On Opening books in 2015

[Uri Blass]

I disagree with lyudmil and I think that at least in 99% of the cases engines play the right move if you give them a long time to calculate.

if there are some cases when they do mistakes you can find it by the evaluation and if white play the right move based on long analysis in every move but black get the advantage after move 15 or move 20(at least 0.1 pawns for black) then it is better to do analysis to see where white went wrong in order to fix the book.

Starting point if we build a book for white is simply play the moves the engines suggest after a long analysis for white and moves that are in the database for black.

Same idea if we build an opening book for black but in the last case I suggest to try to fix only cases when the evaluation for white is at least 0.5 pawn for white because small advantage for white does not mean that black did a mistake.

Engines might play the right move in 99% of cases in simple endgame positions, or simple middlegame positions, but not in the opening, as such positions are usually very complex, both for humans and engines to understand thoroughly.

Give current top engines not 1,2 or 5 hours thinking time, but 10 days thinking time on standard opening positions, and they still will not find the right move in at least half of the cases, I think in 70-80% of cases, based on what I have seen from engine play in openings.

It simply can not be another way: it took humans decades, and even centuries of analysis to ascertain with a fairly high degree of certainly what is the best move in specific opening positions, so how would you want engines to solve the puzzle in an extremely limited time in comparison?

Openings are very difficult to play, very complex, not so specific middlegame and endgame positions with reduced material and play opportunities.

I think that in many cases there is more than one best move in the opening.

If engines do not play the move that you like it does not mean that they do a mistake.

If you claim that engines do a mistake then you need to demonstrate it by playing against the engine and getting the advantage.

You do not need to play until mate and if the engine admit based on evaluation that you are better with black(at least 0.1 pawn for black) or you are significantly better with white(at least 0.5 pawn for white) then it is enough for me.

Uri

[Lyudmil Tsvetkov]

Another funny thing is: you test your engine's performance and improvement with long books, say 8,10,15 moves-long or even longer.

All the relevant search and eval parameters, but also the time management, are tuned to perform well after move 8 or 15. Not before that.

So how would you expect an engine, whose search and eval is tuned to statistically perform well starting late into the opening, to also perform well early into the opening, on move 3,5,6? The specific chess conditions early in the opening and in late opening are not quite the same, so an engine could play well into the early opening and find good moves, only if it has tuned its search and eval parameters with very short, 2-5 moves book, preferably shorter, for years.

How many engines out there have tuned their parameters with a very short book for years? I am afraid none.

So, when SF says d4 is the best move, I strongly doubt it, as there are much better chances it is c4 or e4.

Same from the black side: if SF suggests the French is a good option on e4, I strongly disbelieve that, as Sf has never tuned its parameters to perform well starting from the very first move.

[Haikouichthys]

I disagree. I have looked at TCEC games for example (long time control) and I see a lot of mistakes in the opening, some of them serious.

For example take this position from the Marshall Gambit:

[d] r1b2rk1/5p1p/p1pb4/1p1n2pq/3PR3/1BP3P1/PP1N1P1P/R1B2QK1 b - - 0 17

The main line goes Bf5! f3 Nf6!. f5 is usually what engines choose and it is playable, but usually considered inferior after Bd1!. Engines don't like Bf5 at all.

Marshall games are usually won or lost in the endgame and the opening is just full of decision points where you get a good endgame if you go the right way and a bad endgame if you don't. So I think book knowledge is still important, at least if you are not in correspondence where your search time is in days.

--Jon

I checked this position with both Komodo 8 and Stockfish (latest development version). Komodo quickly chose Bf5 (it briefly likes Bh3, but soon settles back on Bf5). Stockfish liked Bf5 at low depths and Bh3 at high depths. Neither engine selected f5 at any point.

(I agree with your general point, though)

[Laskos]

I disagree with lyudmil and I think that at least in 99% of the cases engines play the right move if you give them a long time to calculate.

if there are some cases when they do mistakes you can find it by the evaluation and if white play the right move based on long analysis in every move but black get the advantage after move 15 or move 20(at least 0.1 pawns for black) then it is better to do analysis to see where white went wrong in order to fix the book.

Starting point if we build a book for white is simply play the moves the engines suggest after a long analysis for white and moves that are in the database for black.

Same idea if we build an opening book for black but in the last case I suggest to try to fix only cases when the evaluation for white is at least 0.5 pawn for white because small advantage for white does not mean that black did a mistake.

That seems wrong. Engines' PV in a position, especially in the openings, are useless after 2-3 displayed half-moves. Even identical engines will play the expected reply only 70%-80% of the time, the next in PV 50%-60%, and so on decreasing. Engines have only a very fragmented outlook of the game, changing randomly "plans" and PV lines. Books need a trustworthy PV of 20 or so half-moves, a thing engines don't give.

The books are not built by leaving long analyzes of openings. Most lines are chosen by outcome statistics. Most games used are fast, for example PlayChess 1' and 3' per game.

[jefk]

[quote="Uri Blass"]
I think that in many cases there is more than one best move in the opening.
If engines do not play the move that you like it does not mean that they do a mistake.Uri[/quote]

you are both right (you and L.Tsvetkov) up to a certain point i guess.
The optimum as found after decades of analysis eg preferring Bb5 Spanish above the more drawish Bc4 Italian, ie discerning between playable opening moves is rather shallow. Depending on the sharpness of a line, there usually is a range of playable moves, which can be relatively 'wide' (8 moves or so).

However, everytime the comp is choosing a suboptimal line, especially if its not considered theory, eg 3.a3!? or whatever instead of 3.Bb5, the chances of a win are getting dimmer. And this is a cumulative effect !

Think of the chess theory by Steinitz, accumulating positional advantages, and only then starting combinations or a solid attack culminating in a win. Lyudmil is a strong player (2100) and knows more about such things than the average engine programmer. And it confirms my experience, that trying to select the 'best' opening move during the opening phase (average eg 10-20 moves)
helps the winning chances, or drawing chances for Black (winning with Black becomes more rare for me, unless the opponent is a weaker engine).

Choosing arbitrary opening moves is certainly possible if you have a faster a comp, you might get lucky now and then and get into some unknown 'theory' deeper in a sideline. But a better 'optimized' deeply thought-out book,
helps, besides human experience (except Carlsen maybe) this also is my experience, and i repeat my statement, its worth about 100 points against average. Example, i'm now at 2900 at ICC, at nr 1 in standard, with other comps as O. and P. both with about 3 times as much cpu's ranking about 50 points lower; although those ratings fluctuate, ofcourse.

NB if you continue to choose for the fastest, and thus most expensive comps, you might also choose to do without endgames bases. This also costs you some 80 points or so, or maybe more nowadays with the syzygy.
But why should you. I would do it only if i would have a positionally superior private engine which gives me winning games anyway, against stockfish and so on. Then indeed L.T.'s comment about the importance of the middlegame would be confirmed in practice. Such an engine is not easy to make however.
jef

jef

[jefk]

[quote="Lyudmil Tsvetkov"]
Give current top engines not 1,2 or 5 hours thinking time, but 10 days thinking time on standard opening positions, and they still will not find the right move in at least half of the cases, I think in 70-80% of cases, based on what I have seen from engine play in openings.[/quote]

agree, although it would depend on some parameters, eg such as you were suggesting in your 'compendium' (2010; is there any update ?),
my overall impression is that this is true, and thus building a new book from long engine analysis for a single position, without working on that book lateron by applying preferences, imho is a waste of time.

jef

PS as for the first option mentioned above, building a book with solid lines, rather than using long Stockfish analysis you might just as well use the Sanbaz 'perfect' book, which at least gives solid opening moves. But then i would prefer to update the preferences, not only by book learning, but a priori by indicating those moves which i consider 'best' from my experience. And although not super-GM quality (theory which is changing every month anyway), i already wrote it down:
http://www.lulu.com/spotlight/jefk
NB i don't think i'm spamming, it did cost me lots of time (years) to find out such things, and thus i think it would save engine bookmakers lots of time if you would use such a method at least for a start instead of trying to make a new book from scratch with engine analysis (some slight updates to my book later-on planned as well, to keep it up to par. In most cases its a matter of preference anyway, would you aim for a 'solid' draw -if you know the lines- against the Slav Meran with e3, or would you like to play sharp gambits as the Botvinnik, or, more likely, the anti-Moscow gambit, even although the latter is more risky..)

[Lyudmil Tsvetkov]

Let's be specific.

Two Knight's Defence:

[d]r1bqkb1r/pppp1ppp/2n2n2/4p1N1/2B1P3/8/PPPP1PPP/RNBQK2R b KQkq - 0 4

Which side favours that one?

Because, according to my SF, in the main line after d5 and Na5, the position is fully equal, or even black has some advantage.

According to Fischer however, white is close to winning, or straightforward wins this.

Who is right?

I bet that Fischer.
Fischer analysed that at home some 40 moves ahead. Engines will not be able to go 40 moves ahead/80 plies under whatever conditions in the current computer chess state.

Same with countless other opening positions. Engines simply lack the necessary depth of search to come up with the right results.

If anyone proves that indeed on the above diagram black is better, or the position is equal, I would be very happy.

Unfortunately, I am very much afraid white has substantial advantage above, in the range of at least 60cps, so winning or close to winning.

[Lyudmil Tsvetkov]

Another one.

[d]r1bqkbnr/1ppp1ppp/p1n5/1B2p3/4P3/5N2/PPPP1PPP/RNBQK2R w KQkq - 0 4

Ruy Lopez, Morphy Defence/a6

Which is the best move here?

As, according to my SF, Bc6, the Exchange Variation, is the best continuation, with Ba4 ranking many centipawns behind.

So, if I have to trust SF in the opening, I have to play Bc6 every single time.

Those positions have been analysed extensively by humans, and it is proven beyond any doubt that Ba4 is better than Bc6, much better indeed.

How do you want me to trust SF here?

I know where to trust SF: in open middlegames and endgames with abundant tactics; there the engine excels.

However, I do not trust SF in closed positions, specific theoretical endgames, as well as in a large bunch of openings.
The search depth and understanding of engines for such positions is simply too low.

[Lyudmil Tsvetkov]

I try quick shootout with SF after Bc6, and get the following position 15 moves later:

[d]2kr3r/1ppq1bpp/p1pb1pn1/8/3NP3/2Q1BP2/PPPN2PP/3RR1K1 b - - 0 14
SF says here full equality instead of the 30-35cps white edge it was seeing initially.

In the other line, Ba4, after a quick shootout lasting the same length, SF gets to the following position:

[d]r2q1rk1/2pnbppp/3p4/p1p1p3/2P1P3/1PNPBQ1P/P4PP1/1R3RK1 w - - 0 17

which is evaluated in exactly the same way SF evaluated Ba4 initially, with some 5-10cps white edge.

But please note, that SF plays unbelievable opening moves here, c4 instead of c3, so completely untheoretical.

I am afraid what will happen if some inveterate theoretician looks at the way SF plays the Ba4 line of the Ruy Lopez, the man might not die, but a heart attack is a probable consequence.

[zullil]

Let's be specific.

Two Knight's Defence:

[d]r1bqkb1r/pppp1ppp/2n2n2/4p1N1/2B1P3/8/PPPP1PPP/RNBQK2R b KQkq - 0 4

Which side favours that one?

What do you mean when you say a position "favors" one side or another? (Serious question---please consider deeply.) The value of any position is simply White wins, Black wins or draw.

More specifically, what does "White is better by 60 cp" mean? Is this different from "White is ahead by 20 cp"? Is it different from "Black is ahead by 20 cp"?