White – Zillions of Games
Black – Zillions of Games

1. d4 d5
2. c4

Since the Zillions of Games implementation of Piece Promotion Chess has no opening book, I entered these moves myself. Now the computer takes over, playing both sides of the board at 10 seconds per move with maximum move variety. These settings, on the 1.8 GHz Pentium 4 I was then using, tend to produce wild and spectacular albeit not terribly accurate play, as we shall see.

2. ... dxc4
3. Bf4 Nc6
4. e4 Qxd4
5. Bxc7=B'

It is unusual to have a promoted piece permanently on the board this early. In most PPC games, 15-30 moves go by before a piece promotes without immediately being captured. This lends PPC a “feel” of starting out as essentially a chess game but eventually diverging. This feeling is incorrect insofar as the strategy, and even the values of the pieces, are significantly different than in chess from the opening move. On the other hand, it is not entirely an illusion. The things that are important in chess—king safety, material, development, center control, attack, defense, initiative—are also important in PPC, just not in the same proportions. What would be a plausible idea early in one game is usually also plausible early in the other. What works is another matter.

5. ... Qxb2
6. Nd2 Bg4
7. f3 Nd4
8. Rc1 c3
9. Nc4 c2

10. Nxb2 cxd1=N'+

As you can see, a quiet, uneventful start to the game—no underpromotions until the tenth move! Seriously, this game manifests the typical wildness of PPC play in accelerated, exaggerated form.

It should be noted that this underpromotion is in fact mildly motivated, since it leaves no alternative but to capture the Kniking. On the other hand, cxd1=Q+ would allow Kf2—not that there's any good reason White would want to play that, but why give your opponent an extra choice?

11. Nxd1 Be6?

With low search depth, perhaps the computer was hoping to promote on a2, but Black's King is too exposed. Bd7 will have to be played anyway and should have been played immediately rather than this self-blockading, time-wasting move.

12. B'xb7 Rd8
13. Rc7=R' Bd7

Calmly proceeding with development will not work for Black—e.g., Nf6 14. B'xa7 b6 15. B'b6 threatening the Rook and if 15. ... Ra8 16. B'xd4 or if 15. ... Rd6 16. R'b8+ leads to mate—e.g., Kd7 17. R'd8#.

14. Ba6=B'

A quite unusual PPC position. First of all, after only one piece exchange there are already three promoted pieces on the board. Secondly, White has three promoted pieces, Black has none, Black's kingside is completely undeveloped, and yet—White does not have an overwhelming position. Certainly White should win, but White's play requires great care. Promotion is only a temporary advantage in open positions, since your opponent's pieces tend to promote too—and Black's position has a surprising amount of fight in it.

14. ... e5
15. B'd5?

This was my dubious “contribution.” The original game proceeded with White playing B'a5!, combining attack and defense, and eventually winning after 15. ... Ne6 16. R'c3. When I went back over to analyze, switching Zillions to the slightly stronger 15 second level with minimum variety, I was puzzled by White's inability to wipe Black out quickly. I thought, “Why couldn't White create a situation where, if White attacks the Rook and Black defends it with the Knight, White can just take the Knight?” The game you are seeing is the result (or one of the results) of this “analysis” session—though Zillions played all the remaining moves.

15. ... Bb4+

D'oh! With stunning swiftness, Black is right back in the game. My blunder produced results so interesting that I allowed it to stand.

16. Kf2

White must concede a promotion (at least). If White interposes on c3, Nc2=N'+ wins a piece.

16. ... Bd2=B'!?

Ne6 may have been objectively better, although it leaves White with the clear upper hand in a very complicated position.

17. Ne3 Ne7
18. Nc4 Bb5
19. R'xe7+?

Instead, 19. B'xb5+ Nxb5 20. R'xe7+ Kxe7 21. B'c5+ wins a piece. The text move also wins a piece, but under much less favorable circumstances. The error is hard to assess—but fatal.

19. ... Kxe7
20. B'xb5 Rxd5
21. exd5 Nxb5
22. Nxd2

At this stage I shifted Zillions into one minute mode for the rest of the game to produce more accurate play. White is now a piece up, in an innocuous-looking position. Black has only two pieces, only one of which is developed. Yet surprisingly enough, White cannot hold the extra piece. Worse yet, he is caught in a mating net.

22. ... Nc3=N'
23. d6+ Kd7

Of course, not Kxd6 24. Ne4+, trading away the dangerous Kniking, when White's extra piece should win easily.

24. Ne4 N'd3+

Already, a fundamental problem starts to assert itself. Once one's King is in check from a Kniking, it is very hard to get away from it.

25. Kg3 N'f4+
26. Kf2 Rc8

Like its ecclesiastical comrade two doors down, Black's kingside Rook makes its first move from home with devastating effectiveness. White cannot stop it from promoting. Black now threatens Rc2=R'+ with N'xg2# to follow. If 27. Nd2 Rc3=R' 28. Kd1 N'd3+ 29. Kd1 and either piece takes the knight with mate; or 29. Kf1 R'c1#; or 28. Nb3 (or to other squares) R'c2+ leads to mate.

27. Ne2 N'd3+

28. Kg3

White must give up his extra piece to avoid immediate destruction. If Kf1 then 28. ... Rc2=R', threatening two mates in one, N'xe2 and R'd1, and it's curtains. For example, 29. g4 N'e3+ and mate next move; or 29. N4c3 N'e3+ 30. Kg1 R'd1+! Nxd1 N'xe2#, or 30 Ke1 N'xg2#.

28. ... N'xe2+
29. Kh4 N'e3
30. Rb1 g5+ (?)

This move is sufficient to win, but it misses a complex forced mate. The text move allows White to get back into the game with some counterplay.

31. Kxg5 N'f4+
32. Kh4 N'xg2+
33. Kh5 N'f4+
34. Kh4 N'xf3+
35. Kh5 N'f4+

This move does win, but the failure to take the Knight suggests that Zillions overrated White's coming counterplay. More incisive is N'xe4! 36. Rb7=R'+, and now certainly not ... Kd8?? 37. R'e7#, nor even Kxd6, but Ke6! and if 37. Rxc8 Kf5! threatening N'g5#. If now 38. R'g8 N'f6+, or 38. Kh6 N'f6 with mate to follow, or 38. Kh4 N'f3+ 39. Kh3 N'f2+ 40. Kh4 N'g4#. On the other hand, if 37. R'e7+ Kd5 and now if 38. d7 N'f4+ 39. Kh4 (Kh6 N'g6#) N'f5+ and mate in a few moves.

36. Kh4 N'g6+
37. Kh3 Rc2=R'

Threatening 38. ... N'f4+ 39. Kh4 R'xh2#. White is forced to deliver checks.

38. Rb7=R'+ Ke6
39. R'e7+ Kf5!

Cold-blooded! The computer, being a computer, foresees the next few moves without fear and goes for the jugular. Most human players, I suspect, would simply take the Rooking and get out of these wild complications with a sigh of relief: 39. ... N'xe7 40. dxe7 Kxe7 and Black should win the resulting endgame pretty easily, although he still must be careful after 41. Ng5.

40. Ng3+ Kf4

Now Black no longer threatens mate in two with N'f4+. Instead, he threatens mate in one with N'g5—so White still has to check! Unfortunately, White can no longer do this without surrendering a piece.

41. Nh5+ N'xh5
42. R'xf7+ Ke4
43. R'f1

White sidesteps the threatened Kniking fork on g5, but Black now mates by force.

43. ... N'g5+
44. Kg3 R'd3+
45. Kg2

If Kf2, N'f3#; or R'f3+, R'xf3#.

45. ... N'h3+
46. Kh1

As seen on the PPC home page!

46. ... R'd1!
47. R'xd1 N'f2#

This one-piece mate is a fitting cap to a game Buffalo Bill would have been proud to play.

Email me at semipatz '@' radicalroot.com