Kingside fianchetto

[Lyudmil Tsvetkov]

I am very interested in your opinion why, according to you, all engines are extremely reluctant to play any openings with a kingside fianchetto.

A kingside fianchetto is a very useful structural technique, very sound positionally and very appropriate in almost all kinds of openings.

A kingside fianchetto, bishops developed in the opening/middlegame on g2/g7, has at least the following comparative advantages to other ways of development:

- it develops the king side
- the bishop on g2/g7 strengthens the king shelter, as it is always good to have a minor piece close to the king
- at the same time, such a bishop controls the center, and can possibly develop big activity on the long diagonal
- a fianchettoed bishop is good in both open and more closed, KID-type positions

It is not without reason that openings with kingside fianchettos were preferred by players of different playing styles, sharp attackers, like Kasparov, universal players like Fischer, and also pronounced positional players like Smyslov, for example. They played a wide range of openings featuring kingside fianchettos, not only the KID and the Gruenfeld, but also the Catalan, the Reti, and Smyslov used this technique practically everywhere, from the Closed Sicilian to different lines of the English Opening and the Ruy Lopez for black.

It is to note, that the above players were not only world champions, but also among the most convincing world champions, and very good theoreticians at that. So that the kingside fianchetto should be very useful indeed, and very often preferable to other developing lines.

In distinction to a kingside fianchetto with Bg2/Bg7, a queen side fianchetto with Bb2/Bb7 is almost always not that convincing. The point is that the bishop there only attacks the center and is developed, but does not shelter the king at the same time, so that a queenside fianchetto is less functional than a kingside one.

And indeed, playing the Larsen Opening with 1.b3 seems to be a bit dubious at the least, unless white's aim is to fight for equality that might also not be available. Of course, some might argue that Fischer scored with the opening, the so called Fischer-Larsen Attack, extremely well, but you have to be Fischer to score well in objectively worse positions. Besides, he did that at the height of his career. Fischer also scored very convincingly with the Exchange Variation of the Ruy Lopez, where, according to me, black should have some advantage.

The same should be true of the queen side fianchetto on b7 for black in openings like the Queen's Indian Defence. The Queen's Indian, although engines might assess it as a good opening, almost fully equalising, and also having a good reputation of drawishness in some lines, is as a whole not a very sound opening. White should get some reasonable advantage with very precise play.

So my conclusion would be that, while it is not worth to incentivise queenside bishop fianchettos, it is an absolute must for any respectable engine to incentivise kingside fianchettos. However, Rybka is the only one of the engines I know that regularly or at least on occasion plays such kingside fianchettos. All other engines, including Komodo and SF, are reluctant to do so.

And urging the engine to play kingside fianchettos seems to be a very ordinary task without any observable deficiencies, for example by bonussing bishop psqt for the g2 and g7 squares. It is easy to do, with no additional calculation cost, and should bring at that real added value in terms of improved positional understanding.

I think adding some 20, or even 30cps bonus in psqt for a bishop placed on g2/g7, only for the opening and middlegame of course, should readily resolve the problem.

This could also help engines to improve their knowledge of the KID.

So my question to you is, if kingside fianchettos are indeed so sound and useful, why then no or almost no engine of more than 500 existing ones prefers in any way such way of development?

[PK]

Rodent gives a marginal bonus for a fianchetto on a wing where its king is present and a marginal penalty for a missing fianchettoed bihop. It is also applied in the unlikely event of queenside castling. The bonus and the penalty are however rather small: +4 and -4 centipawns respectively. This comes together with another marginal bonus for a bishop defending a square next to own king, another 4 or 5 centipawns.

[velmarin]

These things are what define human style and the style of silicon,
Open once the diagonal, with the mobility and the defence of the King is sufficient.

Engines do not like the fianchetto, they don't understand,
But they attack him very well.

[Ralph Stoesser]

In SF there is a tricky static eval term that gives an opening/midgame bonus for a minor behind a pawn of the same color. It is not a genuine idea from SF developers if I remember right, but stems from Don Daily. However, that bonus should help a bit to encourage an early fianchetto development, but might not be enough. If you give SF a very early opening position where g2-g3 already has been played, in most cases SF will be happy to develop the bishop to g2, but it will not make the move g2-g3 in the first place. One reason might be that program's king safety code doesn't sufficiently understand that king side fianchetto + king side castle is a very robust setup against king attacks. Also short-term or mid-term mobility and king attacking prospects for a bishop on g2 might not be the best. From my experience, if you give SF enough time, it will find a fianchetto setup, if it's really appropriate, but will not do it too early in the opening phase without a special reason.

[Lyudmil Tsvetkov]

Rodent gives a marginal bonus for a fianchetto on a wing where its king is present and a marginal penalty for a missing fianchettoed bihop. It is also applied in the unlikely event of queenside castling. The bonus and the penalty are however rather small: +4 and -4 centipawns respectively. This comes together with another marginal bonus for a bishop defending a square next to own king, another 4 or 5 centipawns.

Thanks Pawel.

Will you excuse me, if I tell you that I have not played very often with Rodent lately? It is the SF syndrome, I have to play mostly the best, so called rat-racers.

Now I remember, that another strong engine that frequently fianchettos its kingside bishop on g7, is Spark. But why only on g7, only with black, and not when playing white?

As I play most of my games against SF, I must tell you that I did not see this engine fianchettoing its king bishop, eiether with black or white, even once! It frequently does a queenside fianchetto on b7 in the Queen's Indian, obviously because of some mobility considerations, but never a kingside one.

How can this engine play so strong, when it misses the best opening moves?

I would be happy if other authors/enthusiasts remind me about possible other existing engines doing kingside fianchettos without a book.

[Lyudmil Tsvetkov]

These things are what define human style and the style of silicon,
Open once the diagonal, with the mobility and the defence of the King is sufficient.

Engines do not like the fianchetto, they don't understand,
But they attack him very well.

The problem is that why do not they understand it?

It is true they attack very well, but less than 1% of engines doing kingside fianchetto, when theoretically this is one of the best moves?

[Lyudmil Tsvetkov]

In SF there is a tricky static eval term that gives an opening/midgame bonus for a minor behind a pawn of the same color. It is not a genuine idea from SF developers if I remember right, but stems from Don Daily. However, that bonus should help a bit to encourage an early fianchetto development, but might not be enough. If you give SF a very early opening position where g2-g3 already has been played, in most cases SF will be happy to develop the bishop to g2, but it will not make the move g2-g3 in the first place. One reason might be that program's king safety code doesn't sufficiently understand that king side fianchetto + king side castle is a very robust setup against king attacks. Also short-term or mid-term mobility and king attacking prospects for a bishop on g2 might not be the best. From my experience, if you give SF enough time, it will find a fianchetto setup, if it's really appropriate, but will not do it too early in the opening phase without a special reason.

Excellent explanation, Ralph, completely agree with what you say, but I think SF should do kingside fianchettos, both with black and white, even without any special reason, provided similat setups are either the best, or among the best in a variety of lines.

If I remember correctly, minor behind pawn is worth in SF somewhere some 15cps SF values, 8cps standard ones or even lower, hardly sufficient to encourage g2/g7 fianchettos, provided that short-term mobility and speed of development, direct control of center are not optimal.

What should I say, in spite of that SF claims to be a modern engine, it does not play in the hypermodern style, associated with indirect control of center and fianchettos, but more in the classical style, like Tarrasch and Steinitz.

[Lyudmil Tsvetkov]

Struggling to scroll through the miles-long Rybka-dedicated thread, I came here again.

I made the effort to count the number of times Fischer used a kingside fianchetto. For a reference year I took 1970, so when Fischer was already in a top from, but his openings still were not dictated by tactical considerations, related to the forthcoming Candidate matches and World Championship Final.

For the year 1970, I counted 80 serious tournament games played by Fischer (I excluded blitz and simul games), in 40 of which he employed a kingside fianchetto, either with white or black. In another 4 games, he used a queenside fianchetto, and in 36 games no fianchetto appeared. Games with lacking fianchetto are due mainly to Fischer quite often playing the Ruy Lopez and the Sicilian with white, but he also frequently fianchettoed his king bishop in the Sicilian.

So, roughly in 55% of all games Fischer used kingside fianchetto. I noticed at that, that when he played weaker opponents, he often avoided to fianchetto his king bishop, obviously thinking he could win the easy way, even with weaker lines.

The ration of kingside to queensde fianchettos is 10 to 1, indicating that queenside fianchettos are really far less promising than kingside ones.
How do you interpret those numbers?

For me, Fischer of 1970-1972 is the tsrongest player in history, and also theoretically best prepared.

How many engines would play without book kingside fianchettos in more than half of all games?

So if you ask me again, yes, I think a kingside fianchetto is an optimal structural approach, and engines should do everything possible to include in their reperoires this mightly weapon as often as possible.

So I would give not 30cps, but even half a pawn for a bishop on g2/g7 in the opening. At first it might be difficult, but then engines will learn how to play those lines.

But are there any engines out there that actually listen when strictly positional terms are treated?
Engines want tactics, tactics and more tactics. And therefore, they would not like to change also their search so that it benefits also positional, instead of only tactical terms.

So again, how do you interpret the above statistics?

[Lyudmil Tsvetkov]

Eagerly waiting for your comments on the above statistics, just to quickly add, please do not tell me that you do not care about what Fischer played, as modern top engines play much better than Fischer.

The truth is that current top engines play some 300 elo weaker than top GMs positionally, but are some 600 elo better tactically, and also do not get tired as humans do, or get nervous.

So Fischer knew better what is good and not, but engines could perform better even without that knowledge. Still, with new knowledge added, they will perform even better.

Engines play openings terribly as a whole, ask Jon about that.
The reason is simple, while a top engine will outcalculate and outperform you in an unfamiliar situation with limited time to think, the case with theoretical knowledge is completely different, as it is the fruit of thousands of hours of analysis provided by a large number of people over an extended period of time.

Not matter how long you give a top engine to think over a specific opening position, be it Komodo or SF, it will rarely or never pick the most precise move, unless the position is relatively simple. 20 hours will not be enough for Komodo or SF to pick the right move.

That is why engines generally think openings that are good are bad or vice-versa.
For example, engines tend to think the Exchange Variation of the Ruy Lopez is good for white, when it is actually quite the opposite; or that the Queen's Indian and Nimzovich are ideal openings for black, when most likely it is quite the opposite; or that the King's Indian Defence is bad for black, when it is the opposite.

That is why engines, even top ones, do not like to fianchetto their king bishops.

Still, very interested in your opinion, how do you interpret the above statistics?

[Lyudmil Tsvetkov]

Here a typical Fischer game with a kingside fianchettoed bishop, vs Cuellar Gacharna from the Sousse Interzonal in Tunisia, 1967.

[pgn][Site "Sousse izt"]
[Date "1967.??.??"]
[Round "4"]
[White "Cuellar Gacharna, Mig"]
[Black "Fischer, Robert J"]
[Result "0-1"]
[Opening "Benoni, Classical, 8.Bg5"]
[ECO "A71"]
[MLNrOfMoves "41"]
[MLFlags "000000"]

1. d4 Nf6 2. c4 c5 3. d5 e6 4. Nc3 exd5 5. cxd5 d6 6. Nf3 g6 7. e4 Bg7 8. Bg5 h6
9. Bf4 g5 10. Bc1 O-O 11. Nd2 Nbd7 12. Be2 Ne5 13. Nf1 b5 14. Bxb5 Qa5 15. Ng3
c4 16. O-O Rb8 17. Qa4 Qxa4 18. Bxa4 Nd3 19. Bb5 Ng4 20. Nge2 Nxc1 21. Raxc1 Ne5
22. b3 cxb3 23. axb3 a6 24. Ba4 Nd3 25. Rc2 f5 26. Ng3 f4 27. Nge2 f3 28. Ng3
fxg2 29. Kxg2 Bg4 30. Nf5 Nf4+ 31. Kg3 Bxf5 32. exf5 Bxc3 33. Kf3 Be5 34. Ke4
Rb4+ 35. Rc4 Rfb8 36. f6 Kf7 37. Kf5 Rxc4 38. bxc4 Ne2 39. Re1 Nd4+ 40. Kg4 h5+
41. Kh3 Kxf6 0-1
[/pgn]

As you can see from the game, the fianchettoed bishop on g7 stays there all the game, throughout the entire middlegame stage, does not even move once, and the first move it makes on move 32, the game is won.

So the conclusion might be, you just have to develop your bishop on g2/g7, then do nothing, and wait for the win.

[d]r1b2rk1/p4pb1/3p1n1p/qB1Pn1p1/2p1P3/2N3N1/PP3PPP/R1BQ1RK1 b - - 0 16
Look at this position, what is the bishop on g7 worth?
No mobility, attacks nothing whatsoever, is even hemmed in by own pawns, and still, this is probably the strongest piece on the board, that decides the game.
Although Bg7 does not attack directly anything at the moment, it reigns supreme over the h8-a1 diagonal and will soon start attacking. Besides, as said, it shelters the black king perfectly.

[d]1rb2rk1/6b1/p2p3p/3P2p1/B3P3/1PNn2N1/2R2PKP/5R2 b - - 0 29
On move 29, into the late middlegame, and close to the endgame, the bishop is still there, sheltering the king, controlling the center, and already directly attacking.

In most games it is like that: the bishop is developed on g2/g7, stays there the entire middlegame, and the first move it makes, the game is won. So this is a pattern.

I think the above clearly indicates that bishop on g2/g7 is significantly stronger than bishop on f3/f6 and e4/e5, but psqt would suggest otherwise. A Bg2 has nothing to do on either f3 or e4, same with Bg7, nothing to do on f6 or e5.

So why then assign bigger bonus for f3 and e4 squares in psqt. As long as it is on the long diagonal, it should rather be the other way: g2 square is due a significantly bigger bonus in psqt than f3 and e4 squares. But, please note, the same would not be quite true of Bb2, where the c3 and d4 squares are due their normal bigger bonus points than the b2 square.

What to do, psqt, everything on the chess board is asymmetric and strictly specific.