Grandmaster, and accurately determine the differences between their chess thinking and ours. I hope you enjoy the book. Philip Hurtado Birmingham, December 2021 17 I didn’t know Phil before he sent an email via my website, enquiring if I would be interested in collaborating on a book. Having been intrigued by his idea, we had a long chat on the phone, and I was excited to join the project. Part of the initial appeal for me was that working with a co-author seemed less daunting than taking on an entire book project personally, but subsequently I became so interested in the subject that my input expanded considerably. I was attracted to the book’s concept for a few reasons, such as the fact that the majority of the test positions to be solved in the book were chosen by Phil. I thought these examples would be more useful for most readers, as opposed to material that a stronger player would themselves judge instructive. I felt this aspect important, as when talking socially with lower-rated players about my games, or theirs, I am frequently surprised – both by the, to me, difficult concepts that they understand, and by other areas where their understanding seems surprisingly lacking from my point of view. There are lots of books where GMs give their views on what they think other players need to know, but these may not always ask the questions people want answered. An additional attraction of this project was the way in which the puzzles were recorded, where you can observe participants’ thinking in real time, as they give their opinions, judgements and calculations about the position. I found these highly revealing, and subsequently enjoyed spending time reflecting on many of these comments. Phil’s distinguished academic career contrasts rather with mine. I headed off to the World Junior Championships in Australia in 1988, after doing my exams at sixteen. Professional chess seemed a lot more appealing than heading back to school, and I never really returned to my studies. Despite our different educational backgrounds and playing strengths, there was quite a bit of overlap in the key themes that Phil and I identified when analysing the results. (We initially wrote those sections independently, and only later compared our thoughts.) I am particularly happy that after a lot of calculation, Phil has supplied an answer, albeit with some 18 caveats, to the tricky question “How many moves ahead can a Grandmaster see?” If I had a pound for every time I have been asked this question I would have retired long ago – although as I’ve never had the slightest idea about even a ballpark answer until now, perhaps justice was done. To bring the project to a conclusion, I expanded considerably on the written material, reviewed the eyetracker footage and carried out final editing on most of the book. Phil spent a lot of time on his main responsibilities: fine-tuning the scoring system; compiling and analysing the puzzle data; and crunching the numbers to draw conclusions, such as to the question noted above. We both added a lot of content to sections that are not attributed to a main author. Earlier on in the process, I had suggested the idea of introducing the extra variable of monitoring participants’ eye movements as they solved the puzzles. Phil also found the idea interesting, and he was able to set up an appropriate experiment. This was a fascinating day to take part in, as, in addition to analysing my own results and those of others, it was intriguing to observe others thinking in real time. It was also absorbing to inspect the areas of the board that other players spent a long time focusing on. I solved the puzzles in June 2019 and, in my next event a month later, won the British Championship for the seventh time. I hope this book will help you towards similarly satisfying results in the future. Michael Adams Taunton, December 2021 19  by Philip Hurtado Although De Groot’s work certainly served as an inspiration for writing this book, the approach I have used for collecting data and determining differences in thinking between a Super-GM and lower-rated players has been radically different. De Groot focused on the cognitive requirements and thought process involved in selecting a move, and in doing so, discovered that a chess player’s cogitations go through four distinct stages prior to making their move. Namely: * The ‘orientation phase’, in which the player assesses the situation and forms a general idea of what to do next. * The ‘exploration phase’, when the player looks into the possible variations and lines. * The ‘investigation phase’, in which the player chooses the best move. * And finally the ‘proof phase’, where the player confirms that the results of the investigation are valid. In my experimental design, I focused primarily on numerical data which could be analysed, such as: * Time taken to make a move * Time taken to spot the best move (not including time spent checking it, as per the ‘proof phase’ above) * Evaluation of the position (in the familiar plus or minus numerical style of most analysis engines) 20 I wanted to determine mathematically if there were any significant statistical differences in the above metrics between players of various levels. Would there be any demonstrable correlations between these metrics and a player’s FIDE rating? Statistical Analysis Without getting too technical, it is worth explaining what a statistically significant difference means, as this concept is fundamental for understanding the methodology used in this experiment. At the same time, you do not have to read the following section to start using this book. I have included it for the benefit of those readers who are curious about the topic. If, on the other hand, you just want to get on with the main topic of the chess content, then feel free to skip the next few pages. Suppose I wanted to compare the evaluation ability of Super-GM Michael Adams with that of my friend Andrew McCumiskey, an experienced club player with an Elo rating just above 2000. Using Stockfish’s evaluation as a reference point, I found that Michael evaluated the test positions with an average accuracy of plus or minus 0.34 away from Stockfish, whereas Andrew evaluated the exact same positions with an average accuracy of 1.05 away from the machine. How do I know that these differences are meaningful, and not simply due to Michael Adams having a good day and Andrew McCumiskey having a bad day? Well, the solution to this problem is not at all trivial. You see, humans have been able to calculate the average of things and compare them to each other for thousands of years – ancient Greek and ancient Egyptian mathematicians are among those who are known to have done so. But knowing whether or not these statistical averages are significant or not – i.e. whether the result is due to a real difference as opposed to the randomness of nature – was not fully understood until 1935, when British statistician Sir Ronald Fisher published the first edition of the book The Design of Experiments, in which he introduced for the first time the concept of the null hypothesis, and explained the rationale for using significance tests and probability in order to properly analyse the results from an experiment. The Tea-Tasting Lady One of the first, and probably still the most famous example on how to use the null hypothesis to determine the validity of the results from an experiment, is the funny yet brilliant “Lady Tasting Tea” experiment, which was conducted by Ronald Fisher himself and subsequently reported in the aforementioned book in 1935. I will recount it here, as it is one of my favourite stories in statistics and I find it fascinating. The lady in question was Miss Muriel Bristol, a PhD researcher who worked closely with Ronald Fisher in the same agricultural research facility in Hertfordshire, England. One day Fisher brought tea 21 to Miss Muriel, who refused the tea after tasting it because she said he had poured the milk in the cup before the tea. Fisher argued that there was no difference in taste caused by pouring the milk before or after. Miss Muriel disagreed. To settle the matter, Fisher devised an experiment that would conclusively prove whether or not Miss Muriel could taste the difference. After some calculations, he determined that eight cups of tea (four with milk poured in before, and the other four with the milk poured after) would be enough for the test. With this many cups, Fischer calculated that the probability of a person without the tasting ability correctly identifying all four of the ‘wrong’ cups by guesswork alone was only 1/70, or 1.4%. As it happened, Miss Muriel was right. She correctly identified all four of the cups in which Fisher had poured the milk into the cup before adding the tea. Fisher acknowledged defeat, as he knew that there was only a 1.4% chance that Miss Muriel could have achieved that result by guesswork. To put it another way, in layman’s terms, there was a 98.6% chance that she really could taste the difference as she claimed. I hope you enjoyed the story. The important lesson here is that Fisher introduced for the first time the concept of using probability to determine whether or not the results of an experiment are statistically significant. This probability of the results happening by chance is known in statistics as the p-value. Ronald Fisher postulated that for a statistical test to be significant, the p-value would have to be lower than 5%. In the case of the Lady Tasting Tea, the p-value was 1.4%. Going back to our original question regarding the accuracy which with GM Michael Adams and club player Andrew McCumiskey can evaluate a position: what was the p-value, you may wonder? The p-value of the data in our Adams – McCumiskey experiment is 2.4%. In other words, there is only a 2.4% chance that such a difference in performance could have come about by random chance, or by one player having a good day and the other a bad day. Sorry Andrew – if it’s any consolation, my own results were about the same. Calculating p-values is not easy – they would take months to calculate by hand, and it is only thanks to advanced statistical software that these values can now be automatically calculated. Without going into details, I will just mention that p-values depend a lot on the sample size and the variability of the results. Which brings us to our next section. Sample Size In the same way as Ronald Fisher determined he needed eight cups of tea to prove to the desired level of certainty if Miss Muriel could taste the difference in tea, I needed to determine how many puzzles and how many participants I would need in order to draw meaningful conclusions from the data I intended to analyse. By running a few simulations on the computer, I discovered that I needed at least twenty puzzles and twenty participants. Moreover, the playing strength of the participants would have be normally distributed from absolute amateur (Elo 1000) to Super-Grandmaster (Elo 2700). 22 To be on the safe side, I decided to go overboard on the sample size. I doubled the number of puzzles to forty, and gathered data from nearly fifty participants, from whom I managed to collect a total of 1,100 commentaries over the course of nine months. At this point it is worth mentioning that at the end of the experiment, Michael and I decided to replace ten of the original puzzles with ten puzzles of equivalent difficulty taken from Michael’s own games, as we thought it would be appropriate to include some real-life examples from the Super-GM himself. Statistical data for these ten puzzles of Michael were also taken and added to the study. Methodology From the onset of the experiment I was not sure what my findings would be, so I simply set out to record as much data as I could. Each participant was invited to solve all forty puzzles, using as much time as they needed for each puzzle. Participants were asked to talk through their thought processes before deciding on a move and giving their evaluation of the position. I noted down all the commentary, along with the total thinking time at the end of each comment. I finally recorded their chosen move and estimated evaluation, which would be measured for accuracy using the evaluation of Stockfish as a reference point. It is worth mentioning that, in order to keep the book manageable in size, only a small minority of the commented solutions were selected for inclusion in the book. In the solution section for each puzzle, we generally included roughly four to six of the most interesting and instructive commented solutions from a range of participants, showing the differences in thought processes between players of varying playing strengths. Data Analysis After gathering the necessary data, I ran various statistical analyses on the computer to look for significant differences between players of different levels. I noticed that there were no significant differences whatsoever between players of similar rating. But I quickly found out that there were clear statistical differences approximately every 200 Elo points. So for the analysis I grouped the participants in tranches of 200 Elo points (1600-1800, 1800-2000, etc.) and ran new tests comparing the metrics of these groups. One particularly revealing set of tests was the correlation tests, demonstrating a clear linear relationship between a player’s Elo rating and the time needed to find the best move, as well as the same relationship between playing strength and evaluation skills. As I continued number-crunching the data on the computer, additional interesting findings were fortuitously discovered. For example, I discovered that there is a clear correlation between a player’s strength and the depth at which he or she is able to accurately calculate. I was also especially 23 interested in the time it took players of different rating levels to find the correct move, and then comparing this to the time it took them to finalize their decision to play the move. It came as a surprise to me that stronger players actually spend more time finalizing their decision than they spend finding the best move in the first place. All these findings will be summarized in graph or table format and discussed in more detail towards the end of the book. 24 FIDE (Fédération Internationale des Échecs) has not yet issued an official title of Super-Grandmaster. However, among the chess public, Super-Grandmaster is used as an unofficial title to distinguish the very top players – usually those with an Elo rating of 2700 or above. About the Elo rating system Almost all chess players are familiar with the term “Elo rating” and we will certainly see a lot more of it in this book – but it’s worth saying a few words about what this number actually signifies. The Elo rating system, which is used by FIDE to rate all chess players, was first introduced in 1960 by Arpad Elo, a Hungarian-American chess master and physics professor. The mathematical model he devised is so beautiful and effective that it has been widely adopted for rating other games and applications. Some other examples that use the Elo rating system are the FIFA World Rankings and the Universal Tennis Rankings. It was even used by Mark Zuckerberg to rank the popularity of female Harvard students in the early days of Facebook – and it is used today by Tinder to rank and match user profiles. The Elo rating system in chess is based on the probabilistic outcome of a game played between players of different strengths. The bigger the difference in Elo rating, the greater the expected percentage score of the higher-rated player. For example, a rating gap of 100 points equates to an expected score of 64% for the stronger player. If the rating gap is 200 points, the expected score of the stronger player increases to 76%. Obviously the probability in any individual game will be affected by which player has the white or black pieces, so the above figures are merely a broad average. Another interesting feature is that the drawing percentage increases in line with ratings. For example, when the players are rated 1700 and 1800, the probability of a draw is 22%. For players rated 2500 and 2600, the draw probability increases to 42%. We could spend many pages analysing this and other statistical data relating to Elo, 25 but to do so would take us a bit far off-topic. I will, however, leave you with one last observation that I took from the probability tables: a player with a GM-calibre rating of 2500 only has an 8% probability of defeating a Super-GM with a rating of 2700. The difference between GM and Super-GM The requirements to become a Grandmaster are for the player to attain a 2500 Elo rating along with three ‘GM norms’, each of which requires an Elo performance of 2600 or above. As of July 2021, out of the 367,397 players alive and registered by FIDE, there are only 1,730 GMs, accounting for just 0.47% of all rated players. Clearly, becoming a Grandmaster is, in itself, a mammoth accomplishment. As mentioned earlier, the informal title of Super-Grandmaster is generally reserved for those who attain an Elo rating of 2700 or above. This is a feat attained by just thirty-eight players on the current rating list: that is just 0.01% of all rated players worldwide. For a visual representation of how extraordinary it is to become a Super-GM, look at the worldwide distribution of chess players below, and see how few individuals are capable of playing at or above the 2700 level. Histogram showing number of players in each rating category. The average chess player has a rating of 1750 Elo points. A Super- Grandmaster has a rating of 2700. The featured Super-Grandmaster in this book is co-author Michael Adams. Michael became a grandmaster in 1989 at the age of seventeen, and he has been consistently rated above 2700 level for over twenty years. (His rating dropped below 2700 for short periods in 2009 and 2019/20, but these were brief blips in a long career playing at the highest level.) Michael was ranked fourth in the world in several rating lists from 2000 to 2002, and he attained a 26 peak Elo rating of 2761 in both 2013 and 2017. He reached the final of the FIDE World Championship knockout in 2004, eventually losing in a rapid tie-break. In short, Michael has achieved almost everything there is to achieve in chess, short of becoming World Champion – and he has been competing at this elite level for more than two decades. 27  by Philip Hurtado This book contains forty challenging puzzles for you to solve and monitor your own score. In addition, there are eight bonus puzzles and seven further non-scoring puzzles in the Eyetracker chapter. These are chess positions which you should set up using a real board and pieces, and imagine you are playing a competitive game against a tough opponent. The instructions are simply to select the best possible move you can, and give an evaluation of the resulting position. For each of the forty main puzzles, you will score a maximum of ten points if you find the best move, and an additional three points if you give the correct evaluation as well. You need to find the best move in order to be eligible for any bonus assessment points. The full scoring system for each puzzle will be presented at the end of the solution section for that puzzle. Time to solve a puzzle There is no set time limit per puzzle. However, to mimic tournament playing conditions, we suggest that you aim to solve each puzzle in fifteen minutes or less. You can certainly use a bit more time in some of the more complicated puzzles if you feel it is necessary, as you would in a tournament game. Our panel of solvers spent an average of eight minutes per puzzle, although some will be solved quite quickly while others encourage deeper thought. Puzzle difficulty The puzzles have varying degrees of difficulty, which we have assessed on a scale with 1 being easiest and 10 being hardest. The average difficulty level of the forty main puzzles is 7. 28  When and how should I check my solutions? We recommend checking the solution for each puzzle soon after solving it, as you will gain the most value from the thoughts of others, plus the analysis and comments on that position directly afterwards, when your own ideas are fresh in your mind. It is important to remember that the solution section of this book is completely different from most puzzle books. In an ordinary puzzle book, each solution will contain a few key variations and comments. In this book, the solution for each puzzle starts by recapping the thought processes of several players from the panel of solvers, before moving on to the model solution with detailed commentary from Michael, followed by concluding comments from both authors. How many puzzles should I solve in succession? If this was a normal puzzle book, we would recommend solving five, ten or possibly more puzzles in one sitting before checking the solutions. However, in this book the solution section is packed with a lot more content than you would normally expect. Here you will be able to compare your thought processes with those of a range of other players who solved the same puzzle, before seeing Michael’s detailed analysis and comments – but you will need to have enough energy and alertness to take all this in. It’s up to you how many puzzles you solve in each sitting, but our main piece of advice is to avoid solving to the point of exhaustion, as in that case you will fail to get the most out of the solutions: you will either be too tired to follow the thought processes of the other solvers, or you will have to take a decent break before checking the solutions, by which time your own thoughts about each puzzle will not be as fresh in your mind. To begin, you may wish to solve just one or two puzzles and then check the solutions before going any further. That way, you will get a sense of what you can expect from the solution section in order to gauge how many puzzles and solutions you will be able to manage per session. Do I score points even if I don’t find the best move? Yes. Each puzzle solution has a unique scoring system. The maximum number of points you can get for solving a puzzle correctly is ten points. However, in most puzzles you will still earn a certain number of points for the second or third best move, depending on the position. Why is the minimum score set at 1 point? You may notice in the solutions that even weak moves are still awarded a single point. Why is this? First, it solves the problem of not having to enumerate the whole list of moves that have a similarly 29 low value. And secondly, it conveniently helps us align each player’s performance to the current Elo rating system. Note that the Elo system generally starts at 1000, rather than at 0. In other words, if a player fails to get a single puzzle correct, that player will still end up with a points total in line with a beginner-level 1000 Elo rating. How do the evaluation points work? Once you have decided upon a move, you should assess the resulting position and settle on an evaluation. A player can score up to three bonus points if, after having chosen the best move, they have also evaluated the position correctly by giving an assessment close to that of the chess engine. Most readers will already be familiar with the way that computer engines evaluate positions, having got used to consulting the electronic oracle about their own games. Some, however, might not be so accustomed to working with computers, and with how to express a position in the same numerical terms as modern chess engines do. In that case, in order to give your assessments on the positions, the guidance below will help you to translate your views on the position into a numerical assessment. There is a range built into the scoring system, so absolute precision is not required. Computer numerical evaluation Human evaluation Between –0.09 and +0.09 The position is equal or drawn. Between +0.1 and +0.29 More pleasant for White, but nothing too serious yet. White is slightly better; or White is a pawn up, but Black Between +0.30 and +0.59 has some compensation. Between +0.6 and +0.89 White is significantly better. White has a significant to large advantage, but not Between +0.9 and +1.39 necessarily decisive yet. Between +1.4 and +2.49 White has a substantial, probably winning advantage. More than +2.5 White is clearly winning. You will use the same evaluation criteria, but with a minus sign, for describing an advantage to Black. You can score between one and three bonus points according to how closely your evaluation matches that of the engine. Full details will be given in the score table at the end of the solution for each puzzle. There are no evaluation points if you totally misjudged the position. It is worth repeating that the evaluation points only apply if you have played the best move highlighted in bold, on the puzzle solutions page. You cannot award yourself evaluation points if you did not select the best move. 30  How were the computer assessments arrived at? Some of you may investigate the puzzle positions with your own versions of Stockfish and/or other engines. Depending on many variables, you may well get slightly different assessments to mine. The power of your computer, the amount of thinking time you allow it, the depth of moves considered, the version of Stockfish used and various ways of tweaking the program settings are just some of the factors that can affect things. However, I believe any differences in evaluation will be relatively minor, and definitely should not change the assessment as regards to which move is the best. After several tests, I decided to use a consistent methodology, which gave an accurate and stable evaluation after running all the positions on my computer for a long time. For the evaluation of all the puzzles in the book, I have used the numbers provided by Stockfish 10+ WASM at the publicly available website Lichess.org, running at the default depth of 23 half moves deep and 5 best continuations, and then corroborating the results at a much greater depth of 45 moves on my own machine. In the event of a discrepancy in evaluation, the average of both evaluations was taken, where that was appropriate. All the chess puzzles in this book have been solved by Michael, in my presence – and thanks to the data collected, we have been able to gain new insights into his thought process. Analysing the metrics of Michael’s puzzle-solving performance sheds new light on the mystery of what makes Super-GMs such outstanding players. You will now be able to solve the puzzles yourself, before comparing your answers with his and discovering some of the secrets of Super-GM thinking. 31  by Michael Adams How to score well and gain maximum benefit from the puzzles 1. Some of these puzzles are challenging, and will be especially difficult as you are coming to the position cold, and can’t follow the thread of your thinking as during a game. In this situation it is often not easy to immediately get to full concentration levels, so try to take some time to orientate yourself. 2. One of the biggest and most common mistakes is to fail to notice a move in the starting position, or reject one too swiftly, or to forget about a promising candidate move you previously considered in a fog of variations. In most of the puzzles that I failed badly on, I was guilty of one or the other. It is a common failing in chess players to get carried away with calculating long lines, while totally missing a stronger possibility available on the first move or two. As Bent Larsen once said: “Long variation, wrong variation.” 3. Looking for a puzzle-like solution won’t always be the correct approach, so try to consider each position normally as you would in a game. 4. When you have identified the best move, bear in mind that finding the best reply will be crucial in accurately giving an assessment of the position. Stronger players often outperform by getting the full picture on the board by being more aware of their opponent’s resources. 5. Perseverance will often pay dividends. I and other GMs took much longer than you might expect to come up with the answers to some relatively simple puzzles – but by soldiering on, we arrived at the correct solution to those positions in the end. Trust your intuition, and don’t give up on your instinctive ideas without serious thought. 6. To score points, only the first move and assessment are required, but the stronger players nearly 32 always gave a more detailed, concrete line of best play to justify their judgement. Trying to do this yourself will certainly prove rewarding, and you will be able to compare your thoughts to other solvers in the answers. The deeper you dive into a position, the stronger you will become. After Phil had tested me on several of his own games, I was able to return the favour. So you will be able to see if you can match, and perhaps exceed, my performance in those positions. Sadly, yet again I didn’t get everything right, either over the board personally, or when solving Phil’s puzzles. Good luck! 33  by Philip Hurtado Before moving on to the solutions, I would like to give a few more details about the solvers featured in the book. They are arranged in descending order of Elo rating (apart from the final four, who are grouped together due to being from the same school). GM Michael Adams – Elo 2701 The English gentleman of chess, Michael was a World Championship finalist in Tripoli in 2004, and has been a World Championship Candidate several times. He has won the British Championship seven times, and was ranked number four in the world at his peak. The first time I saw Mickey playing live was in the early 1990s, at a weekend rapidplay in London. It was a pleasure to see him play. Many of the amateurs competing would finish their games early, just to go and watch him play on first board, from where he would never move. Because of his style of play, his moves were unfathomable to us. There were no direct blows, so amateurs never knew what was going on until the very end, when suddenly, at some point later in the game, his opponent’s position collapsed like a house of cards. Now, thirty years later, I have the opportunity to find out how he did it! GM Julio Granda Zuniga – Elo 2630 Julio was born in Camana, a historic town conquered by the Spanish in 1539, situated on the South Pacific coast of Peru. He was brought up on a farm, where his father taught him chess at the age of five using an old book series written by Argentinean master Roberto Grau. Julio is considered an 34 incredible natural talent, since it took him many years to begin using modern computers to study chess – long after his rivals were using them extensively. He has won the American Continental Championship five times, and was World Senior (age 50+) Champion in 2017. After many years living in Spain, and playing actively, Julio has returned to his roots in Camana. He has bought his own farm where he rears free-range chickens, as well as growing his own organic orchards of apricots, plums, peaches and other stone fruits. GM Eduardo Iturrizaga – Elo 2607 Born in Caracas, Venezuela in 1989, Eduardo became a grandmaster in 2008, the first Venezuelan to achieve the title. He scored an amazing 97% in the puzzle tests. GM Renier Vazquez – Elo 2551 Renier moved from Cuba to Madrid where he arrived empty-handed, with no material possessions whatsoever. But his head was packed with the highly valued training methods used at the Cuban school of chess. Cuba has produced one of the highest number of grandmasters per capita in the world; forty-three Grandmasters from a population of only 11 million. In Cuba, talented players are scouted to join the famed high performance schools for chess and sports. When Renier arrived in Madrid he joined our chess club in Collado Villalba, where he coached the first team, helping us gain promotion to the top division of the Madrid League, and eventually to win it six years in a row! IM Juan Reyes – Elo 2451 Juan is an International Master from Peru. I first met him more than twenty years ago in Madrid, where he gave me and a friend private chess lessons. It was thanks to those training sessions that I reached my peak Elo rating of 2112. GM Keith Arkell – Elo 2429 A highly likeable, sociable Grandmaster, with superb intuition and endgame technique, Keith has won countless weekend tournaments in the UK. He was English Champion in 2008, losing a playoff for the British Championship after sharing first with Stuart Conquest. Other personal highlights, showing how he has maintained his playing strength over a long and successful career, were winning the European Senior (age 50+) Championship in 2014, and sharing first (placing second on tiebreak) in the World Senior (50+) event in the same year. Keith’s great endgame technique comes out to shine when you play him in blitz. When we played in a pub, after the eyetracker puzzle solving, Keith won five consecutive games against me giving time odds of three minutes to one. That’s how quick he is! IM Harriet Hunt – Elo 2414 Harriet is an International Master (as well as Women’s Grandmaster), and has won the British Women’s Championship five times. She made her debut for the English women’s team at the age of sixteen, and has achieved excellent results for the team, including the best performance on Board 1 at the European Women’s Team Championship at Batumi 1999. Harriet’s academic career, where she is a research associate at the McDonald Institute of 35 Archaeological Research at Cambridge University, has limited the amount of chess she has played in recent years. However, she was back in fine form when representing England again at the FIDE Online Olympiad in 2020. Anonymous – Elo 2305 A professional who works in the finance industry and plays chess as a hobby, this individual wishes to remain anonymous. During his university years he reached a rating close to 2400, yet has never registered for the FIDE Master title. I can confirm that he is capable of analysing chess well above the level of his current rating, which was demonstrated by his impressive scores in the puzzles. Sebastian Reyes – Elo 2272 Sebastian is the nephew of IM Juan Reyes. Juan told me I should test the puzzles with Sebastian, who was then champion of Madrid in his age category. He was 100 Elo points lower when he did the puzzles last year, and I expect his rating will continue to climb with his uncle’s support. FM Akshaya Kalaiyalahan – Elo 2149 By the age of eighteen, Akshaya had already won the British Women’s Chess Championship twice, as well as representing England at two Olympiads. An intelligent thinker with an instinctive understanding of chess, she is currently studying law at Oxford. WFM Louise Head – Elo 2090 Louise was a top scorer in our team when playing for Solihull Chess Club. She won the English Women’s Chess Championship in 2017 and 2019, and was a member of the English Women’s team at the 2018 Chess Olympiad. Louise is currently doing a PhD in Physics at the University of Edinburgh, one of the UK’s most prestigious universities. As I write these biographies, I am beginning to realize there may be a correlation between playing chess and making it to an elite university. Jesús Cao – Elo 2060 There are people you meet in life who, for very distinct reasons, you immediately trust and admire. Jesús was one of them. I met him for the first time when I joined a chess club in Madrid in the late 1990s. He had a handsome appearance, white hair, glasses, and always wore a nicely fitting suit and tie. Apart from being President of our club, he was a permanent member of the Board of the Bank of Spain, and you could feel the confidence and tranquillity that made him an ideal fit for such a senior position. Jesús has always been very kind to chess. He has generously contributed to all the chess clubs he has presided over, and has helped many foreign players from Latin America to settle in Madrid, offering them classes, simuls, and in some cases, personal financial help. Andrew McCumiskey – Elo 2017 A great, intelligent guy, Andrew was one the first participants in the puzzle experiments. We played together at Solihull Chess Club in the first division, the year our club won the Birmingham and 36 District Chess League. Andrew works as an IT Consultant, specializing in Automotive, and is married with four sons. He learned chess at the age of eleven and has played regularly for over forty years. He rarely studies openings, but relishes long subtle endgames. He is a great believer that chess helps with both concentration levels and decision-making in the wider world. Nathanael Paul – Elo 1993 Nathanael is a talented club player with great fighting spirit, who won’t concede a draw until only kings are left on the board. I was impressed with his determination when we played at the 4NCL, the game ending in a combative draw. Philip Hurtado – Elo 1924 I was born in Brighton, England, but then moved to Madrid at the age of four. I became fascinated with chess when I read William Hartston’s book on chess openings, and recall how accurately he described the way in which beginners bring out their queen too early – exactly as I used to do as a teenager when playing against my uncle Rufino at his restaurant in New York. Taran Jina – Elo 1912 Taran first started playing aged six, and went on to represent England in several junior team events across Europe. An excellent student, he went to Warwick University to study Mathematics, where he is the president of the university’s Chess Society. Away from the board, he is keen fan of football and cricket, playing both in his free time. Tony Sadler – Elo 1893 Tony was the first person I played when I joined my local chess club. He has been a stalwart member of Solihull Chess Club in the Birmingham and Leamington League for many years. After he lost to me in a friendly game, he generously recommended that I should play a board above him in the team. Francisco Marcos – Elo 1842 Paco, which is the hypocorism of Francisco, is a FIDE arbiter. He graduated in Computer Science at the Polytechnic University of Madrid, and worked many years as an IT consultant for the Spanish Ministry of Defence. His two passions are chess and cryptocurrencies. Paco was one of the early bitcoin miners in Spain using ASIC computers, and has second-to-none knowledge in this field. Raymond Carpenter – Elo 1767 Raymond is one of the founders of the Solihull Chess Club, and captain of the first team who won the Birmingham Chess League. Due to Covid restrictions, Ray still has the cup at his home. Ray is a very likeable person, as well as a tough competitor over the board. Agnieszka Milewska – Elo 1677 Aga is a chess teacher and coordinator at the Chess in Schools and Communities Charity, as well as an International Arbiter. She is also now an ECF Director of Women’s Chess. A hard-working single mother, and a good trustworthy friend whom you can speak openly to. A word of warning: Aga’s rating seems to increase when she drinks vodka, as I discovered when 37 I played some blitz games against her at the Royal Automobile Club in London after having dinner with some friends. Jorge Granell – Elo 1673 Jorge is the President of the “El Ocho” chess club, a club which I used to play for in Valencia, Spain. A lovely person with a passion for chess, he is always looking for ways to promote chess in schools and nurture new talents. In Jorge’s own words: “I am grateful to Philip for letting me participate in this book. I also want to encourage the readers to inspire youngsters to take up chess as a learning tool that helps us think strategically throughout life.” John Green – Elo 1563 John is the President of Solihull Chess Club and a former police officer, as well as a loyal friend. Highly inquisitive, he is always looking at ways of promoting chess locally, and recruiting younger people to play the game. Every time I speak to him on the phone, he prompts me to ask Michael if he wants to join Solihull Chess Club! Fermin Rodriguez – Elo 1544 Fermin is a good friend of mine. I met him for the first time when I was seventeen, at the very first chess tournament I played in – we were both young then! Fermin was carrying a book on pawn endings by Ilya Maizelis, which I’m not sure he ever actually read, and he gave it to me in an act of friendship and generosity. Fermin has a prodigious memory, and has tons of chess-related stories he can recount in great detail. One memorable tale involves his loss against Spassky at the Oviedo tournament in 1991, where Spassky beat him playing the Hippopotamus Defence in 23 moves. Fermin’s great memory may have been a mixed blessing for his chess, as practically all of his study has centred on opening theory. He plays the first dozen or so moves like a Grandmaster, the middlegame like an amateur and the endgame... well, I don’t think he has won an endgame in years. Penny Wood – Elo 1443 Penny plays at my chess club in Solihull. I asked her to say something about herself for the book: “Married, two daughters, four grandchildren. Admin officer in local prison for two decades; can’t wait to retire. Likes fine chocolate, good coffee and reading in cafes. Involved for many years in the lower echelons of club play.” She then wittily added: “Once drew with Bill Hartston, but I think he was being polite.” Leo Hurtado – Elo 1345 My beloved son. Leo has only ever played in one tournament: the FIDE World Amateur Chess Championship in Cagliari, Italy. Chess is not a particular passion of Leo’s, but he enjoys playing casual games against his friends or cousins. He is nineteen years old, and is studying Computer Science and Artificial Intelligence at university. Oscar – Elo 985 38 Osky was my first chess student when I migrated from Spain back to England in 2016. He is eight years old, and has now graduated to work with stronger coaches. I would be surprised if he doesn’t win one of the junior sections at the British Championships in the coming years. Victor – unrated I don’t know why, but I’ve somehow always been privileged in meeting extraordinarily intelligent people. This thirteen-year-old chap of Russian origin has the deep, penetrating, inquisitive eyes of a genius. I had the pleasure of teaching him and his siblings how to play chess. Theo – unrated Oh Theo! A dancer, a poet, a connoisseur – and he is only eight! Theo is class. One of my favourite moments during our chess lessons was when tea and freshly baked biscuits were served. If you ever have afternoon tea with Theo, you will know the difference between eating and savouring! Alex O’Loghlen – unrated Alex was born in Cape Town on December 1st, 2008, and now lives with his family in Dublin, Ireland. He first started playing chess at the age of five, taught by his dad. He enjoys playing against his grandfather. Apart from chess, Alex enjoys spending time with his friends, playing guitar and rugby. My distinguished co-author is a Chess Ambassador for the Stephens and George Charitable Trust based in Merthyr Tydfil. In the summer of 2019 we both visited Caedraw Primary School where Michael played a simul, and I tested the players below on some of the puzzles. Anshul Swamy – Elo 1300 Anshul was introduced to the game at St Mary’s Primary School at the age of seven, and reached the Mega Finals of the Delancey UK Chess Challenge that same year. He has enjoyed repeated success in that competition, qualifying for successive Giga Final stages every year thereafter. Now aged thirteen, he has received tuition from former Welsh Champion and Olympiad team member Tim Kett and his wife Sarah, as well as Russell Doddington, and is currently mentored by Pete Heaven. Anshul is a member of the Welsh junior national squad, and has frequently represented the Welsh Chess Union. Alex Lawrence – Elo 1285 Alex is twelve years old, the captain of the junior team at his local club, and a member of the Welsh Under-14 Team. He has won a number of junior chess congresses, and has recently graduated to his first senior congress. His puzzle-solving ability impressed me greatly, as did his resilient play against Michael in the simul. His current Elo rating, just one year after he solved the puzzles, is an impressive 1673. Tabitha Ryan – Elo 1100 Tabitha is ten years old, and the captain of St Mary’s Primary School chess team in Merthyr Tydfil. She is the current Welsh Under-11 Girls Champion, and a member of the Welsh Under-12 Team. 39 Ioan James – unrated Ioan is ten years old. He has only recently learned how to play chess, and has enjoyed playing in internal school competitions. He is also a member of his school team in Merthyr Tydfil in South Wales, and has begun playing in local tournaments. Many thanks to all these participants, for putting in the time and effort to solve the puzzles that appear in this book, and revealing their thinking process throughout. As a reminder for readers, each player’s Elo ratings or equivalents will be listed alongside their answers throughout the puzzle solutions. 40 41 42  Puzzle 1 Black to play – Level 7 Show/Hide Solution Puzzle 1 – Deceptively Difficult We have equal material. 00:20 I have a nice passed pawn in the middle of the board. 43 00:45 I could play maybe 1...Re4. 01:30 He can take my a5-pawn. I could move my king up, 1...Kg6 to have a go at the g-pawn. 1...Kg6 seems a bit feeble. White can just grab my pawn on a5. 02:00 1...Kg6 with the idea of Kf5 and Ke4 supporting the d-pawn. 02:40 Moving the king up and getting him into action is probably not a bad idea. 03:00 1...Re2 is no good. The rook is not going anywhere. Maybe I can place the rook behind the d- pawn. 03:40 But I am giving up the a-pawn and leaving the e-file which does not look like a great idea. 04:00 1...Rd8 is possible. The pawn needs a friend behind it. 04:40 Yeah. I would probably go for 1...Rd8. Evaluation: Black has a chance here. (–1.0) Counting pawns: five against five. My a5-pawn is threatened. 00:30 I can go and grab his g5-pawn. 01:20 I absolutely need to defend my a-pawn: 1...Ra8, but 2.b4! doesn’t look good for me! So that is not great. Let’s see some alternative candidate moves. 02:00 The other plan is 1...Rd8, putting the rook behind the pawn, trying to push it up quickly. 03:30 1...Rd8 2.Rxa5, and now I am thinking of 2...d4. This is another good position you picked, Phil. 1...Rd8 2.Rxa5 d4 3.Ra4 d3 And White cannot get back to stop it. 04:00 So White cannot take. Maybe he has to play 2.Rd4 to blockade the advance. What about 1...Re5? 05:00 1...Re5 2.Rxa5 d4 3.Ra4 And now I can push d3, and his rook cannot come behind it. 07:00 I like 1...Re5; also it threatens to take his g-pawn. 07:30 Not sure whether to play 1...Rd8 or 1...Re5. This position is harder on a screen. I am much stronger over the board. 07:45 1...Re5 is my favourite. I think I am winning. He probably cannot take my a-pawn. And I can take his g-pawn. Evaluation: –3.0 44 Okay. The pawn on a5 is under attack. 00:40 If 1...Ra8, 2.b4 looks problematic. Black needs to play something active. 00:50 Maybe 1...Rd8, in order to push the pawn forward. 1...Rd8 2.Kg3 d4 01:15 The king needs to get over, to stop the pawn. Is 1...Re4 possible? 02:00 No, it doesn’t work. White just takes on a5. What about 1...Kg6? 02:30 1...Kg6 2.Rxa5 Kxg5 and it is not so clear. 03:00 1...Rd8 2.Rd4 Kg6 and White is in trouble. 03:30 1...Rd8 2.Kg3 d4 3.Kf2 d3 4.Ke1 Rd5 5.Rxa5 Rxg5 looks good for Black. 04:00 1...Rd8 2.Rxa5 d4 and it’s difficult to stop the pawn. 04:30 The key line is: 1...Rd8 Preparing the advance of the pawn. 2.Kg3 (or 2.Kg1) 2...d4 3.Kf2 Rd5 And Black is better. 04:45 I think it is clear that 1...Rd8 is the best move. Black is better. Evaluation: –1.1 Deeper Analysis by Michael Adams N.N. – Philip Hurtado Internet (blitz) 2018 45  Black’s main asset is the protected passed d-pawn, but the protected factor isn’t so important with rooks on the board. There are lots of other loose pawns, so activity is paramount. 1...Rd8! It’s normally good endgame technique to place a rook behind a passed pawn, and it is the most challenging approach here. Neither 1...Re2? 2.Rxa5 nor 1...Re4? 2.Rxa5 d4 3.Ra8 makes any sense for Black. 1...Ra8 gives White a choice of decent responses: a) The obvious 2.b4 was mentioned by some participants but is less effective than it looks – although White should not be in danger. 2...d4 3.bxa5 (3.b5!? cxb5 4.Rxd4 is equally playable) 3...Rd8 4.Rb4 d3 5.Rb1 Kg6 6.a6 Kxg5 7.a7 Ra8 8.Rd1 Rxa7 9.Rxd3 Rxa2 10.Rd6 Ra6 Black has an extra pawn but can hardly hope to win with such a passive rook. b) 2.Kg3! Kg6 3.Kf4 is simple and good, with no worries at all for White. 1...Re5 is met by: 2.Rxa5 d4 3.Ra4 Rd5 (if 3...d3 4.Rd4 Rd5 5.Rxd5 cxd5 6.c6 both sides get a new queen and the endgame is a draw) 4.Rc4 The rook is in time to block the d-pawn. 4...d3 5.Rc1 Kg6 6.Kg3 Kxg5 7.Kf3 White is out of danger. 1...Kg6 Improving Black’s king was played by Phil in the game, and creates some practical chances. Play continued: 2.Rxa5 Kxg5 3.Ra6 Rc8 3...Re6! makes things trickier for White, who requires the tactical solution 4.b4! to create sufficient counterplay. Play continues 4...d4 5.b5! cxb5 (5...d3? 6.bxc6 d2 7.c7 d1=Q 8.c8=Q gives White fine winning chances) 6.Rxe6 fxe6 7.c6 d3 8.c7 d2 9.c8=Q d1=Q 10.Qxe6 with a level queen endgame. 46 4.Ra7 4.Kg3! would have avoided trouble. 4...Rd8 4...Kf4! 5.Rxf7† Ke4! leaves Black’s king well placed to support the advance of the d-pawn. 5.Rc7? 5.Kg3! d4 6.Kf2 was necessary. 5...d4! 6.Rxc6 d3 7.Rd6 Rxd6 8.cxd6 d2 White resigned. 2.Kg3! White’s king must monitor the passed pawn so that the rook can remain active. 2.Rxa5?? d4 and the d-pawn is unstoppable. 2.Rd4? puts White’s rook in a passive spot, and 2...Kg6 collects the pawn on g5. 2...d4 3.Kf2 Rd5 Now the black rook threatens two pawns along the fifth rank. 3...d3!? 4.Ke1 Rd5 5.Rxa5 Rxg5 6.Kd2 Rxg2† 7.Kxd3 h4 8.Ra4 h3 9.Rh4 h2 transposes to the main line. 4.Rxa5 47 4...Rf5†! A clever finesse, forcing White’s king away from the g-pawns. 4...Rxg5 5.b4 d3 6.Ra3 Rd5 7.Ke1 and White is safe. 5.Ke2 5.Kg3? d3 6.Ra4 d2 7.Rd4 Rd5 wins for Black. 5...Rxg5 6.Kf3 6.Kd3 and 6.Ra4 are likely to transpose. 6...d3 7.Ke3 Rg3† 8.Kd2 48 8...h4 The most direct. 8...Rxg2† 9.Kxd3 h4 10.Ra4 is another transposition. Quite a tricky try is: 8...Kf6!? 9.Ra4 9.a4!? Ke5 10.Ra8 Kd4 11.Rd8† Kxc5 12.Rf8 should also draw. 9...Kg5 The alternative is 9...Ke5 10.Rh4 f5 11.Kc3! (11.Rxh5? Ke4 12.Rh4†? f4 13.Rh2 Re3–+) 11...f4 12.Rxh5† Ke4 13.Rh8 Ke3 14.a4 Ke2 15.Rd8! and White holds on. 10.Ra8 Kf5 11.b4 Ke4 49 12.Kc3! Improving the position of White’s king is a recurring theme in the defence. 12...Rxg2 13.Re8† Kf3 14.Kxd3 Rxa2 15.Rf8 Ra7 16.Rc8 Ra6 17.Rf8= 9.Ra4 Rxg2† 10.Kxd3 h3 11.Rh4 h2 12.a4 Ra2 13.a5 13.b4!? Rxa4 14.Rxh2 Rxb4 15.Re2 should also lead to a draw with careful play. 13...Rxa5 13...Kg6 14.a6 Kg5 15.Rh8 f5 16.a7 Rxa7 17.Rxh2 holds. 14.Rxh2 Rxc5 50  Black has emerged with an extra pawn. It shouldn’t be enough to win, but White still needs to defend carefully. 15.Kd4 15.b4 Re5 16.Rc2 should also work. 15...Rd5† 15...Rc1 16.Ke5 holds. 16.Ke4 Kg6 17.Rc2 Rd6 18.Rc5 Re6† 19.Kf3 f5 20.b4 Kg5 21.b5 White liquidates to a clearly drawn endgame. 51  Puzzle Commentary by Phil Siegbert Tarrasch’s famous edict comes to mind: “Rooks belong behind passed pawns: behind their own pawns in order to support their advance; and behind the enemy’s pawns, in order to impede their advance”. Stockfish endorses this advice here. Most strong club players, including myself, tried to work out the consequences of attacking the g-pawn instead, and overestimated our chances. The strongest players such as Adams and Granda made the right choice based mainly on precise calculation. Another interesting point here is that many of the weaker players, who hardly calculate any lines at all, and base their moves mainly on feeling and intuition, also ended up selecting the right move, just by following Tarrasch’s principle. Adams Insight In this complex position there are many enticing ways to proceed, and calculation regarding 52 supporting the advance of the d-pawn is important at every moment. The main line, which is quite forcing, results in the d-pawn being corralled by White’s king, but this gives Black the opportunity to create dangerous passed pawns on the kingside. McCumiskey gave 1...Re5 with a winning position for Black here (–3.0); in his defence, this was partly based on a tactical error that 2.Rxa5 is impossible. However, stronger players would realize from their own experience that even if Black picks off the weakling on g5, this is still a rather hopeful assessment. Imbalanced rook and pawn endgames with an extra pawn are, regrettably, rarely that simple! Puzzle 2 Black to play – Level 6 Show/Hide Solution Puzzle 2 – How Accurately Can You Evaluate? 00:05 Immediately I see a fork: 1...Qe3† I can also play 1...Re8 with the idea of getting my rook into White’s camp. 00:50 If his king moves with 2.Kf1 then I have 2...Bc4†. 02:30 In fact, coming to think of it, my first move would be 1...Bc4, to prevent 2.Kf1. Followed by 2...Qe3† and if the king moves to h1. I could play 3...Re8 and 4...Qe2. 03:05 Oh! But then white has 5.Re1 and I lose my rook. 53 04:20 If I were to take the f3-pawn with my bishop: 1...Bxf3 2.gxf3 Rxf3 White can then do whatever he feels like. And then: 3...Rxg3† 4.hxg3 Qxg3† 5.Kh1 Rf8 And now what does White do? I am playing 6...Rf2 with a good attack. 06:00 Other ideas are 1...Re8 followed by 2...Re3 and 3...Bb3 and 4...Qe7. White will obviously not let this happen, but that would be a great position for me. 08:00 White’s king is well protected at the moment. So to win I would probably need to fire some ammunition shots against it. 08:25 So I think I would probably play 1...Bxf3 and start bombarding the king’s fortress. Evaluation: Black is probably better. 00:55 I am looking at something like...h5-h4. 04:00 1...h5 2.Qe1 (to prevent...h4) 2...Bxf3 11:00 I am still looking at 1...h5 2.Qe1 Bxf3 3.gxf3 h4 4.Kg2 hxg3 but don’t think it is good for Black. 13:00 I am going to go for 1...h5 and if 2.Qe1 Qg6. Evaluation: –0.50 I am a pawn down. Other than that, I am happy with the position. 01:05 I am wondering if we can sacrifice something on f3, e.g. 1...Rxf3 2.gxf3 Qe3† followed by 3...Qxf3†. I am going to have a look at that, and see if it works: 1...Rxf3 2.gxf3 Qe3† Black wins. 04:30 So White cannot take the rook; he has to move something else. 06:00 1...Rxf3 2.Re1 (or 2.Rc3) 2...Re3 Keeping the tension. 07:00 White doesn’t have any counter-attacking chances. They are left suffering on the light squares. 1...Rxf3 2.Re1 Re3 followed by taking control of the e-file with...Rce8. Evaluation: –0.8 54 I am attracted to the idea of 1...Rxf3 just because it looks dangerous for White to take. 00:11 1...Rxf3 2.gxf3 Qe3† 01:05 White is not forced to capture the rook. He can play 2.Qd2 to try and swap queens. But in that event, Black has much the better bishop. (Thinks in silence.) 04:35 1...Rxf3 is the strongest move. And White should not take it. Evaluation: –0.4 Deeper Analysis by Michael Adams N.N. – Philip Hurtado Internet (blitz) 2019 Positions with opposite-coloured bishops can throw up unexpected attacking possibilities, and here White’s misplaced queen gives a passing chance. It’s important to understand the value of the untouchable bishop on d5, unopposed in the middle of the board. 1...Rxf3! Black has to seize the moment here, as White is only a move away from bringing the queen back 55 into the game from the queenside, thus defusing any attacking ideas. It’s important to give up the rook, rather than the bishop, as control of the light squares around White’s king is key to generating threats. 1...Bxf3? 2.gxf3 Qe3† 3.Bf2 Qxf3 And now 4.Rd2, 4.Rf1 or 4.Qd2 all see off the attack. 1...h5? has the reasonable idea of playing...h4, but that proves too slow as White can recover coordination with either 2.Qd2 or 2.Qe1. 1...Qe3†? is met by 2.Bf2. 2.Qd2! An important defensive idea, not weakening the eroded kingside further, while preventing...Qe3† and seeking relief in exchanges. The naive 2.gxf3? is impossible due to the attacking potential of opposite-coloured bishops. 2...Qe3† 3.Kf1 (3.Kg2 Qxf3† 4.Kh3 Be6† 5.Kh4 Qg4#) 3...Qxf3† 4.Ke1 Re8† 5.Be5 Qe3† 6.Kf1 56 6...Be6! Threatening...Bh3 mate. (The simpler 6...Rf8† 7.Qxf8† Kxf8 also wins.) 7.Kg2 Bh3† 8.Kh1 Qf3† 9.Kg1 Qg2# However, White doesn’t have to capture the rook and, by continually aiming to swap major pieces, can steer the game towards a draw. 2...Re3 A tricky practical try would be: 2...Qg4!? Attempting to keep more material on the board in an effort to pose more problems to the defence. 3.gxf3! This is the most accurate, but calculating the only drawing line over the board would not be easy. 3.Rf1 Rb3 offers Black slight but enduring pressure with no real risk. 3...Qxf3 4.Rf1 Qh1† 5.Kf2 Rf8† 6.Bf4 Qe4 6...Qxh2†? 7.Ke1 Qh4† 8.Rf2 leaves Black without a convincing follow-up. 57 7.Kg3 7.Kg1? g5 wins. 7...Rf6 7...Qg6† leads to a draw after: 8.Kf2 (or 8.Bg5 Qd6† 9.Bf4 Qg6†=) 8...Qe4= 8.h4 Rg6† 9.Bg5 hxg5 10.hxg5 Rxg5† 10...Qe7!? 11.Rf5 (11.Kg4!? is also possible) 11...Re6 12.Re5 Rxe5 13.dxe5 Qxe5† 14.Qf4 Qxb2 15.Re1 reaches a double-edged position which should be a draw, but in practice any result would be possible. 11.Qxg5 Qg2† 12.Kh4 Qh2† 13.Kg4 Qg2†= With perpetual check. 58 3.Rc3! Now it is hard for Black to avoid major liquidation. 3...Rce8 3...Rxc3 4.Qxg5 hxg5 5.bxc3 Ra8 6.Bd6 is close to a draw. 4.Rxe3 Qxe3† After 4...Rxe3 5.Kf2! Rb3 6.Qxg5 hxg5 7.Rd2 the doubled g-pawns reduce Black’s chances. 5.Qxe3 Rxe3 6.Kf2 Rb3 7.Rd2 b4 8.axb4 8.a4 Kf7 and Black can keep trying. 8...Rxb4 The position should be a draw but White still has some suffering ahead, due to the two isolated pawns on the queenside and the more passively placed rook. 59  Puzzle Commentary by Phil This is one of those moves that you either spot quickly or miss completely. Most players, myself included, once they find 1...Rxf3 go for it, and evaluate the position very favourably to Black. All strong players, and especially top grandmasters, spend more time on finding their opponent’s best reply and find 2.Qd2! leaving Black with just a minimal advantage. I noticed early on when conducting the experiments that stronger players would not play their move immediately, no matter how obviously good it looked. Michael, for example, even though he saw 1...Rxf3 after only eleven seconds, did not play the move until four-and-a-half minutes later. By then he had seen White’s best reply 2.Qd2, and assessed the position incredibly accurately – with the precision of a chess engine. Adams Insight I felt that those who missed 1...Rxf3 here were probably not sufficiently aware of the power of opposite-coloured bishops whilst attacking. If the rook on f3 is captured, White is helpless on the 60 light squares. It is also worth noting that White’s major pieces don’t influence the kingside at all in this puzzle position; they are all on the other side of the board, and the best defensive idea is to bring them back into the game as quickly as possible, aiming to minimize the damage with exchanges. Although the rook can’t be captured immediately, the queen retreat means that Black will have to allow exchanges as White consolidates. Black’s bishop is especially well located on d5, as the pawn on d4 prevents any white pieces from disturbing it. Puzzle 3 White to play – Level 7 Show/Hide Solution Puzzle 3 – Simple, Yet So Difficult Materially, Black and White are even. Oh, no they are not! White is a pawn up. 00:30 Black is threatening to take the bishop on e2. And also to take the pawn on a2. 01:00 The white king is on the back rank. There could be back-rank issues. White has to protect his pieces. 01:45 Can we play an active move? Can White play 1.Rd7? No, I can’t. I can’t move the rook because I lose the bishop. And I can’t move the bishop because I lose the rook. 02:00 I think I have to move the rook. 61 03:30 White might have an edge, but has to be careful. I like the idea of 1.Rd2 at the moment. Protecting the bishop and attacking the queen. 04:00 White cannot develop his pieces. Whereas Black can. 04:45 1.Rd2 would be my move. Unless I am missing something. If he gives me check, I can play Bf1, or Rd1. 06:30 And if he moves 1...Qe4, I can play 2.Rd7. 08:00 1.Rd2 is my move. White is better because Black’s pawns are weak and isolated. Evaluation: +1.0 My first thought is that White is a pawn up. 00:25 Black’s queen on c2 is really active though. It can take the pawn or the bishop. Let’s see. Black’s king is well protected, so there is no immediate attack on the king. 01:30 1.Rd2 is the first move to consider. 1...Qc1† 2.Bd1 and I maintain my material advantage. What I said is not true. 2.Bd1?? Loses to 2...Bxf3. 02:00 So after 1.Rd2 Qc1† I have to play 2.Rd1 Qc2, with a repetition. So if I don’t want a repetition, I have to think of something else. Maybe 1.Kf1, or 1.Re1. 05:00 To 1.Kf1 there is 1...Ba6!. 1.Re1 Qxa2 looks quite equal. 1.Re1 Qxa2 Maybe I can now think of an attacking move, like 2.Ng5 or 2.Bc4. 07:30 2.Bc4 Bxf3 3.gxf3 (to avoid more exchanges with 3.Qxf3 Ne5). I believe White preserves a small advantage after 3.gxf3. The good thing is that the bishop on c4 is well placed, and I can follow-up with h4-h5 and putting pressure on f7. 09:00 Black’s pawn on c5 is weak also. 11:00 1.Re1 looks passive, but it is the only good move. The other move 1.Nd2, I wouldn’t even consider it. 1.Re1 Evaluation: +0.75 White has an extra pawn. White appears to be better. The first move I can think of is 1.Re1. 01:00 1.Kf1 Ba6 can be annoying. 62 01:45 After 1.Re1 I like White. They have a better pawn structure. And White can play Bc4, h4 and h5. 03:30 Another interesting variation could be: 1.Nd2 Rd8 2.Nf1 Rxd1 3.Bxd1 Qxd1 4.Qb8† 04:00 Also 1.Nd2 Qxa2 2.Qc7 Ba6 3.Qxa7 No. I am not totally happy with 1.Nd2, it allows too many exchanges. 12:00 I’ll probably stick to 1.Re1. (Thinks in silence for a while.) 13:00 1.Nd2 Qxa2 2.Qc7 Bd5 3.Qxc5 Bxb3 4.Nxb3 Qxe2 And now I can play 5.Rf1 or 5.Rc1 and White is better. I can manoeuvre my knight to d4 and f5 and keep the pressure. 1.Nd2 Evaluation: +0.9 00:30 The bishop is attacked. I have three options: 1.Rd2, 1.Re1 and 1.Kf1. This requires some calculation. 01:30 1.Kf1 is looking bad because of 1...Ba6!. 01:50 1.Rd2 Qc1† and White is forced to go back to 2.Rd1, with a repetition of the position. 02:30 1.Re1 Qxa2 and now 2.Bc4 or 2.h4 could be interesting. 03:45 1.Re1 Qxa2 2.Qc7 could be slightly better. 1.Re1 is probably the strongest move. 04:30 1.Re1 Qxa2 2.h4 Qxb3 3.h5 Ne7 Now I can’t play 4.h6 because of 4...Nf5. But the position seems promising for White. 05:00 I would go for 1.Re1. After 1...Qxa2 White can continue with either 2.Bc4 or 2.h4, with small pressure. Evaluation: +0.3 Deeper Analysis by Michael Adams Philip Hurtado – N.N. Internet (blitz) 2019 63  White has an extra pawn, but this will be regained once White meets the threat to the bishop. White’s advantage lies in the fact that the bishop can settle on an excellent post on c4, which will set up attacking possibilities such as h4-h5 or Ng5. The advance of the h-pawn has the additional positive of negating back-rank problems, blending attack and defence. 1.Re1! White calmly plays the move that most comfortably consolidates. The rook is securely defended by the knight here. 1.Rd7?? fails to 1...Qxe2, as the bishop can’t be captured due to the back-rank problem. 1.Rd2 Qc1† 2.Rd1! (2.Bf1? Bxf3–+) 2...Qc2 repeats the position, giving White a second chance to play 3.Re1!. 1.Kf1 It appears logical to keep the rook on the d-line, but White’s pieces aren’t soundly placed enough for this move, and importantly Black can swap White’s bishop with a tactical idea. 1...Ba6! 2.Re1! The alternatives 2.Bxa6? Qxd1† and 2.Rd2? Qc1† 3.Ne1 (3.Rd1 Qxd1†–+) 3...Qxd2 win for Black. 2...Qxa2 Black’s problems have mostly been solved. 1.Nd2?! This move was chosen by Julio based on the line 1...Rd8? 2.Nf1 Rxd1 3.Bxd1 Qxd1 4.Qb8† Nf8 5.Qxb7 and White wins – but he missed an important resource. 64 1...Ba6! 2.Bh5! 2.Bf3? Rd8 wins for Black. 2...Rd8 3.e4 3.Qc7? runs into the sublime 3...Rd5! when Black wins easily, as the knight can drop back to f8 to defend along the back rank. (3...Rxd2 4.Rxd2 Qxd2 5.Qb8† Nf8 6.Bxf7† Kxf7 7.Qxa7† allows White to fight on.) 3...h6! 3...Rxd2? 4.Qb8† Nf8? 5.Bxf7†! Kxf7 6.Qf4† with good winning chances for White. 4.Qe3 Qxa2 Black regains the pawn with dominant pieces. 1...Qxa2 65 2.Bc4 The bishop is powerful on this outpost, securing the b3-pawn and eyeing f7. Having consolidated his own position, White has many threatening possibilities such as h2-h4, Ng5, Qd6 or Qc7. Black will have to take swift action to take the sting out of a potential kingside attack. This will involve making concessions to negate White’s attacking ideas, which is likely to mean drifting towards an unfavourable endgame. I was also tempted by targeting the black knight immediately: 2.h4!? Qxb3 The idea is that 2...h5 3.Bc4 now gains in strength by threatening to remove Black’s knight. 3.h5 Ne7 3...Bxf3 is well met by 4.gxf3! (4.Bxf3 Qc3! 5.Rd1 Ne5 solves Black’s problems) 4...Qc3 5.Rd1 Ne5 6.Rd5 and Black’s knight lacks a stable outpost. If 6...f6? 7.f4 Nc6? 8.h6 White is already winning, as 8...g6 9.f5 g5 10.Qc7 decides. 66 4.Qg5 Qb4 4...Qe6 5.Qxc5 is also problematic for Black. 5.Rd1 f6 6.Qg3 Nf5 7.Qc7 Qb6 8.Rd7 Bc6 9.Qxa7 Qxa7 10.Rxa7 Nd6 Black should be able to hold on, but White is still pressing. 2...Ba6 Black looks for relief in exchanges, and sensibly proposes a trade of the powerful bishop on c4. 2...h6?? allows 3.Qxg6. 2...Qa5 3.Rc1 leaves Black worse, due to the split queenside pawns and less active pieces. 3...Qa3 4.Rd1 The one-move threats have enabled White to transfer the rook back to the d-file, while Black has achieved nothing shuttling the queen back and forth. 2...Bxf3 makes sense in terms of reducing material, but now it becomes clear how strong the bishop on c4 is. 3.Qxf3!? (3.gxf3 is also quite potent: 3...a5 4.f4 a4 5.f5 and White’s play is arriving first.) 3...Ne5 4.Qe2! A small trick to maintain the bishop. 67 4...Qxe2 5.Bxe2 Rb8 6.f4 Nc6 7.Bc4 White has a pleasant endgame advantage. Notice that 7...Na5 doesn’t force an exchange in view of 8.Bd5 Nxb3? 9.Rb1 and White wins. 3.Bd5 Ne7 4.Be4 Black has dislodged the bishop from its strong diagonal, but at the heavy cost of destabilizing both of Black’s own minor pieces. 4...Qxb3 4...f5?! 5.Bb1! Qxb3? 6.Qd6 with decisive threats. White has several good continuations here. 68 5.Qd6 5.Qe5!? Ng6 6.Qxc5 should enable White to win the a7-pawn sooner or later. For instance, 6...Qb6 7.Qa3 Bb5 8.Rb1 a6 9.Nd4 Rb8 10.Bd3± and White will reach a 4–3 scenario on the kingside. 5...Qe6 6.Bxh7†! A convenient tactical idea. 6...Kh8 6...Kxh7? 7.Ng5† wins. 7.Qxe6 fxe6 8.Bc2 Black will struggle to defend the four pawn islands in this endgame. 69 * 1.Rd2 is assessed by the computer as being on a par with 1.Re1, but only because of the repetition 1.Rd2 Qc1† 2.Rd1 Qc2, when 3.Re1 gives White the advantage. For the purpose of the exercise, 1.Re1 is the key idea needed to fight for the advantage. Puzzle Commentary by Phil I found myself in this situation as White in a blitz game. I remember considering 1.Re1 for a brief moment, but quickly rejected it as it felt too passive, and settled for 1.Rd2, followed by a draw by repetition. When I reviewed the game with Stockfish, I was surprised at the big advantage it thought White had. I wondered what fabulous move it had seen. I was amazed at the simple yet unnatural solution 1.Re1!. The move is particularly difficult to play, for two reasons. First, it involves moving your rook away from the only open file, which contravenes one of our most instilled principles; and secondly, it throws the a2-pawn to the wolves. I showed this position to many club players, and although the best move occasionally fell into their radar, nobody below 2300 dared play it. Adams Insight 70 Sometimes you have to take a backward step before you can go forward. Although 1.Re1 is undesirable for a few reasons, it secures White’s pieces and leaves White free to undertake more positive action on the following moves. The bishop can head to a good post on c4, and attacking ideas with h4-h5 are available. There are enough positives to White’s position to enable White to exert considerable pressure, even once the rook has temporarily