Chess960 – pros and cons

The third world chess champion, José Raúl Capablanca, believed that chess would be played out in a few decades, with draw being the inevitable result at top level. He came up with his own chess variant, played on a bigger chess board and incorporating 2 new pieces, trying to turn the chess fight into something more complex.

The eleventh world champion, Robert James Fischer, expressed similar concerns, envisioning the “draw death” of chess caused by the incredible development of chess theory. His solution? Another chess variant, where the position of the pieces on the first rank would be randomized. His style of chess was initially known as “Fischer random”. Currently most people refer to it as “chess960”, since there are 960 different starting positions. Other names for it are “freestyle chess” or “chess 9LX”. The idea is simple – the rules are essentially the same, but the entire “home cooking” part in the opening is rendered useless and the game would be decided by pure playing strength.

The 5-time world champion and strongest chess player of all times, Magnus Carlsen, decided not to defend his title after his 2021 victory against Nepomniachtchi, and started to focus more on fast time control games and chess960. The main reason was the same – the fact that modern opening preparation has become so powerful and deep that a well prepared opponent can basically “kill the game” and reach very dry equal positions by force – a strategy used by many of his opponents. The stronger player has only 2 options – play a more complex but objectively inferior opening line, or play principled best-move chess just to test the opponent’s preparation and memory.

Top engines, much stronger than any human player, paint the same picture. Any mainline opening, if played correctly, leads to complete equality. Capablanca’s intuition was right.

It comes as no surprise that Magnus and other top players are trying to make chess960 more popular. With the right sponsors, they might succeed. Some people love the idea, others hate it, and most are somewhere in-between.

PROS

  • No opening preparation. For most players, it’s already a daunting task to memorize their current opening repertoire, and that would be just one starting position out of 960.
  • Since everything about the starting position feels new, the fight starts very early in the game. There are important strategic decisions to be made as early as move 1.
  • Early tactics are common. It’s not unusual to see even top grandmasters losing a pawn or even a piece in the opening stage of the game, because they cannot rely on the tactical patterns they’re used to.
  • There is a lot of room for creativity. Playing a gambit in a chess960 game is a matter of intuition and feel for the game. Playing a gambit in something like the poisoned pawn variation of the Najdorf is a matter of spending weeks trying to understand engine novelties at move 30. As they say, nowadays we all have the same coach and read the same opening book, authored by Stockfish.
  • Even if the starting position is sometimes “weird”, as the game progresses it looks more and more like regular chess. The concepts we are familiar with regarding king safety, piece activity, space, pawn structures and so on are still valid, the importance of being able to calculate correctly is still the same. The strongest chess players in classical chess are still on top in chess960.
  • The percentage of draws is much lower. There is no safe way to reach an equal middlegame or endgame. There is no well known move repetition in the opening. Mirroring white’s moves usually does not work, on the contrary.
  • Very often the games are wild and double edged. Without the safety blanket provided by the opening preparation, both players feel uncertain about what’s going on in the initial stage of the game, and that simply makes it more fun to play and watch. The fact that players lose orientation early can be seen as a disadvantage as well.

CONS

  • Some starting positions simply look very “ugly”, there is no piece harmony, both sides are struggling to get developed.
  • Certain starting positions are objectively better for white. However, the advantage is probably not more than +0.6, at least according to Sesse.
  • Since white’s edge can be bigger in some positions compared to others, it would be fair to play pairs of games (one with each color, same starting position). That leads to other problems, since in the second game the player with the black pieces already has an idea of what to do (or what not to do) – even more so if there was enough time to check with an engine.
  • Most chess players have spent a huge amount of time studying current opening theory. Was all of that in vain, wasted time? They would feel betrayed. Part of the rich legacy of the game of chess would be lost.
  • Given the complexity of a fresh opening position, fast time controls lead to way too much randomness and silly blunders. To get something that resembles decent chess, long time controls are needed.
  • There is a lack of continuity, it’s hard to learn from opening mistakes, because it’s unlikely to ever reach the same opening position in the future.
  • The ability to spend countless hours trying to find new ideas in the opening can be seen as a chess skill. That would no longer be the case in chess960.

I confess that I enjoy watching chess960 played by strong GMs. I find even the ugly positions or early mistakes more entertaining than another boring draw in the Berlin defense. But I don’t think it will replace classical chess any time soon. Any other opinions are welcome, feel free to comment.

Does talent matter?

Some of the questions I get asked a lot have to do with the never-ending nature vs. nurture debate.

  • Do we have equal potential to become IMs, GMs, or even world champions?
  • Is there something that we could call talent when it comes to chess abilities, or it’s all about training the right way?
  • If my son/daughter gets the best chess tuition and works really hard, will he/she become one of the best in the world?

I wrote an article on this topic a few years ago (in Romanian), and I am posting most of it below. My position hasn’t changed since then.

Opinions on the relative importance of native versus environmental factors (where environment means everything else, including training, coaching, etc.) vary widely. Traditionally, very strong chess players have been seen as the ideal example of special native endowment, as have great composers or mathematicians. This picture makes sense intuitively. It is very easy for anyone to learn the rules of the game and the basic ideas, but it seems extraordinarily difficult to achieve the performance of winning dozens of simultaneous games or crushing opponents without even looking at the board. Besides, what else but genius could explain the astonishing performances of child prodigies like Morphy, Capablanca or Reshevsky?!

Reshevsky crushing opponents and the age of 8

The opinion was shared by most past champions. You might enjoy hearing the voice of the great Alekhine, who seemed to be very convinced that great chess players are born, not made (around second 35).

The widespread egalitarian fallacies that have become prevalent in the last decades did not forget or forgive chess. An important contribution was made by some studies published in the 90s, but even more so by the success of the Polgár experiment, which proved (in the view of many) that directed, sustained work, from the youngest ages, leads to exceptional results. More on this later.

From the point of view of serious research, a very important moment was represented by a study published in 1993 by the psychologist K. Anders Ericsson and his team. It introduced the concept of deliberate practiceDeliberate practice (DP) is defined as sustained work aimed at improving performance in a specific domain. The authors make a clear distinction between DP and two other forms of activity: work , assumed to be an effort motivated primarily by external rewards, and play , respectively, seen as inherently pleasurable activities lacking a long-term goal. In the original paper, the authors were able to find a correlation between the level achieved by a group of violinists and the number of hours of DP. The best violinists had somewhere over 10,000 hours of DP, compared to mid-level violinists who were around the 7500 level and the not so good ones at 5000 or less. This is the origin of the infamous 10,000 hours myth – supposedly enough to achieve excellence in a field. In later work, the authors extended their research to other groups (pianists, chess players!), maintaining the position that the only difference between those who excel in that field and the rest is the number of hours of DP. However, even these authors agreed that it is entirely possible that the willingness of individuals to engage in DP for very long periods of time is a genetically determined characteristic. Thus, in their view, there is no talent or genius as such in any field, but there might exist a predisposition towards sustained work. This is the only concession they made.

Ericsson’s ideas quickly caught the public’s imagination and were followed by a wave of popular science articles in which various authors sought to show that the only thing that differentiates individuals is the amount of work put in to achieve a goal, and as a result any parent can make their child a genius in any field they want. The practical results, however, are still awaited. However, negative reactions followed, highlighting serious deficiencies in the methodology used by Ericsson and in the interpretation of the results. One of the biggest problems is that he never took into account the very numerous cases of individuals who gave up on their original goal after long periods of DP precisely because they couldn’t get anywhere close to the desired results. In other words,  DP seems indeed to be a necessary condition for success, but not a sufficient one! Other obvious problems stem from the difficulty of accurately estimating the number of hours of DP, which is based on some rough approximations made by the subjects. Also, it is impossible to determine the quality of DP (one person could do more in one hour than another person in 3). Finally, it is highly unlikely that the same threshold (10,000 hours) is universally valid, regardless of activity or goal.

Strictly related to the field of chess, there are at least 6 studies that sought to find a possible correlation between the number of hours of DP and the level reached by the player (expressed by the ELO coefficient). They are listed below (the excerpt is taken from a paper published in 2013 by Hambrick et al.):

The above authors are generally psychologists and chess enthusiasts at the same time. Guillermo Campitelli, for example, is an Argentinian with a maximum rating of around 2200, which intrigued him precisely because he had been striving for many years, including during his junior years, to reach an international title. Despite the large number of hours of DP he failed to do so. Once he became an internationally known researcher, he aimed among other things to find explanations related to his failure. After thoroughly questioning a large number of chess players he discovered “strange” things that contradicted the official DP dogma. His findings confirm that training is indeed important, but at the same time suggest that the speed of progress varies a lot, plus there is an individual ceiling. Let me quote him:

The problem is that among the players who reached master level, some did it in 3,000 hours… while others took 30,000 hours and reached the same level. And there are people who have trained well over 30,000 hours and have not reached this level.

I do not wish to go into technical detail details regarding the statistical methodology used to analyze the results of the above studies. A meta-analysis is graphically represented below:

What the figure shows is particularly important. It appears that, at least among the chess players analyzed to date (nearly 1100 players!),  DP contributes only 34% to the level achieved by the player. The remaining 66% cannot be explained by this element, being very likely the contribution of what we could call talent.

I am attaching another interesting graph from Gobet & Campitelli’s 2007 study.

The chart has intermediate level players (below 2000) as white bars, expert level players (between 2000 and 2200) as gray bars and master level players (over 2200) as black bars. It’s easy to see that the black bars (stronger players) tend to cluster to the right, corresponding to a greater number of hours of study, while the white bars (modest players) seem to be more frequently on the left (fewer hours of study). That makes perfect sense, but the anomalies prove that the number of study hours (DP) does not explain everything. Hambrick’s remarks:

  • There is a major overlap in the distributions, for example about a third of masters had fewer DP hours than the average of the expert group, and a few masters even had less DP than the average of the intermediate player group (players 2 classes below!).
  • Among intermediate players (below 2000), a quarter had more AD than the expert average and a few had more practice than the master average.
  • The number of hours of AD, in the case of masters, varies greatly, between 832 and 24,284 hours – a difference of almost 3 orders of magnitude (!!).

detailed analysis published in 2011 by Robert Howard regarding the Polgár phenomenon leads to interesting conclusions as well. The 3 sisters are known to have undergone an intensive training program from an early age with their father and later with various high caliber trainers. Even if we ignore Sofia, some interesting differences emerge between the two more famous sisters. Judit reached a peak of 2735 after about 60,000 hours of DP, while Zsuzsa had a much lower peak of 2577 after much more training, about 80,000 hours! Equally interesting is the fact that the average number of hours required to reach the maximum rating, for the top 10 players in the world at the time, was only 14,000. So, again, it can be said that a targeted, intense, and long-term training is necessary for high performance, but there are very large differences between individuals related to both the speed with which they progress and the maximum level they can reach. In addition, we should consider the following element: the sisters were already performing amazing feats at very young ages (for example Judit beat a master without seeing the board at 7), when they were not even close to the magical threshold of 10,000 hours of training. It is logical to assume that in the Polgár family there is a set of genes allowing the owners to learn chess with great ease.

Another key factor that several studies have highlighted is the particular importance of the age at which DP begins. It seems that to achieve excellence in chess, it must be as small as possible. A child manages to progress much faster than an adult, and furthermore, it seems that those who start after childhood do not make it to the top, regardless of the number of hours of DP.

There are obvious differences between kids trying to learn chess, clearly visible as soon as they see a chess board for the first time. Some of them grasp everything quickly, others need additional explanations, multiple repetitions and forget more. I have personally witnessed kids who couldn’t master the technique of checkmating with king and rook vs. lone king in 2 hours, while other kids get it in 5 minutes. Long before we can talk about thousands of hours of training, we can see how absurd the blank slate theory is. The same has been reported in all fields, whether it is piano, football or mathematics. The signs of genius are obvious long before the number of hours worked or the quality of any coach can be suspected of having an influence. At the age of 2-3, Magnus Carlsen could solve problems and play games designed for much older children. At 5, he would read geography books and could name all countries and their capitals. His father tried to teach him chess at that age, but he was not interested initially. At the age of 7 he took up chess again, on his own, and within a year he was stronger than his older sisters, who had been playing chess for years. Does anyone think that Magnus is where he is just because he studied a lot?! How many hours of study had he managed to accumulate by the age of 13, when he was already giving Kasparov a very hard time? Was he a product of the powerful Norwegian school of chess? If Agdestein took any other child and trained him, would that kid reach 2870? We all know the answers.

Some have fooled themselves by doing retrospective studies, just like Ericsson. A retrospective study is one that looks at past events. For example, we can compare champions and low-level players, and find that, in general, champions trained more – hence the conclusion that training explains their level. WRONG. This conclusion has been proven false in all cases where prospective studies were used instead. Such studies start with a group of individuals at the same level and track their progress into the future for years. What we find is that the number of hours is actually a poor predictor for the level a student reaches (students who studied exactly the same amount of time ended on vastly different levels, and others even dropped out along the way, precisely because they realized that they were not making progress).

Why are there so many top chess coaches who downplay the importance of talent? How come they don’t notice what any kindergarten teacher, or any parent with two or more children notices? There are several reasons:

  • Some coaches are acutely aware of the reality that certain players are untalented and say so in private, but not in public or when discussing with their students or their families. The reason is obvious, they don’t want to lose their customers. It is a source of income. Car mechanics will find something wrong with the car, doctors know another checkup is needed, lawyers want you to appeal every time.
  • Very good coaches manage to maximize the genetic potential existing in each individual. They manage to get more than the student could have achieved on his own. They notice that, in general, students who work harder get higher, and automatically assume that it’s just the number of hours that makes the difference. In reality, even the disposition to work long hours is likely genetically determined.
  • Very good coaches get to train the best students, precisely those who are the most talented of their generation. There is a serious pre-selection. Kids who don’t like chess don’t even make it to a club or a coach. Of those who like it, a large proportion have no talent and are quickly lost. Of the few who do get to play serious tournaments, an even smaller proportion goes on to win titles and medals, becoming leaders of their generation. And of those, only the most promising end up working with an internationally renowned coach. Being used to such children with an obvious gift for chess, the coach may wrongly extrapolate what he sees to the general population.
  • The egalitarian model is that at some point, by chance, student X begins to prepare more than student Y, and as a result in a few years there will be a clear difference in playing strength between them. The model closer to reality, however, is different: X and Y come into contact with chess, X understands and learns chess a lot easier; as a result X works harder than Y because he enjoys it, and because some positive results show up (rating, medals), which in turn stimulate him to prepare even more. A positive feedback is in place.

I understand that the fact that talent plays a role can feel disheartening to many. They are probably asking themselves “What if I am not one of the lucky ones? Am I just wasting my time?”. The good news is that, for most people, the “ceiling” is way higher than they think, and definitely higher than their current level. As I keep saying, chess has to be fun. Yes, maybe you don’t have 2700+ genes. But you can gain another 200 points, and I hope the path will be enjoyable.