How to beat the Fritz chess program

1.e2-e4 c7-c5 2.f2-f4 winkensmile:In all fairness, CHESSBOT Fritz originally wanted to play 2. ...e6 here, but I wanted to explore the Queen Sac line here, known as The Grand Prix Attack, and Fritz forced the move 2. ...d5.

Fritz adds: [3.Bf1-b5+ Bc8-d7 4.Bb5xd7+ Qd8xd7 5.d2-d3 e7-e6 (5...Nb8-c6 6.e4xd5 Qd7xd5 7.Ng1-f3 e7-e6 8.Nb1-c3 Qd5-d7 9.0-0 Ng8-h6 10.Nf3-e5 Nc6xe5 11.f4xe5 Nh6-f5 12.Bc1-f4 Bf8-e7 13.Nc3-e4 0-0 14.Kg1-h1 Qd7-c6 15.Qd1-g4 c5-c4 16.Rf1-f3 c4xd3 17.c2xd3 Qc6-b5 18.Ra1-f1 Ra8-d8 19.Qg4-h3 Rd8-d4 20.Bf4-d2 Salaun,Y (2386)-Karr,J (2376)/France 2003/CBM 094/0-1 (33)) 6.Ng1-f3 Nb8-c6 7.0-0 Ng8-e7 8.Nb1-a3 g7-g6 9.Bc1-e3 Bf8-g7 10.Be3xc5 Bg7xb2 11.Ra1-b1 Bb2-g7 12.Na3-b5 0-0 13.e4-e5 b7-b6 14.Bc5xe7 Nc6xe7 15.Qd1-e2 Ne7-c6 16.c2-c3 Rf8-c8 17.h2-h4 Bg7-f8 Larsen,B (2515)-Zhu Chen (2490)/Bad Homburg 1998/CBM 066/1-0 d7-d5 winkensmile:I had just battled CHESSBOT Fritz to a draw in a Sicilian, Najdorf, Poisoned Pawn variation, and during that opening, I was doing some research into the various lines of the Sicilian, in efforts to find a more interesting line to play against Fritz. Research of Opening Book, and databases are permitted in correspondence chess, and CHESSBOT Fritz was aware that I was researching the opening lines.

We were well into this variation, when I happened to look at the March 2007 issue of Chess Life, and noticed Dana Mackenzie's article where he defeated both Fritz 9 (3161) and Crafty 19.19 (3106), as well as a respectable human, IM David Puress (2461), with this Grand Prix Attack version of the Sicilian, in which he sac's the Queen on move 6!, and asked Fritz if he would be interested in looking into this gambit line with me after we finished this current game, and he graciously agreed. I have neither Fritz 9, nor Crafty 19.19; so I wondered how a grouchy old wood pusher like me, winkensmile (2163), might fare against a powerful silicone monster like CHESSBOT Fritz (2224+) with a gambit line that has me throwing away my Queen on move 6!?

I had first tried this line on the local club TD's laptop (the night before), which had Fritz 9 loaded on it, in a 40/90 tourney setting, but after I went in for the immediate gain of the exchange with 10. Nf7 and 11. Nxh8, I went on to lose; so I did a little more research, and thought I would try another game with a better understanding on how this opening works, with another Fritz engine, and with little more time and research behind me.
3.Ng1-f3 DayWalker:Here why not: 3.exd5 Qxd5 , 4.Nc3 winkensmile:3. exd5 Qxd5 4. Nc3 is a playable book line of the Grand Prix Attack, and creates a position similar to the Counter-Center, but Black gets a good game after 4. ...Qd8, 5. Nf3 Nf6 6. Ne5 e6 7. Qf3 Be7 8. b3 Nfd7 9. Bb5 O-O, but I wanted to get Fritz into this other variation, and getting a bot out of book early is a good thing, especially where you have done some opening research! ;) d5xe4 4.Nf3-g5 Ng8-f6 5.Bf1-c4 winkensmile:Fritz adds: [5...e6 6.Nc3 a6 7.a4 Nc6 8.Ngxe4 Nxe4 9.Nxe4 Be7 10.d3 0-0 11.0-0 b6 12.Bd2 Bb7 13.Qh5 Nd4 14.c3 g6 15.Qg4 h5 16.Qd1 b5 17.Ba2 Nf5 18.Qe2 Qd7 19.Re1 bxa4 20.Rea1 Hector,J (2420)-Yrjola,J (2415)/Gausdal 1987/CBM 004/1-0 Bc8-g4 winkensmile:Fritz adds: [6.Be2 Bxe2 7.Qxe2 Nc6 8.Nxe4 Nd4 9.Nxf6+ gxf6µ] 6.Qd1xg4 Nf6xg4 7.Bc4xf7+ Ke8-d7 8.Bf7-e6+ winkensmile:8. ...Ke8 forces a draw by repitition after 9. Bf7+. Kd7-c6 winkensmile:In Dana Mackenzie's encounter, Fritz 9 played the inferior 8. ...Kc7; Fritz adds: [8...Kc7 9.Bxg4 Qe8 10.Ne6+ Kb6 11.Na3µ] here, so CHESSBOT Fritz played the far stronger 8. ...Kc6; note that a human opponent (IM David Pruess, 2461, also played 8. ...Kc6, but went on to lose after 9. ...e6; also note that 9. ...e5? is encountered with 10. Nf7 Qh4+ 11. g3 Qxg4?? 12. Nxe5+! 9.Be6xg4 g7-g6 winkensmile:So, here, Fritz took the better alternative with 9. ...g6, opening diagonal lines to release his dark squared Bishop, and allowing White to play 10. Nf7+, forking Black’s Queen and Rook, winning at least the exchange for counter play, and this is where we went out of book, and I was on my own. I took several days to mull over the next move, almost to my 5 day time control, compiling several pages of hand written notes by a real board and pieces.
10.Ng5-e6 winkensmile:Perhaps a bot would have played 10. Nf7+, (which is a playable move, but NOT to win the exchange, but to redeploy it to another square, like e5 later on), but after some thought, I realized that a far stronger move would be 10. Ne6!, where White’s Knight gets a powerful outpost, kicks the Black Queen, blocks the d7 pawn, and controls all the central dark squares; no point in putting such a powerful piece out of play, in the corner on h8, (even if he does win the Rook, which is out of play anyhow), and this move also prevents Black from playing his Bishop to g7 any time soon, the Queen doesn't have too many good squares to seek refuge, (perhaps 10. ...Qd7 is best). 10. Nc3 is also playable, and shows up in a few databases. (*) as does 10. Nf7 (**), but again, only to redeploy it elsewhere, and to secure squares for a Bishop attack in the center.

This is where humans can sometimes outwit an engine, who may underestimate the effectiveness and scope of a minor piece attack and where I can develop with tempo.

As Dana Mackenzie mentioned in his March 2007 issue of Chess Life, "the engine's weakness is that they have no sense of fear", and are completely confident in their calculations and algorithms, oblivious to the looming attack of White's Knights and Bishops working in concert to overwhelm Black's Queen and storm the King. The threats may scroll off their horizon, (known as the "horizon effect"), which often occurs in the various subtle endgame techniques, where pawn formation and tempo are absolutely critical. A well seasoned human player may see these things intuitively, but many bots and engines go for the material advantage without regard to these subtleties.


* 1. e4 c5 2. f4 d5 3. Nf3 dxe4 4. Ng5 Nf6 5. Bc4 Bg4 6. Bxf7+ Kd7 7. Qxg4+
Nxg4 8. Be6+ Kc6 9. Bxg4 g6 10. Nc3 Bg7 11. Nf7 Qd4 12. Nxh8 Bxh8 13. Be6
Na6 14. Bd5+ Kb6 15. Bxe4 Rd8 16. Ke2 Nb4 17. a3 Nxc2 18. Bxc2 Qxf4 19.
h3 Bd4 20. Ne4 Qe5 21. d3 Ka6 22. Rb1 b6 23. b4 cxb4 24. Rxb4 Kb7 25. a4
Qh5+ 26. g4 Qe5 27. Rb5 Qc7 28. Bb3 e6 29. Bg5 Rf8 30. Bxe6 Re8 31. Bd5+
Kb8 32. Kd1 a6 33. Rb4 Qe5 34. Bf7 Rc8 35. Rf1 Ka7 36. Bf4 Qh8 37. Bd5
Rf8 38. Ke2 Rd8 39. Bh6 Rxd5 40. Rf7+ Kb8 41. Rf8+ Qxf8 42. Bxf8 Kc7 43.
Bd6+ Kd7 44. Bf4 b5 45. Be3 Bxe3 46. Nf6+ Kc6 47. axb5+ Rxb5 48. Rxb5 Kxb5
49. Kxe3 h6 50. Ng8 a5 51. Kd2 Kc5 52. Nxh6 Kd4 53. Nf7 a4 54. h4 a3 55.
Kc2 a2 56. Kb2 Kxd3 57. h5 gxh5 58. gxh5 Ke4 59. h6 Kf5 60. h7 Kg6 61.
h8=Q Kxf7 62. Qh6 Ke7 63. Kxa2 Kd7 64. Kb3 Ke7 65. Kc4 Kf7 66. Kd5 Kg8
67. Ke5 Kf7 68. Qh7+ Ke8 69. Ke6 Kd8 70. Qd7+ 1-0

** 1. e4 c5 2. f4 d5 3. Nf3 dxe4 4. Ng5 Nf6 5. Bc4 Bg4 6. Bxf7+ Kd7 7. Qxg4+
Nxg4 8. Be6+ Kc6 9. Bxg4 g6 10. Nf7 Qd4 11. Nc3 Rg8 12. Ne5+ Kb6 13. b3
Bg7 14. Nc4+ Ka6 15. a4 Nc6 16. Be2 b6 17. Nd6+ Ka5 18. Nb7+ Kb4 19. Na2+
1-0
Qd8-a5 winkensmile:Fritz adds: [10...Qd7 11.Na3 Bg7 12.c3µ] 11.b2-b4 winkensmile:After Black's 10. ...Qa5, I took another day or so to find a way to forge on with the attack and to develop my pieces with tempo, and so, in keeping with the theme with this thread, I threw another pawn at him with 11. b4!

The point being that it kicks the Queen again, and allows me to mobilize my dark square Bishop, and removes Black's pawn on c5 so that I can play a timely Nd4+, also, seeing that Black's King is exposed, and on the Q-side, it allows for a possible semi-open attack by my Rook on the b file later on. Hey, I already sac'd my Queen, what's another pawn for the sake of the attack!?
c5xb4 winkensmile:Fritz adds: [11...Qxb4!? is an interesting idea 12.Nd8+ Kb6³], but on 11. ...Qxb4, I might consider 12. Ba3 or 12. Nc3. 12.Bc1-b2 winkensmile:After 11. ...cxb, 12. Bb2, I have some serious threats levied against Black, (aside from the obvious BxR) including Nd4+, Be5, and possibly weaving a mate net here! Fritz sees the dangers and offers up his Rook again with 12. ...h5, (allowing 13. Bxh8 hxg4 14. Nxf8), but then I would have my active Bishop on h8 and my killer Knight on f8; seeing that both Black's Rooks are already out of play (and may as well be off the board), and Black's Bishop on f8 is also useless, I stayed true to the theme of the position and played 13. Bh3, keeping my minor piece attack alive.
h7-h5 13.Bg4-h3 winkensmile:Fritz adds: [Inferior is 13.Bxh8 hxg4 14.a3 Qh5] Rh8-g8 14.a2-a3 winkensmile:Fritz's 13. ...Rg8 took it out of being en-pries but also threatened to get his Bishop mobilized with Bg7; however, I struck another gambit shot with 14. a3!
winkensmile:In Fritz's post mortem analysis, he gives this move 14. ...e4-e3?, with an alternative: [¹14...Bh6!? 15.axb4 Qxb4+-] e4-e3 winkensmile:Up until now, Fritz had assessed my position in the negative numbers, down about a pawn or two, but here, suddenly, it assessed me at a few points ahead! After a few tweaks to Fritz’s engine here, Fritz found the amazing 14. ...e3!! which caught me by surprise! I had not even considered this move! I took another few days to mull this one over. I could not capture with 15. dxe3?? which would allow both 15. ...bxa3+, (winning my dark square Bishop), or even just 15. ...b3+ giving me a cramped position.
15.a3xb4 Qa5xb4 16.Bb2-c3 e3xd2+ winkensmile:I played 15. axb4, attacking Black's Queen, and Fritz, played, 15. ...Qxb4, not sure if Fritz had anything better; (and here, I also mulled over 16. Ba3, but I figured I'd better stay on this diagonal with Black's pawn threatening exd2+), and then, he still gets 16. ...exd2+!

Again, I still can't capture with 16. Bxd2, (16. Kxd2 leads to 16. ...Qd6+), which allows Black to seize the initiative with 16. ...Qb2, and when you have an attack going, initiative is everything! I considered 16. Kf2, but my inner senses told me to stay close to that d2 pawn, and so, I played 16. Kd1, possibly mobilizing my other Knight with an eventual Nxd2, also, this move allows me to play Re1 when it's ripe; Fritz told me in our PM that he didn't expect this move; this is the beauty of a human vs. bot game, in that we are often surprised at our opponent's responses.
17.Ke1-d1 Qb4-b6 18.Ne6-d4+ winkensmile:After 16. ...Qb6, I thought again for a day or two, trying to find the best way to continue the attack, and came up with 17. Nd4+, wondering where Black will place his King? This move also threatened 18. Be6, getting my Bishops into the center.

There aren’t too many good squares left for Black's King and Queen; 17. ...Kd5 allows 18. Be6, attacking the Rook on g8, and brings the light square Bishop into the fray, with mate nets beginning to form, while 17. ...Kd6 allows 18. Nxd2, which threatens 19. Nc4+ forking the King and Queen, and White's Knights and Bishops mount a fierce attack to overwhelm Black's King and Queen, and the scent of checkmate is already in the air.

Again, take a look at this position, notice how White's pieces are fully deployed, and ready to spring into attack, while Black's minor pieces are totally useless, still on the back rank, and may as well be off the board! Amazing when you consider that White is playing against a bot, (with a 3100+ rating), and without his Queen!
winkensmile:Also, in this position, 18. ...Kc5 is met with 19. Nxd2 e5 22. fxe5 and now, Black is running out of squares; moves like 22. ... Qd8, Bg7, and Nc6 don't prevent mate after 23. Ne4+ Kc6-c7 winkensmile:Still amusing to note that ALL of Black's minor pieces are all on the back rank, and out of play, while White's pieces are about to unleash some fury! 19.Bc3-a5 winkensmile:Black finally acquiesces to the loss of its Queen with 17. ...Kc7, (which allows 18. Ba5, skewering Black's lady), and with 18. ...Bg7 for some counter play. At this point, Fritz was tweaked again, disabled book, with at least 60 minutes of analysis, with Deep Position analysis set, and maxed out on hash tables; the engine was also set to full bore defensive mode, hoping to salvage a draw out of this.
Bf8-g7 20.Ba5xb6+ Kc7xb6 21.Ra1-a4 Bg7xd4 22.Ra4xd4 Nb8-c6 23.Rd4xd2 winkensmile:After the following series of exchanges and after the dust settled, White is a full piece up, regaining the Queen with interest.
e7-e5 24.Nb1-c3 winkensmile:24. Nc3 finally develops my last piece and defends the f pawn (24. ...exf4, 25. Nd5+ followed by 26. Nxf4).
Ra8-d8 25.Bh3-d7 Nc6-b8 winkensmile:My 25. Bd7 nullified Black's pin with his Rook on d8, and threatened Black's Knight, but Black's 25. ...Nb8 took me by surprise again!

I didn't even consider this move in my analysis (again)!

26.Nc3-d5+ Kb6-c5 27.Nd5-f6 Rg8-g7 28.Rd2-d5+ winkensmile:I took another day or so mull it over, and found a way to continue the theme of the game by developing with tempo, and played 26. Nd5+ with 27. Nf6 to follow, which allowed a strong outpost, (again), for my Knight, attacking Black's Rook, which, after 27. ...Rg7, allowed 28. Rd5+ and now my Rook is also in on the attack!
Kc5-b6 29.Bd7-e6 winkensmile:At this point, I figured that it would be good to trade off pieces, and free up my Bishop, so I played 29. Be6, and 30. Rd2, keeping my Rooks close to my pawns.

Black has some serious counter play with his connected Q-side pawns, and I need to find a way to deal with this. I have been in similar positions before against masters in OTB games, being a full piece down, and managed to trade off Rooks, and pawns, and then sac my remaining piece for the last remaining pawn and it is not possible to mate with just a Bishop and a Knight, (nor either one alone)!

Even though I have material and spatial advantage, the game is FAR from being over. Pawn structure is very important in these endgames, and I must not fail to keep that in mind. One misstep here with Fritz, and he gets at least a draw! I also need to be aware of the position of my King so that I can defend the advancing Q-side pawns.

I need to double up on Black's pawns and win another pawn or two if I can. It's going to be a real white-knuckle finish!
Kb6-c7 30.Rd5-d2 Nb8-c6 31.f4xe5 Nc6xe5 32.Rh1-e1 Ne5-c6 33.Nf6-d5+ Kc7-b8 34.Nd5-f4 Rd8-f8 35.g2-g3 Rf8-e8 36.Re1-e2 winkensmile:Finally, I have my Rooks connected! a7-a5 37.Be6-d5 Re8-c8 38.Re2-e6 Nc6-b4 winkensmile:I played 38. Re6 double attacking Black's pawn on g6, but Fritz continues to surprise me with 38. ...Nb4! (38. ...Ne7 39. Be4), which attacks my Bishop and puts pressure on my pawn on c2. I'm not too sure I have the luxury of time to capture on g5 here, as Black's real threats are on the Q-side.

There are several possibilities here, and I have to deal with Black's plan of pushing a5-a4 followed by b7-b5, followed by Rg7-Rc7, also, Black has Rc3, with either Rgc7, or Ra3, and so, I decided on the 39. c2-c4 push, which moves the passed c pawn, frees my Rook to move to the Q files, protects my Bishop, prevents b7-b5, also, neutralizes the threats of any capture on c2, and any penetration with Black’s Rook on the c file. If Black captures 38. ...Nxd5, 39. cxd5, puts the d pawn on the fast track with d5-d6-d7 supported by both Rooks, and with threats of a timely Nd5, to Nb7.

39.c2-c4 h5-h4 40.g3xh4 winkensmile:Fritz adds: [Less advisable is 40.Rxg6 Rxg6 41.Nxg6 hxg3 42.hxg3 Re8+-; 40.Nxg6?! hxg3 41.hxg3 Rc5+-] g6-g5 41.h4xg5 Rg7xg5 winkensmile:But Fritz again finds some clever ways of breaking through! After 39. ...h4, 40. gxh4, (doubling up my h pawn), then 40. ...g5, 41. hxg5, Rxg6! Fritz threatens to drop his Rook to g1+, and take control of my home rank, plus reduced my pawns to only two isolated loners.
42.Re6-e7 winkensmile:Fritz adds: [42...Rg1+ does not help much 43.Ke2 Rc7 44.Rxc7 Kxc7 45.Be4+-] Nb4-c6 43.Re7-h7 Rc8-c7 winkensmile:Fritz's 43. ...Rc7 baited me to play 44. Rxc7, (hoping for 44. ...Kxc7, with 45. Ne6+ winning a Rook); but Fritz has 44. ...Rg1+. I had a gut feeling that Fritz had cooked up something even better, and was expecting me (even baiting me) to play 44. Rxc7, so I played 44. Rh8+ instead; besides, my Rook on the h file has more play than his Rook on the c file; why trade a better piece for a weaker one?

44.Rh7-h8+ Kb8-a7 winkensmile:Here, 45. Ne6 (forking both Black's Rooks) doesn't work because 45. ... Rg1+ 46. Kc2 Rg7, which could cause me some problems. 45.Rd2-a2 Rc7-d7 46.Kd1-e2 Rg5-e5+ 47.Ke2-d2 winkensmile:Fritz adds: [47.Kf3 seems even better 47...Kb6+-] wow! I didn't look at this move! Maybe this illustrates the human weakness, as I was trying to stay close to the Q-side while I still could, to deal with those Q-side pawns.

Fritz also continues with: [47...Re4 is no salvation 48.Nd3 Rd4 49.Ra3+-] Re5-e7 winkensmile:Black finally connects his Rooks, but not for long. 48.Kd2-c3 Re7-e3+ 49.Kc3-b2 Ka7-b6 50.Rh8-h6 winkensmile:[50...Re5 doesn't change anything anymore as White has 51.Rf6+-] Kb6-c7 51.h2-h4 Rd7-d6 52.Rh6xd6 winkensmile:Time to trade off a pair of Rooks. Kc7xd6 53.h4-h5 winkensmile:Fritz adds: [53...Nb4 cannot change destiny: 54.Rxa5 Ke5 55.Ng2 (or 55. Ng6, or Ne3), Re2+ 56.Kb3 Nxd5 57.Rxd5+ (57.cxd5?! Rxg2 58.Ra2 Rg5±) 57...Kf6 58.Rd6+ Ke7 59.Rg6+-] Re3-e7 54.Kb2-c3 Re7-e3+ 55.Kc3-d2 Re3-e7 winkensmile:I want to get my King to d3, but Black's Knight threatens Nb4+, so I have to take it off. 56.Bd5xc6 winkensmile:After 55. Bxc3, Black faces the loss of another pawn, and only has a few useless checks which only helps me bring my King over to escort the h pawn along.
Kd6xc6 57.Ra2xa5 Re7-h7 winkensmile:Black is down another pawn, and my h pawn is gathering strength. My Rook and Knight will prevent Black's King from getting anywhere close to the K-side, and Black will have to eventually sac his Rook on the h pawn promotion, (or worse, get forced into trading off Rooks, where White gets to keep the Queen after promotion), and face the inevitable. 58.Kd2-e3 winkensmile:At this point, Fritz conceded, "there are no more good moves left, and it's time to hit the Resign Button." 1-0