Sunday, April 19, 2020

The French Connection: Volume 36

Hello everyone and welcome to the thirty-sixth edition of The French Connection. The featured game in this article comes from the World Zone Individual Championship Preliminaries. Sounds like an ear full, doesn't it? It is a correspondence event played on the biggest world wide correspondence site, the "ICCF", or "International Correspondence Chess Federation". There are four stages, and the stages are staggered. Not everybody starts at the same stage. You have the Preliminaries, Semi-Finals, Candidates, and Finals. Where you start depends on your rating. Those over 2400 start at the Candidates stage. Those 2100 to 2399 have to qualify for the Candidates via the Semi-Finals, and those below 2100 have to qualify for the Semi-Finals via the Preliminaries.

Now why is this information important? Correspondence chess is a lot different than over the board chess. With the aid of books and engines, risky, unsound lines that you can try over the board really don't work here. Also, the draw ratio is very high, especially at the higher levels. With the additional information available, even low rated players can end up playing GM level moves, and so not only is it difficult to win, but it is extremely difficult to win with Black, which should explain why you are about to see a drawish system played by Black against a lower rated player in the feature game.

So why isn't everyone rated 2400 and above given that engines can be used? It isn't until you actually play correspondence chess that you realize that computers are not as strong as people make them out to be! Computers that have played against GM's, like Deep Blue in the 1990's against Garry Kasparov, are juiced up by programs developed by human GMs. Without these additional programs, like opening power books, endgame table bases, along with others, you see computers playing a lot of strange moves with highly inaccurate evaluations, especially in the opening and endgame. Have you ever gone on a site like, examine a game you played, and wonder why after Black's 6...e5 in the Classical King's Indian, the computer claims White is up almost a full point? Have you ever wondered why, in a position where immediate tactics to win a piece are not available, it thinks that a King, Rook, and Knight for White versus a King and Rook for Black, no pawns for either side, is plus 3 for White? You ever see it recommend a move in the middle game for Black, saying it is -0.7 (meaning better for Black), you play that move, it gives a non-pawn move for White, and then the next move for Black, it wants you to go right back from where you came from, and still thinks it's -0.7? If you keep going back and forth, it should read 0.00 due to a draw by repetition. Every see a computer say that one move is -0.7 for Black, another move is +0.3, meaning slightly favors White, but once you make the first move, it flips to 0.00, and once you make the second move, it suddenly thinks Black is slightly better with a -0.4 evaluation? Long story short, computer evaluations cannot be trusted, and so there is still a major human factor in correspondence chess, and so a 2000 over the board player may not be a 2000 correspondence player. He could be a 2400 correspondence player as there is no time pressure, or he could be a 1600 correspondence player as he relies too many times on computer moves and doesn't realize that further human analysis and judgment is required.

So what can computers do for you in correspondence? They are excellent at figuring out very deep, forcing lines that humans may not see, and they are great to use as a blunder check, and so you are not going to win games with 3-move tactical blunders very often as the computer will tell them when they try move "X" and see that it drops a Rook to a weird fork 5 moves later that is totally forced.

Therefore, the next time you come to me and say "but the computer says blah blah blah is best", you see why I will often blow it off unless it leads to a forcing tactical sequence that wins material with no positional consequences.

It should also be noted that the ICCF has three "zones". The Europe Zone, the Africa/Asia Zone, and the World Zone. The World Zone includes Argentina, Australia, Brazil, Canada, Chile, Cuba, Ecuador, Guatemala, Japan, Mexico, New Zealand, Nicaragua, Panama, Peru, Turkey, Untied States, and Venezuela. This event is restricted only to players from those countries, which shows the difference between a Zonal event, and something like a Norm tournament or World Correspondence Championship event.

With this background information in mind, let's take a look at the feature game.

World Zone Individual Championship, Preliminary Round
W: Gerald Thomas (1794 - USA)
B: Patrick McCartney (1920 - USA)

1.e4 e6 2.d4 d5 3.Nd2 dxe4

This line, which can also be played against 3.Nc3, is known as the Rubinstein Variation. Black gives up the fight for the center, and accepts a slightly passive position. White has a space advantage, but Black has no real weaknesses in the position, unlike the weaknesses seen in many other lines of the French Defense. For example, in the Open Tarrasch (3...c5), Black often has to either bring his Queen out early (4.exd5 Qxd5), or else deal with an isolated pawn (4.exd5 exd5 with a later dxc5 or ...cxd4). In the Closed Tarrasch (3...Nf6), Black will usually have a backwards e-pawn on a semi-open file. In the Winawer (3.Nc3 Bb4), Black has to worry about the dark squares on the Kingside.

The downside is that Black is passive to start the game, and this line is often used as a drawing weapon, and at the GM level, White tends to win fewer games here than in the more highly theoretical lines, but wins by Black tend to be very close to non-existent. So why would I play this line against a player of lower rating? The reason is two-fold. The first is that against a lower rated player in correspondence chess, minimizing risk with Black can be a good thing. In a bracket with all players below 2100, you can bank on there being more wins overall than in the later rounds, and so getting a draw with Black could mean the difference between advancing and not advancing. The second is that lower rated players are likely to make a few second-best moves. Again, with the use of computers, banking on a tactical blunder is highly far-fetched, but a positional error in a position that isn't full of fireworks isn't out of the question, and actually, we will see that happen in this game.

4.Nxe4 Nd7 5.Nf3 Ngf6

This is the main starting position of the Rubinstein. While there may be a few sidelines, like the Fort Knox, where Black develops his light-squared Bishop early, this position is the position you will see the majority of the time when Black takes on e4 on move 3. White has a significant number of choices here, which can be viewed as the other downside to the Rubinstein. Black has no weaknesses right now, but White has numerous choices at this point. He can:

  • Trade Knights on f6 and then follow up with 7.Bd3 (highly popular at the amateur level), 7.c3 (Kasparov's specialty which can be highly dangerous for Black if he doesn't know what he's doing), 7.Be3, or even go for a fianchetto approach with an early g3.
  • He can pin the Black Knight with 6.Bg5, and once kicked by 6...h6, only then trade the Knights on f6 and follow up with the retreat 8.Bh4, maintaining the pin, trading of Bishop for Knight with 8.Bf6, or relieve the pin and retreat with 8.Be3, seeing the h6-pawn as a weakness compared to the 6.Nxf6 and 7.Be3 line shown in the previous bullet.
  • He can play 6.Bd3 and castle Kingside, which was Khalifman's Recommendation in "Opening for White According to Anand" back in 2006.
  • He can play 6.Bd3 and take a more dynamic approach by continuing development and castling Queenside.

So as we can see, there are a significant number of moves that Black needs to know how to respond to. Outside of Kasparov's 6.Nxf6 and 7.c3, none of them are extremely dangerous, but Black can very quickly get into an inferior position if he doesn't know precisely what he is doing.

6.Bg5 h6 7.Nxf6+ Nxf6 8.Bh4

So White decided to go with the 6.Bg5 approach and maintaining the pin on the Knight.


This pawn break is Black's main weapon in the vast majority of lines of the Rubinstein, and this line is not one of the exceptions. Now, once again, White has a lot of options.


White decides to relieve the central tension a little too early, and this line causes very few problems for Black. Other options by White include:

  • 9.Bb5+ should not be a problem for Black, especially given that Black's worst piece is being traded. After 9...Bd7 10.Bxd7+ Qxd7 11.Qe2 Be7 12.O-O-O O-O 13.dxc5 Qc6 14.Ne5 Qxc5 15.Bxf6 Bxf6 16.Nd7 Bxb2+, White has nothing better than allowing the perpetual after 17.Kxb2 Qb4+ as if he tries to escape the draw by walking his King to d2, Black can pin the Knight with ...Rfd8, and while White can hold on to the Knight with Ke1, precise play is required for White to survive, and even then, the line has been figured out to a draw.
  • 9.Bc4, which can lead to tactical issues for Black if he takes on d4 immediately, but 9...a6 10.O-O and only now 10...cxd4 and Black is fine here. If White tries to stir up an attack by sacrificing a pawn with 11.Qe2 Qb6 12.Rad1, Black has the time to take the pawn with 12...Qxb2 and again, the most White can really hope for is a draw here.
  • The pawn sacrifice line 9.Ne5 has been figured out. Black equalizes with 9...Qa5+ 10.c3 cxd4 11.Qxd4 Bc5 12.Qf4 Bd6 13.Bg3 O-O 14.Be2 Nd5.
  • I think that if White wants any hopes of a win with this line, he has to try 9.Bd3 or 9.c3. Black should be ok in both of these lines, but more accuracy is required from Black, and it's not a simple one-liner that equalizes the position for him.


Trading Queens would be a mistake here. After 9...Qxd1+?! 10.Rxd1 Bxc5 11.Be2 Ke7 12.Ne5 g5 13.Bg3 Ne4 14.Bh5 Rf8 15.Bf3 Nf6 16.h4 Rg8 17.hxg5 hxg5 18.Rh6, White's advantage is significant.

10.c3 Qxc5


This move does not make much sense. White has a slight lead in development, and while 11.Bd3 Bd7 12.Qe2 Bd6 13.Bg3 Bxg3 14.hxg3 Bc6 is equal, there is a lot more room for error here, and White has a few approaches he can take. He can play the safe 15.O-O, or he can try 15.Rh4 and castle Queenside. Equal does not mean drawn, but this Queen trade simple eases things for Black.

11...Qxd4 12.Nxd4 Bd7 13.Bxf6

I don't like this move for White at all. Yes, it slightly wrecks Black's pawn structure, but the extra central pawn and the Bishop in an open position is more important than the pawn structure. As we will see in the game, White is never able to take advantage of Black's kingside. I didn't like White's 11th move either, and so it's hard to recommend anything, but if I had to play this position and was White, I'd probably recommend 13.O-O-O here.

13...gxf6 14.Bb5 Bc5 15.Bxd7+ Kxd7 16.O-O-O

This is really the first time that Black needs to make a decision. Should he move his King to c7, e7, or move one of the Rooks to d8 before moving the King?


I think this is Black's best move. The problem with 16...Rad8 or 16...Rhd8 is 17.Nb3+, which forces 17...Bd6 and Black's pieces are cluttered on the d-file, making the Rook somewhat ineffective as it has to wait for the King and Bishop to both move. It also puts the Bishop on a more passive square than c5.

It is a little less clear whether 16...Kc7 could be better or not. White has a 3-on-2 pawn majority on the Queenside. While Black may have both doubled pawns and an isolated pawn, his majority is still on the kingside. With pieces still on the board, there is no reason for Black to take a defensive stance and use the King to block the White queenside pawns. It makes a lot more sense for it to join its pawn majority. In addition, it also creates a tactical defense to the isolated h-pawn that we are about to see a few moves later.


Now 17.Nb3 can be answered by 17...Bxf2 rather than 17...Bd6, which would be winning for Black.


This move looks like it drops the h-pawn, but it doesn't. This is the correct Rook to put on the d-file as otherwise, they will become disconnected as the Black King is about to be forced to the back rank with an upcoming check. With this move, the Rooks remain connected.

18.Nf5+ Kf8!

The correct move as otherwise, Black's next move wouldn't be possible.


Taking the h-pawn loses: 19.Nxh6? f5! traps the Knight. After 20.g4 Bxf2 21.Re2 Bh4, the White Knight can't get out and 22...Bg5 is coming.


Had the King gone to e8 on the previous move, this move wouldn't be possible as White could just take it with the Knight due to the pin.

20.Rxd8+ Rxd8 21.Rd1

So now a critical decision from Black. It is inevitable that the Rooks will be traded. It makes no sense to re-locate the Black Rook to a passive position and give White the d-file. So the Rooks are going to go. The bigger question is the minor pieces. Should Black trade the minor pieces first, since taking the Knight comes with check? Or should Black take the Rook, which allows White to recapture with the Knight and keep the minor pieces on the board?

Here is where one must be careful about using computers and solely relying on their assessments. When people ask "What is the numerical assessment of the position?", I question why that is relevant. Sure, if the computer says it's +6, I would have zero doubt that White is winning, and if it says -8, then yes, Black's winning. Sure! But here, if I run the current position on Shredder,'s computer engine, it says that 21...Bxe3+ is "-0.55" and that 21...Rxd1+ is "-0.48". So if you blindly rely on the computer, you would argue that 21...Bxe3+ is a slightly stronger move. I am here to tell you that this is 100 percent inaccurate!

Let's say Black plays 21...Bxe3+ right now. White will of course take the Bishop. After 22.fxe3 Rxd1+ 23.Kxd1 Ke7, yes, Black does get to the fourth rank before White does, but it's not enough. After 24.Kc2 Kd6, White has the critical move 25.c4!, taking the d5-square away from the Black King, and giving him only one square, e5, and not two squares for Black to toggle upon. After 25.c4, it still claims a slight advantage for Black after 25...a5 or 25...Ke5, but as you play each of these, suddenly it changes its assessment to 0.00 in both cases, 25...a5 being responded to with 26.b3 and 25...Ke5 being responded to with 26.Kd3.

Now, any player can run random moves through the bot and see if the favorable assessment maintains itself or if it dies like it did here. You need to be able to do more than that in correspondence chess. I figured out myself that taking the Knight was not the way to go for the following reasons, beyond just seeing the triple zeroes:

  • The position is fairly open with pawns on both sides and no immediate tactical threats by the White Knight. Black has the better minor piece!
  • In King and pawn endgames, assuming all other factors are equal, like how far up the board your majority is, the side with the smaller majority has the advantage. Therefore, 3 on 2 is better than 4 on 3 (the situation in this game), 2 on 1 is better than 3 on 2, and 1 on 0 is better than 2 on 1.
  • White's majority is in its ideal formation. Three pawns that are not isolated or doubled. Black's majority is crippled. Black would have e, double-f, and h vs e, g, and h.
  • Black's majority is closer to the center of the board. In a King and pawn endgame, outside pawns are stronger.

So we see here an actual occurrence where relying on computer assessment would be an error. Black should still be able to hold the position, but he is by no means better like the computer originally said. Also, Black has to be careful not to over-press and suddenly be in a lost position. For example, in the line above after 25...Ke5 26.Kd3, the move 26...f4 would not be answered by a capture, but rather, 27.b4, and while this should still be drawn, both sides have to be aware of what they are doing, and not rely on computer assessments until either one side suddenly has an alarming advantage, like +3, or the total number of pieces becomes 7 or less as then table bases can be used to determine if one side wins or if it is a draw.

So after this assessment, Black's choice should be easy.

21...Rxd1+ 22.Nxd1 Ke7 23.Kc2 e5

Black has the advantage due to the advanced pawn phlanx and the better minor piece, and so White must be thinking from a defensive mentality here.


I think White would be better off advancing a Kingside pawn. Which one? That raises an interesting question.

Playing 24.g3 seems counter-intuitive because it puts the pawns on the dark squares, which is the color complex that Black's Bishop occupies. That said, it also takes a bunch of dark squares away from Black. Personally, I think this is White's strongest move. Black has to be careful. For example, after 24...e4 25.Ne3! Ke6 26.a4 Bxe3?! 27.fxe3, black must find 27...Kd5 just to maintain equality. Black's majority is stopped, and White has the Queenside majority, and only two results would be possible here, a White win or a draw.

Of course, Black is not forced to play 24...e4, and can play something like 24...Ke6, but after 25.a4, I don't see how Black is going to make progress without taking such a risk. Sure, he can play waiting moves like 25...Bd6, but without an error by White, I don't see how either side can make progress, and a draw is the likely result.

While I think 24.g3 is best, I also think that 24.f3 is interesting, though Black might be able to take advantage of the open dark squares. After 24...Kf6 25.Kd3 Bh1 26.h3 Kg5 27.Ne3, Black has two viable options:
  1. 27...Bxe3 28.Kxe3 f4+ 29.Ke4 f6 30.a4 Kh4 31.b4 Kg3 32.c4 Kxg2 33.c5 f5+ 34.Kxf5 (34.Kxe5? Kxf3 35.b5 Kg3 36.c6 bxc6 37.bxc6 f3 38.c7 f2 39.c8=Q f1=Q is clearly better for Black with the extra pawn) 34...Kxf3 35.b5 Kg3 36.c6 bxc6 37.bxc6 f3 38.c7 f2 39.c8=Q f1=Q+ 40.Kxe5 and Black is better as his King is harassing one of the White pawns directly, but it's unclear whether it is enough for Black to win.
  2. 27...Kf4 and now I think 28.Nc4 is stronger than the overly passive 28.Nf1. After 28.Nc4, Black can try 28...e4+ 29.Ke2 Bc5 30.b4 Be7 31.Ne3 Bg5 or 28...h5 29.Ke2 Bc5 30.b4 Bf8 31.Kf2 h4, but in neither case do I see Black getting anywhere.

I still think 24.g3 is safer than 24.f3 due to line A, but the latter may turn out to be playable as well.

24...Ke6 25.b4 e4+ 26.Kc4?

This move leads to problems for White. I think that 26.Ke2 was forced, attempting to stop Black. The problem with advancing the King is that Black will be able to create a zugzwang position where White will run out of viable moves and will be virtually forced to get out of Black's way!

26...Bd6 27.h3 f4 28.b4

It does not help White to play 28.Kd4 as the King can be chased away via 28...f5 29.a4 Be5+ 30.Kc4 Bf6 and the King can then enter via e5 and re-route the Bishop.

28...Ke5 29.a5 Be7 30.g3?

Other moves are likely not to work either and White is likely already lost, but the move played makes Black's task very easy. After a move like 30.f3, Black can win by advancing the e-pawn and once again, making White run out of productive moves. For example, 30...e3 31.Kd3 h5 32.Nb2 Kd5 33.Nd1 Bf6 34.c4+ Kc6 35.Nc3 Bxc3 36.Kxc3 h4 37.Kc2 a6 38.Kd3 b6 39.axb6 Kxb6 40.Kc2 a5 41.bxa5+ Kxa5 42.c5 Kb5, winning.

30...fxg3 31.fxg3 Bg5

The Bishop is headed to e1.

32.Nf2 f5 33.Nd1 Bd2 34.b5 Be1

This will allow Black to force through the f-pawn as White either has to advance the g-pawn, or lose it.

35.g4 f4!

Far stronger than trading on g4.

36.a6 b6!

White is in Zugzwang! If he moves his King, Black will advance his e-pawn. If he moves his Knight, Black will advance his f-pawn.


Or 37.Kb3 e3 38.Kc2 Ke4 39.h4 f3 and the f-pawn can't be stopped.

37...f3 38.Nd1 Kf4 39.Kd4 e3! 40.Kd3

Or 40.Nxe3 Bf2 wins the Knight.

40...e2 41.Ne3 Bxc3 42.Nc2

White can win the Bishop, but then Black promotes his e-pawn.

42...e1=Q 43.Nxe1 Bxe1 44.g5 hxg5 0-1

So what we saw here was an instructive game that illustrates many aspects of the game:
  • An introduction to the Rubinstein Variation of the French Defense
  • Paying close attention to detail and correctly deciding between two moves that look to be equal in value. For example, Black's 16th and 18th moves, both of which feature significant differences between two moves that at first glance appear to be equally playable.
  • Understanding when to trade and when not to! Black was right to keep the minor pieces on the board, and while the analysis shows a few scenarios where Black can trade, the first opportunity, back at move 21, it was most certainly in Black's interest not to trade the minor pieces.
  • An illustrative example of how a Bishop can certainly be better than a Knight, despite its inability to reach half the squares on the board. The Knight was dominated by the threat of Black's pawns advancing. The Bishop was able to improve its position over the course of the endgame while the White Knight was virtually frozen for the duration of the endgame.
  • DO NOT ASSUME that all computer moves are gospel! Computers are extremely strong at calculating long, forcing sequences that the human mind will often fail to find because of an obscure move in the sequence, whether that be a move that violates principles, like a Knight going to the edge of the board, or a major sacrifice, like giving up a Rook that 13 moves later leads to regaining the material and then some or mating the opposing King. That said, there are other aspects of the game where computers are weak, and positional evaluation can be one of them, along with openings and endgames. We saw the computer advertising that Black should trade the minor pieces on move 21, but we saw here that trading the pieces only causes trouble for Black, and while leaving them on might not be a forced win for Black, it is White, not Black, that has to prove it in the endgame. Computers also have a tendency of overvaluing material above positional aspects, and so don't just assume that what the computer says must be right. That is often not the case at all!

This concludes this edition of The French Connection. Til next time, good luck in all of your French games, Black or White!

No comments:

Post a Comment