Recently one of my readers asked whether I play against the computer for training or just for fun. I answered that it was probably 75 percent for fun and 25 percent for training. But perhaps I was wrong, because the two are not mutually exclusive! Here’s a game I played yesterday that, in my opinion, was both fun and had a lot of training value.
Position after 27. Na4. Black to play.
FEN: 7r/1r2kp1p/2bnp1pP/p2p4/N1pP4/P3P3/2B1KPP1/1R5R b – – 0 27
I was Black in this position, where Shredder (set at a rating of 2201) has just moved its knight from b2 to a4. Obviously Black is in good shape, with an extra pawn on c4 that is protected and passed, but that doesn’t mean the game will win itself. After White’s move there are going to be some exchanges, which means that the position may change dramatically in the next few moves. Because this was a key transitional moment, I thought it was a good time to take my time-out.
(Explanation for people who are new to this blog: One of my training tools when playing the computer is to allow myself one time-out per game to analyze the position as deeply as I want. I call this Matrix chess. I like the technique because it’s like analyzing a position from a tournament game at home, but with an important difference: the game isn’t over. You don’t have 20/20 hindsight to guide your analysis, and you get immediate feedback on whether your analysis was right or wrong.)
What do you think Black should do here? I considered three moves, which have different goals: 27. … Bxa4, 27. … Rxb1, and 27. … R8b8.
My gut reaction here was to play the first of these, 27. … Bxa4. The argument is simple: The worst piece in Black’s position is the bishop, so why not trade it off? I may not get another chance.
The trouble is that White doesn’t have to recapture right away. Instead he throws in the intermezzo 28. Rxb7+! Nxb7 29. Bxa4. The trade has put Black’s knight in a bad position and allows White to take control of the only open file after, say, 29. … Nd6 30. Rb1.
In fact we have a position here where two general principles come into conflict. One is that Black should trade off his bad bishop while he has the chance. The other is that Black should fight for control of the open file. But I can’t do both! When two principles conflict, how do you tell which one is more important?
I’m not sure if I can answer this question in general, but I’ll tell you how I resolved it in this case. If I want to control the b-file, 27. … R8b8 is definitely the move to play. If White were to trade both pairs of rooks, Black would just be winning, because I can eventually round up the h6 pawn. If White leaves the b-file unchallenged, then I’ll get a rook to the seventh rank and surely the game will win itself, right? For maybe a minute or so I thought this was just an easy win, and I was already starting to compose a blog entry in my head about the value of open files.
But is it so easy? Suddenly I had a ghastly thought: what if White trades off only one pair of rooks and then moves his knight to c3? After 27. … R8b8 28. Rxb7 Rxb7 29. Nc3! Rb2 White simply plays 30. Kd2!
Position after 30. Kd2 (analysis). Black to move.
FEN: 8/4kp1p/2bnp1pP/p2p4/2pP4/P1N1P3/1rBK1PP1/7R b – – 0 30
My rook is on the seventh rank but it has nowhere to go, and it’s going to be evicted next move with Kc1. This is a very important type of position to recognize: a position where I have “maxed out” and can’t make further progress. Well, I could try sacrificing the exchange with 30. … a4 31. Kc1 Rb3, which is really interesting — but let’s face it, that’s a risky way to play for a win considering that just a couple moves ago I seemed to have a very safe advantage. There must be a better way.
No matter how long I looked at it, I just couldn’t find a winning plan in the above position. If I can’t do anything with my rook on the b-file, then how can I make progress? Ordinarily the thing to do would be to try to create a second weakness, by pushing pawns on the kingside. But that is fraught with risk. The g-pawn can’t move, and even the f-pawn has to move with caution because Black has to be aware of the possibility (in some positions) for White to sac on g6 and queen his h-pawn.
That h6 pawn is, in fact, a very interesting feature of the position. On the one hand, it’s a weakness that needs constant tending; it’s why White doesn’t want to trade off his last rook. On the other hand, if the h7-pawn is removed the h6-pawn turns from a weakness into a game-winning advantage. I really don’t want to open new files that might allow White’s rook to penetrate and threaten h7.
So at this point I decided to back up and take a closer look at the first line. After 27. … Bxa4! 28. Rxb7+ Nxb7 29. Bxa4 instead of moving the knight, I can protect it with my king: 29. … Kd6! 30. Rb1 Kc7! 31. Rb5 Ra8. (Diagram)
Position after 31. … Ra8. White to move.
FEN: r7/1nk2p1p/4p1pP/pR1p4/B1pP4/P3P3/4KPP1/8 w – – 0 32
At this point I realized I had been guilty of some false assumptions:
Assumption 1: The knight is badly placed on b7. False! It defends the a5-pawn, and the c5-square as well, and it cuts off White’s access to the back rank.
Assumption 2: It’s the end of the world if White controls the b-file. False! Just like the previous position we saw, this is a position where White has “maxed out.” He can’t improve the position of his pieces any more. All I have to do is lift my rook to a6 and b6 and trade them off. Then the endgame should be a win! Not only that, if White realizes the danger and runs away with his rook, then I will control the b-file and it will be much better than before. Now that White has only the bishop left to defend with, not the bishop and knight together, Black’s rook (and knight) should be able to do some damage.
In fact, Shredder obediently let me follow the plan: 32. Kd2 Ra6 33. f4 Rb6 34. Rxb6 Kxb6 and now I “should be winning.”
My goal now is to corral the h6 pawn. But that’s not so easy! I have to make sure that White doesn’t get a chance to penetrate with his bishop to e8 or d7, and I also have to watch out for the aforementioned bishop sacrifice on g6, if I ever push my f-pawn. So I had to continue to “retreat to victory,” bringing my king back to d8.
The game continued 35. g4 Nd6 36. Kc2 Kc7 37. g5 Kd8! (Diagram)
Position after 37. … Kd8. White to move.
FEN: 3k4/5p1p/3np1pP/p2p2P1/B1pP1P2/P3P3/2K5/8 w – – 0 38
It’s just amazing to me how many of the moves in this lengthy maneuver look like “passive” or “defensive” moves. Moving my knight to b7. Moving my king to c7 to defend the knight. Moving my rook to a8. Bringing my king back to d8. Of course, one reason I have the luxury to play this way is the fact that the position is totally closed. White’s king and bishop can’t generate even a shred of counterplay. It’s fascinating how the two pawns at c4 and a5 keep White’s king locked out forever. (In fact, White’s pawn at a3 is part of the conspiracy; put it at a2 and White’s king could penetrate. So the a-pawn is what Mike Splane calls a “traitor pawn.”)
Here Shredder played 38. Kd2 f6 39. gf (otherwise I will trade and win the g5-pawn) 39. … Ne4+ 40. Ke1 Nxf6 41. Bc2 Ke7 42. Kd2 Kf7 (Being very careful not to allow the bishop sac) 43. Ba4 Ng8 44. Bc6 Ke7! (Patience, patience, patience! We don’t want to allow e4, breaking up my pawn formation). 45. e4 Kd6 46. Ba4 Nxh6.
Finally, my patience is rewarded! From this point the win is very easy, with two extra pawns, so I’ll stop here.
Non-chess players sometimes ask, “How many moves ahead can you see?” It’s a hard question to answer. In this case you can argue that I saw 20 moves ahead, because the decision to play 27. … Bxa4 was based on the plan that was finally completed with 46. … Nxh6.
Anyway, the take-home lesson from this game, for me, was to examine your assumptions. A bad bishop doesn’t necessarily have to be traded. A good bishop isn’t necessarily effective. (Take a look at the last diagram! White’s “good bishop” can’t do anything. In fact it turns out to be worse than useless because it can’t defend his pawns on dark squares.) A rook on the seventh rank isn’t necessarily a winning advantage. A rook on an open file can’t do anything if all of the entry squares are defended. “Retreating” moves can be winning moves. Basically, everything you thought you knew about chess is (sometimes) wrong!
A second lesson is to learn to look for and recognize “maxed out” positions. And finally… be patient. If you want to defeat a master-level player, whether it’s a human or a computer, you can’t rush.
What do you think? Was my judgement of the position after 27. … R8b8 28. Rxb7 Rxb7 29. Nc3 too pessimistic? At any point in this 20-move “combination,” did Shredder have a better defense that might have drawn? Let me know in the comments!
{ 7 comments… read them below or add one }
In diagram 2, I too couldn’t find a clear winning plan for Black if White is allowed to play f4 on his next move. But Black can try 30… Ne4+ 31. Kc1 Nf2 32. Rf1 Nd3+ 33. Bd3 Rg2 34. Bc2 Rh2. The h6 pawn falls and Black will have 4 pawns, 3 of them passers, for a knight. I like Black’s chances. My biggest worry would be that White might get enough counter-play if he can somehow win Black’s bishop. which s a bit short of squares.
In diagram 3, What is stopping White from playing his king up to f6, or better yet to g7? . Won’t you have to play f7-f6 and weaken your kingside? Then the bishop goes back to c2 and your h7-g6 pawn complex is very weak. Your plan of Ra8-a6-b6 seems to be too slow. White’s king doesn’t have to stay near the c-file to prevent the advance of the c4 pawn, his bishop handles that job, so it looks like Shredders choice of Kd2 is a bad one
I think that Rhb8 is the principled move, taking control of the b-file.
In re “traitor pawns,” I’ve always enjoyed inventing my own terminology to describe chess ideas.
I think this is a great comment. In your first line, I think you noticed something I missed, which is that 30. … Ne4+ 31. Nxe4? is met by the crushing 31. … Ba4!, winning the pinned bishop. So 31. Kc1 is forced. However, you may have missed something a bit later. After 33. … Rg2 White doesn’t have to play 34. Bc2. He can play 34. Be2 Rxg3 35. Rf3! and now Black doesn’t get 4 pawns for the knight, he only gets 3. The resulting position is very complicated and Black runs some risk of losing.
I agree with you that in diagram 3, if White plays 32. Kf3 instead of 32. Kd2, then 32. … Ra6 is too slow. I think Black has to change plans and play 32. … Nd6 33. Rc5+ Kd8. If 34. Kf4 Ra7! White’s king does not dare come to either e5 or g5. Meanwhile, Black still has the plan … f6, … Ng7, … Nxh6 in reserve. So I think there are still winning chances, although the win can’t be proven with the same kind of mathematical precision as in the line that occurred in the game.
The bottom line is that I’m still not sure which is really better, 27. … R8b8 or 27. … Bxa4. I agree with you that 27. … R8b8 is more principled. 27. … Bxa4 is a lot more subtle. As Mary Kuhner said, I’m trying to lure White into the rabbit hole. The danger of this sort of variation is that if I miscalculate a little bit, then I’m just luring White into a position where his pieces are tremendously active. In chess as in boxing, the “rope-a-dope” strategy is very difficult to pull off successfully.
In your first paragraph you are suggesting 34. Be2 Rxg3 35. Rf3! But in my first comment I was suggesting 34. Bc2 Rh2 going after the h-pawn, not Rg3 going after the e-pawn. Also, I don’t see anything to capture on g3, maybe I have the position wrong,
The other problem with 34. Be2 is it overloads the knight (I think the White king is on c1., right?), so Black gets the chance to play 34 … Ba4 and activate his bishop. Both of White’s minor pieces are tied down. If 35 Bf3 Rc2+ wins the knight and if 35 Na4 Re2 36. Rf3 f5 seems like it is winning; the h6-pawn and the a3-pawn can’t both be guarded.
BTW, I really enjoyed your patient strategic approach to this ending. Very impressive.
My bad! When I set up the position to go over your comment I mistakenly put the pawn on g3, not g2. So my “correction,” 34. Be2, has no purpose.
As for the second line you suggested, I took a deeper look at 32. Kf3 Nd6 33. Rc5+ Kd8 34. Kf4 and eventually realized that Black doesn’t even need to play 34. … Ra7. I can simply play 34. … Ne4 forking the rook and f-pawn. It seems incredibly risky, I know, to play this way when White’s king is threatening to march to f6/g7, but after 35. Rc6 Nxf2 the knight comes back to e4 in time to shut the king out. Winning that f-pawn is a big deal because the knight can’t be driven away from e4. It looked pretty good to me. Another possibility is to wait a move, play 34. … Ra7 first and then follow up with … Ne4. (That was my first idea, until I started asking myself, “Why am I even playing … Ra7?”)
So it looks as if either way, 27. … R8b8 or 27. … Bxa4, is probably winning.
As for your last comment, that’s why this was such a good training game! Historically I don’t play a lot of patient, maneuvering, strategic chess, and so it’s great to have a practice game where I was able to pull it off.
” In fact we have a position here where two general principles come into conflict. One is that Black should trade off his bad bishop while he has the chance. The other is that Black should fight for control of the open file. But I can’t do both! When two principles conflict, how do you tell which one is more important? ”
I think this is a great question too. My own rule-of-thumb in complex endings where I have a lasting advantage is stop counter-play first. Because Bxa4 cedes control of the b-file I would be reluctant to play it when I have another strong alternative. Shereshevsky, in his book “Endgame Strategy”, has an entire chapter devoted to the importance of stopping counter-play first, which he calls “Do not hurry.”
On the other hand, I think you mentioned something at one of my chess parties about the principle of taking advantage of opportunities that are only going to be there for one move only.
One of the big contrasts in our playing styles is in my games I tend to place too much trust in a patient approach, and in your games you tend to place too much trust in forcing moves. So maybe the answer to your question is to not worry too much and just play the move that “feels right” to you.
I also like Rob Radford’s comment about the rooks on the b-file are active simply by controlling the file, that penetration further down the file is not really necessary to add value to that piece.
The principle of two weaknesses says the superior side will need to create a second weakness in order to win the game, a single weakness can be defended. The h6 and the h7-g6-f7 pawn complexes are both weaknesses. The b-file is a potential weakness for both sides. It’s hard to see how either player can give away control of that file and hope to win. There doesn’t seem to be an easy way to create a second weakness once your opponent has control of the file.
As the game played out, White did have two weaknesses. He had to protect the c-file to prevent the c4 pawn from queening, and he had to guard the fragile f4-g5-h6 pawn structure. White also had no counter-play, as long as Black was patient. So the principles worked, even if some of the moves that were actually played were suspect.
In last year’s Washington Open I triumphantly played my queen to f6 against my opponent’s kingside castled position, only to find that (in a highly clogged position) she did nothing there and was time-consuming to extricate. My opponent said, “You went down the rabbit hole.”
In this year’s Washington Open I put a queen on f6 again, immediately thought “rabbit hole!” and retreated her the next move. It was true, too. So I agree, nomenclature is great. For those of us who are verbal thinkers it’s next to essential.
(Maybe next year I can avoid putting my queen on f6 at all? 🙂
It might be that Black is just winning in the initial position and that both Bxa4 and doubling the rooks on the b file are good. One hunch is that after 27…R8b8 28. Rxb7 Rxb7 29Nc3 Black is not obliged to play his R to the 7th (2nd rank), with the idea that threats are strong too. The R on the open file is not doing nothing. White needs to pay it heed. So then, can Black play f5 with ideas like Kf6 showing that the h pawn is overextended? I too like your variation using quiet patience but only after seeing your continuation. I would have assumed Black was not actually making progress.