The Secret of Chess

by David Smerdon
12/22/2017 – Other chess reviewers have been at best dismissal and at worst harshly critical of The Secret of Chess, by Lyudmil Tsvetkov. However, according to GM David Smerdon, this book is a one of a kind work that legitimately has the potential to revolutionise how we think about chess. | Photo: Smerdon at the Tromsø Chess Olympiad, by Andreas Kontokanis CC BY-SA 2.0 via Wikimedia Commons

Fritz 16 - He just wants to play! Fritz 16 - He just wants to play!

Fritz 16 is looking forward to playing with you, and you're certain to have a great deal of fun with him too. Tense games and even well-fought victories await you with "Easy play" and "Assisted analysis" modes.

More...

Nimzowitsch, Kmoch...Tsvetkov?

The AlphaZero phenomenon has reinvigorated interest in chess engines and the way they "think". This story originally appeared on GM David Smerdon's blog, and is re-published with kind permission.


A couple of months ago, I received a curious email:

Lyudmil-TsvetkovHello, Mr. Smerdon,

I recently published an innovative book on chess knowledge/evaluation, ‘The Secret of Chess’. It is written very much in the vein of Nimzovich’s and Kmoch’s works, but is of much larger scope and accuracy of assessment…

Please, try to take a look at it, I guess you will not be disappointed.

Best regards,
Lyudmil Tsvetkov, author of the book

As you might imagine, I was pretty sceptical that this book would indeed revolutionise chess — especially after discovering Tsvetkov’s a 2100 player who has been inactive from tournaments for more than a decade. Still, the author claims to have spent the past five years almost entirely devoted to the study of chess, and especially the use of engines. Given the 'big data' revolution in many sciences today, a part of me has always believed that there must be more sophisticated ways to train in chess. And so, despite the somewhat outrageous claims and poor written English of the introduction, I pushed on, hoping to find at least one or two useful training nuggets. And so, I read the whole book.

Before I get into the details, I want to make two paradoxical statements. First, I completely understand why other chess reviewers have been at best dismissal and at worst harshly critical of The Secret of Chess. Second, however, this book is a one of a kind work that legitimately has the potential to revolutionise how we think about chess.

In fact, it’s not even written as a regular book, but more of a mathematics textbook and an encyclopaedia (ironically, Amazon classifies it under ‘Humour and Entertainment’). It’s essentially just a list of hundreds of essential chess themes or patterns that together comprise the key heuristics to success, each with a definition, frequency and value (written as a bonus or penalty in terms of centipawns, or hundredths of a pawn). Here’s an example:

Blocked pawns on squares the colour of the bishop

Definition: pawns, blocked by enemy pawns on squares the colour of the bishop

Value: additional penalty, -10cps [centipawns], both for the mg [middlegame] and eg [endgame]

Additional information: the over-penalty is due because:

— the condition of being blocked makes the pawns fixed targets, unable to move; fixed targets are easier to attack and destroy

— blocked in general represents a more durable condition, further highlighting the weakness

Frequency: very frequent

The themes are typically accompanied by one or more illustrative diagrams, and occasionally also so-called 'piece-square tables', listing the centipawn values for pieces on each of the 64 squares of the board.

Such tables, like the values, are presumably designed for chess engine programmers, as their goal is primarily in arriving at the most accurate evaluation of a position as possible. And indeed several members of the computer chess community have apparently spoken highly of Tsvetkov’s identifying and quantifying these features. And you can actually find quite a few interesting threads on the Talk Chess forum in the past in which the author has contributed to improving the world’s strongest engines with his suggestions. I do get the impression from reading the book that chess programmers comprise Tsvetkov’s primary audience.

From a regular reader’s perspective, the tables and precise values aren’t very helpful. We’ll never be able to remember all of these numbers, let alone implement them in a live game. What would have been useful is to have these values distilled down into the more human “small advantage”, “clear advantage” and so forth, along with other simplifying heuristics. But after a chapter or two, one does get used to the book’s structure, and the going gets easier. What I did was make a list of the key themes that I wasn’t aware of, together with a more human assessment of their relative value and some clues for how I can remember them. It took some time and effort that I wish had been done for me in the book, but at the end of the day, I consider this written summary to be immensely rewarding and a reflection of the true value of this book.

Many of the themes are well-known to seasoned players, such as outposts, open files, the bishop pair, etc. But there are plenty of others that aren’t so obvious, at least to me, and which I found really interesting. Some of these concepts are so (post)modern that the author has had to come up with his own original terms for them, such as “spearhead pawn”, “unbackwarded pawn” and “double-root pawn”. Indeed, a huge bulk of the themes have to do with various pawn features, which leads to some of Tsvetkov’s most controversial arguments: That 1.c4 is White’s strongest first move, that the French is close to a losing opening for Black, and that the Stonewall Dutch, for either colour, is to be recommended. (Given my own experience, I’m not sure how I feel about this…)

It follows from his concept of “twice-backwarded pawns” that after 1.c4 e5 2.Nc3 Nf6, White should play 3.e4!, with an advantage. To be honest, I feel that Tsvetkov could be a little biased here by his large experience playing against engines; we know that engines struggle to play blocked positions, so a human would have a better chance of success by trying to close things up as soon as possible. Is 3.e4 objectively as strong against another human? (And, conversely, is 3.e3 as bad a move as Tsvetkov makes out? For example, he gives the position:

 

He says:

In spite of the big development advantage white enjoys, black has already a winning position. And the main reason for this is the familiar central pawn pattern in the form of the e4,e3,f2,g2 pawns. The central e4 black backwardmaker is having a cramping effect upon the entire white king side and renders the white shelter inflexible.

He goes on to explain this rather outrageous claim, which I’ll reprint in full:

Top engines, of course, completely misunderstand and misplay the variation. Most of them will still assess above position as very favourable for white, but of course, black is winning. As the associated lines are very deep, engines basically see nothing and rely on their positional evaluation, which is far from perfect and, in many cases, like this one, rudimentary.

Winning is pretty much straightforward: black continues with slow attack on the king side, pushing pawns there, and gradually transferring pieces to this focal point, including the ones on the queen side. As the position largely carries a closed character, emphasised by the e4 central backwardmaker, black has all the time in the world for regrouping and coordination. White, on the other hand, can do almost nothing, as the white shelter is inflexible, and attempting to break free with f2-f3 or f2-f4 will easily backfire, creating multiple weaknesses.

Of course, in a practical chess game, matters are far from clear. But objectively…? Honestly, I don’t know. I’d need to hear the opinions of correspondence and freestyle chess experts before I’m convinced. But still, I like hearing the way Tsvetkov thinks about these positions, mainly because it’s a new way of evaluation that I haven’t encountered before. And I’m all for diversity.

Here are some other specific examples that I’m sure will have many grandmasters scratching their heads:

 

One of the key insights from The Secret of Chess is the need for flexibility in one’s pawn structure. The diagram is an example of this, for which the author writes:

In spite of the enormous lead in development and massive centre, white is actually a bit worse… Fact is, white’s pawns are quite broken down in different very small groups, while black has a great number of interconnecting pawns. Look at the h6-g6-f7-e7-d6-c6-b5-a5 weaving snake!

It seems hard to believe that Black could be anything but worse here — White has the centre, more space, better development… But on the other hand, it’s true that Black’s position is tough to crack, and after further thinking, I couldn’t find a convincing plan for White to gain a clear advantage. I still have my doubts that Black is better, but at least Tsvetkov’s principle of pawn flexibility seems plausible to me.

Tsvetkov’s definitely not afraid of making bold statements about how a game will finish from a very early stage. He often writes that “Of course Black is lost here…” at a point where material is equal and there are no clear tactics or attacks at play. Here are two nice examples:

 

I don’t know how I would have evaluated this position if I got it in a game — probably I would have thought White was doing quite okay. I certainly wouldn’t have guessed that White is lost! But Tsvetkov writes:

Black has already won the game, because of the white twice backward shelter f2 pawn…Black should get its dark square bishop to f6, sacrifice it on h4 for 2 enemy pawns, later transfer a knight via h7 to g5 and f3, with smashing attack. If necessary, both rooks could be enrolled in the approaching army to support the assault, by taking long-range aim along the h and g files.

Compare the previous position to this one (I have flipped the colours for ease of comparison):

 

White will win this at some point, although this is not currently easily recognisable.

The reason is precisely the complete lack of a pawn shelter for the black king (with the f5,g4 and h5 pawns all advanced), while its white counterpart still enjoys the immediate contact of 2 own shelter pawns, g3 and f2. It might take 30, 40 and even 50 moves, but at some point, the position will be opened, the white pieces will penetrate and mate the black king. The black king does not have any pawn shelter on its current place, the king side, as well as in the center, while walking to the queen side is simply impossible because of impending attacks.

It’s so hard for me with my classical chess training to believe that this position is already decisively won for White, while the previous one is lost. I’m still not sure that I believe it. And yet, as with most of the contentious claims in the book, these evaluations are backed up by Tsvetkov’s surprisingly robust theory of chess evaluation. Coupled with a second book of 100 annotated victories by the author over chess engines, we can get a sense for where the book’s strengths lie: Closed positions. Indeed, I get the impression that Tsvetkov’s approach works quite well in extremely blocked structures, and I suppose one could tailor an opening repertoire to maximise the likelihood of achieving said positions (Tsvetkov’s opening choices against Stockfish and Komodo would be a good place to start). Given his success both in improving and beating chess engines, I feel like it would be unwise to ignore his lessons on how to evaluate closed structures, and it would not surprise me if computers tell us exactly the same thing in ten or fifteen years’ time.

And this goes to the heart of the issue. Whether or not you believe in Tsvetkov’s chess philosophy or even just his evaluations, there’s no question that his approach to chess is fresh and different, something that’s been missing in the chess literature for a long time. I probably won’t end up a convert, but I have definitely spotted several interesting new heuristics that I will be trying out in the future. Concepts such as "vertically isolated pawns", "twice backward pawns" and "spearhead pawns" are not things that I consciously think about when I analyse, though in a sense they sit somewhere in my chess intuition for assessment. But knowing which of these heuristics are relatively more important than others — a feature that Tsvetkov’s quantifications can address — as well as automatically bringing them into one’s decision processes, might be quite valuable. In any case, I’m going to try it out.


Komodo Chess 11

The multiple computer chess world champion comes in a new and yet more powerful version. Thanks to co-author US Grandmaster Larry Kaufman, Komodo is the strategist among the top chess programs!


Unfortunately, the combination of the textbook style without much in the way of descriptions, combined with difficult English, makes The Secret of Chess a really hard read. And given the knee-jerk rejection that many 'classical' chess players will feel, there’s a fair chance this book will be largely ignored by the wider chess community. But I sincerely hope this doesn’t happen. I’m almost surely in the minority, but I believe Tsvetkov’s insights could really change the way we think about chess, from how beginners learn the game to how experts improve. It’s bold, completely different and sometimes conflicts with a lot of established chess wisdom, but, just like big data analysis, meditation and veganism, that doesn’t mean there aren’t some lessons to be learned, even if you don’t subscribe to the whole package. If you’re willing to open your mind to a new way of thinking about chess, and you’re determined enough to power through the text, this book is definitely worth a read.

Tsvetkov’s The Secret of Chess is available from Amazon in both Kindle ($5.96) and paperback ($25.99) editions.


Links


David is an Australian chess grandmaster and economist. He is the second highest ranked chess player of Australia. Smerdon has played for the Australian team in the Chess Olympiad since 2004. He goes by the online persona "Smurfo".

Discuss

Rules for reader comments

 
 

Not registered yet? Register