A Fascinating Experiment

In my last post I complained about the futility of playing against computers. I got some excellent comments, one of which inspired me to try a new experiment. I decided to play a 10-minute game against Shredder where I get one “time-out” per game. The idea is that I would play until I reached an interesting position that demanded an in-depth analysis, and then I would take as much time as I needed to get it right.

My first try with this idea was, I think, an outstanding success. I took a time-out at exactly the right moment. Here is the position:

Position after 18. … f6. White to move.

FEN: r4rk1/ppn3pp/2pbpp2/8/qPPPQ3/P4N2/1B3PPP/3RR1K1 w – – 0 19

I’m playing White here. What would you do? Take your time. I spent about three hours (!) analyzing this position, and I ended up making the move that was originally tenth (!!) on my list of candidate moves.

White is at a crossroads, where he could play for immediate tactics or continue maneuvering and improving his position. My initial list of candidate moves (written down right after I took the time-out) shows my preference for immediate action:

• 19. d5 (top choice)
• 19. c5
• 19. b5
• 19. Qd3 (the only “waiting move” on my list)

As I analyzed the position further, my list of candidate moves grew to include

• 19. Ne5
• 19. Qg4
• 19. Qh4
• 19. Rd3
• 19. Rd2

And finally the tenth move I wrote down, after more than an hour of analysis, was:

• 19. Bc1!

How do you make a choice among so many possibilities? Well, first let’s look at the forcing moves. All of them have drawbacks. 19. d5!? has the idea of 19. … ed 20. cd cd 21. Rxd5! Rae8 22. Qc4 when the discovered check threat is very strong. This would be great for me; I’ve used tactics to solve some of my positional problems, the hemmed-in bishop and the pawn on d4 that didn’t seem to be going anywhere. Unfortunately, after 19. … ed 20. cd Black has the much better move 20. … Rae8! right away. This messes up White’s timing, as 21. Qc4 Rxe1+ 22. Rxe1 deflects White’s rook from d1 and allows 22. … cd. Bummer!

Next, 19. b5!? looks very attractive if Black plays 19. … cb? 20. Qxb7! Unfortunately, White is hanging the c-pawn, and after 20. … Qxc4 21. bc bc (21. … Qxc6 is also good) 22. Rc1 Qd5 White does not really have compensation.

Next, 19. c5 has some positive points. It forces the bishop back to an awkward square, 19. … Be7, and after 20. Qd3 it looks as if White has a positional bind. Black’s queen seems to be shut out of the game, and White has the plan of Bc1 and Bf4. (By the way, this was when Bc1 first appeared on my mental radar screen.) The trouble, though, is that Black’s queen is not as “shut out” as she appears. Black plays 20. … a5! and after 21. Qc4 ab 22. ab Qa2! Black’s pieces are starting to become more active than White’s.

Finally, the wild and crazy piece sacrifice 19. Ne5!? did not work, according to my analysis, because of 19. … fe 20. de and now Black plays the counter-sacrifice 20. … Bxb4! 21. ab Qxb4. The threat against the loose bishop on b2 is annoying, because it slows White’s attack down by a crucial tempo. White can’t afford to ignore the attack either, because 22. Rd7?? would run into 22. … Qxb2 and Black threatens mate in three starting with 23. … Qxf2+.

What we see from all these tactical variations is that White just isn’t ready for tactics yet. There are too many weaknesses in his position. The pawn on c4. The bishop on b2. The rook on d1, which is constantly being tickled by Black’s queen. The back rank. Even f2 in the last variation. So I finally, grudgingly, admitted that the position needs to be taken more slowly. Of course, if you’re playing against a computer this is not a big surprise — it has probably seen through all the tactics already (even though I set it at a 2165 strength).

Now if we’re going to take it more slowly, what should we be thinking about? In my earlier post Simple Chess (from 2013) I gave a list of questions that Mike Splane and Gjon Feinstein came up with. So I started going through the list one at a time.

• What are the pawn breaks? Well, that’s easy. b5 and d5 are the breaks, and we’ve already seen that they don’t work. However, White should still keep them in mind.
• What are the best and worst-placed pieces for me and my opponent?  Again, easy! For me, it’s the bishop on b2. It is the only piece that is not doing anything and has no prospects. Everything else is pretty well placed, and in fact what annoyed me about having to consider moves like Qd3 and Rd2 and Rc1 and so forth is that I’m moving pieces that are already well-developed.
• What are the targets? Easy! For me the number one target is the pawn on e6. I still want to find some way of undermining that pawn. Unfortunately, my own position is riddled with targets, as we’ve seen: a3, b4, c4, b2, d1 are all targets in various ways. This is why I have to be a little bit careful.
• What are the possible trades, and which ones are good for me? Well, there aren’t too many possible trades at the moment, but I wouldn’t mind trading my ineffective bishop for his bishop, or even better for his knight, which is the only thing holding the e6 pawn.

At this point I fast-forwarded through questions 5, 6, 7, and 8 because it was already becoming obvious what I should do. The move 19. Bc1, first of all, takes away one of the targets. As we’ve seen, several tactical lines for White didn’t work because Black was able to gain a tempo by attacking the bishop. Second, it improves my worst-placed piece. At the very least, it can go to f4 and trade itself for one of Black’s minor pieces. We’ll see that it has an even more interesting destination in the game. And finally, 19. Bc1 doesn’t spoil the already good placement of my other pieces. Moves like 20. c5 or 20. Ne5 are still on the table, and in fact 20. c5 has become a more serious threat because it can be followed up (after … Be7) by Bf4.

So I played 19. Bc1. Probably Black’s safest response is 19. … Rfe8, after which 20. Bf4 or 20. c5 Bf8 would be about equal. But Shredder surprised me with a move I thought was bad: 19. … Qb3?

“Hot diggity dog!” I thought, and with scarcely a moment’s thought I threw down the gauntlet: 20. Ne5!?

There’s no question that this sacrifice is better now than it would have been on move 19. The one caution flag is that I’m playing against a computer, and it’s doubtful that I can out-calculate it. Against a human, 20. Ne5 would probably get an “exclam” without the question mark. But objectively, Black does have more resources than I thought, and it’s possible I should have played the simpler 20. c5 followed by 21. Bf4.

The next few moves were easy: 20. … fe 21. de Be7 22. Rd7 Bd8 (diagram)

Position after 22. … Bd8. White to move.

FEN: r2b1rk1/ppnR2pp/2p1p3/4P3/1PP1Q3/Pq6/5PPP/2B1R1K1 w – – 0 23

Decision time again! Actually, I wish I could have taken another time-out here, but the rules were only one per game. White’s candidate moves are:

• 23. Bh6
• 23. Red1
• 23. Qg4

Which of these moves would you play?

My intention had been to play 23. Red1, but I forgot my analysis and panicked after I saw 23. … Nd5? In fact, this is nothing to be afraid of because I can play 24. Qg4 and take the knight after 24. … Rf7 25. Rxf7 Kxf7 26. cd. Rybka says, however, that Black can defend successfully with 23. Red1 Kh8! preparing … Rg8. This is frankly a little bit difficult to believe, and I would be happy to play this position against a human. But maybe not against a computer.

Rybka’s top choice is 23. Qg4, the main point being that if Black ever plays … Ne8 (his best way of defending g8) he will lose the e6-pawn. However, I think that the move I played was not worse and, if you believe Rybka’s more in-depth analysis a few moves later, actually better. The move I chose was the most forcing one:

23. Bh6! …

Remember how I wrote earlier about a “more interesting destination” for the bishop? Here it is! The previously befuddled bishop on b2 has now become an Amazing Apostle of Archbishoply Awesomeness on h6! Black’s reply is forced, since 23. … gh?? 24. Qxh7+ would be mate.

23. … Rf7 24. Rxf7 Kxf7 25. Qxh7 Ne8 (diagram)

Position after 25. … Ne8. White to move.

FEN: r2bn3/pp3kpQ/2p1p2B/4P3/1PP5/Pq6/5PPP/4R1K1 w – – 0 26

Here is when I finally went wrong. Of course I would like to lift my rook to e3 and then to f3, but there’s that annoying matter of the back rank that needs to be taken care of first. According to Rybka, I should just play 26. h3. White has two pawns for the piece but it is very difficult for Black to untangle his pieces in time to deal with the Re3-f3 threat. In fact, Rybka says that Black’s best defense is to give back a piece on f6 at an appropriate time and enter a pawn-down endgame with about a +0.6-pawn advantage for White. Whatever that means.

Instead, the move I played was 26. h4? which was a mistake simply because you shouldn’t give pawns away when you don’t have to. Shredder correctly captured: 26. … Bxh4! 27. Bxg7 Nxg7 28. Qxh4. I willingly went into this because I didn’t think Black’s king could survive, but I way overestimated my attack. After 28. … Qxa3 (another cold-blooded computer move) 29. Qf6+ Kg8 White’s rook just doesn’t get into the attack fast enough. The game ended 30. Rf3 Qc1+ 31. Kh2 Rf8 32. Qg6 Qxc4 33. Rg3 Qh4+ 34. Rh3 Qf4+ 35. Rg3 Qxe5 White resigns.

In spite of the fact that I ended up losing, I thought this game was a lot of fun and infinitely more instructive than my usual blitz games. I loved the move 19. Bc1, which I never would have even considered in a blitz game and might not even have gotten around to analyzing in a tournament game. I also loved 20. Ne5, even though it may not have been objectively best, and I think it would have given me really good chances to win against a human opponent.

All in all, I think I will probably try this again some time!