Confused by engine analysis

Looking for an easy quick game against a 1200 player I foolishly thought the first few moves were weak enough to allow me some early tactics so I jumped in with my knight to see what I could make of it...and before I could fully formulate all the sage advice I would happily share after my victory the game turned around. Within a few moves their position looked really solid and I was a shambles struggling to find useful moves. I hung on just to check they had the endgame to finish me off and they did.

Game 4173602

Next I thought I'd run the game through my Sigma engine giving it 2 mins per move to see what I could learn...already recognising that the early "weak" moves could be a tactic to pull people of their databases. I was playing quickly but then so was my opponent.

Anyway what I can't understand is how few of my own (ultimately losing) moves are offered improvements by the engine and how many of my opponents moves get suggestions for improvements. I would expect it to be the other way round. I've run this a couple of times to be sure it's not just skipping my moves but the result is the same.

Any suggestions.

This is what spewed out of Sigma - I have the black pieces - notice how the improvements in curly brackets so often follow whites moves:
1. d4 Nf6 2. e3 e6 3. a3 { +0.22 3. Nf3 d5 4. Be2 c5 5. O-O Nbd7 6. Nbd2 c4 }
Be7 4. h3 Ne4 5. Bd3 d5 6. Qg4 { +0.35 6. Bxe4 dxe4 7. Qg4 O-O 8. Qxe4 Bd7 9.
Qf4 a5 10. Nf3 Ba4 } g6 { +0.03 6... h5 7. Qd1 b6 8. c4 c5 9. Bxe4 dxe4 10. Nc3
} 7. Bxe4 f5 8. Bxf5 exf5 9. Qd1 { +0.68 9. Qf3 O-O 10. Nd2 c6 11. Ne2 a5 12.
O-O b6 } Na6 10. Nf3 { +1.12 10. Nd2 Qd6 11. b3 Be6 12. Ngf3 O-O 13. Ne5 c5 14.
Bb2 } c5 11. c3 { +1.09 11. Nc3 Be6 12. dxc5 Nxc5 13. O-O Ne4 14. Nxe4 dxe4 15.
Qxd8 Rxd8 16. Nd4 } cxd4 12. cxd4 { +1.29 12. exd4 O-O 13. b3 f4 14. O-O Bd7
15. Ne5 } Qd6 13. Bd2 { +1.22 13. Ne5 O-O 14. O-O Be6 15. b4 Rfe8 16. b5 Nc7
17. Nc3 } Bd7 14. Qb3 { +1.37 14. Ne5 O-O 15. Nxd7 Qxd7 16. Nc3 Rac8 17. Qf3
Nc7 18. O-O } Qc7 15. Nc3 { +2.13 15. Qxd5 Bc6 16. Qb3 O-O-O 17. O-O Kb8 18.
Rc1 b5 } Be6 16. Rc1 Qb6 17. Qxb6 axb6 18. O-O { +2.02 18. Ne2 b5 19. Ne5 g5
20. O-O O-O 21. Ba5 Rac8 } O-O 19. Na4 { +1.93 19. Ne5 Rfe8 20. Nb5 Rab8 21. b3
Kh8 22. f3 Rbc8 23. Rxc8 Rxc8 } b5 20. Nc3 { +1.72 20. Nc5 Nxc5 21. dxc5 Bf6
22. Nd4 Bd7 23. Bb4 Be5 } Nc7 21. Ne2 Ne8 22. Nf4 Ng7 23. Rc7 Bd6 24. Rxb7 {
+5.33 24. Rxg7 Kxg7 25. Nxe6 Kg8 26. Nxf8 Rxf8 27. Rc1 Rb8 28. Ne5 Re8 29. Bb4
Bxb4 30. axb4 } Rab8 25. Rxb8 { +6.10 25. Rxg7 Kxg7 26. Nxe6 Kg8 27. Nxf8 Rxf8
28. Rc1 Rb8 29. Rc6 Be7 30. Rc7 Bd6 31. Rd7 } Rxb8 26. Nxe6 { +2.69 26. Ne5 Rc8
27. Rc1 Rxc1 28. Bxc1 g5 29. Nxe6 Nxe6 30. Bd2 f4 31. g3 } Nxe6 27. Rc1 Rb6 28.
Bc3 { +2.52 28. Ne5 Kg7 29. Nd7 Rb7 30. Nc5 Re7 31. Bb4 Ng5 32. Nd3 Ne4 33.
Bxd6 Nxd6 } Kf7 29. Nd2 { +2.48 29. Ne5 Ke8 30. Ba5 Rb8 31. Rc6 Ke7 32. Rb6
Rxb6 33. Bxb6 Kf6 } b4 30. axb4 { +2.58 30. Bxb4 Bxb4 31. axb4 Rxb4 32. b3 Rb8
33. g3 Ra8 34. Nf3 Kg8 35. h4 } Bxb4 31. Bxb4 Rxb4 32. b3 Kf6 33. Ra1 { +2.25
33. Rc6 h5 34. Nb1 Ke7 35. Nc3 Kd7 36. Nxd5 Rxd4 37. exd4 Kxc6 38. Ne7 Kd6 39.
Nxg6 Nxd4 } Rb5 34. Ra4 { +2.31 34. Ra7 h5 35. f4 Rb6 36. Rd7 Rb5 37. g3 Ra5 }
Nd8 35. Ra6+ { +2.43 35. g4 Ne6 36. gxf5 Kxf5 37. Ra7 h6 38. Rf7 Kg5 39. f4 Kh5
40. Re7 } Ne6 36. g4 { +2.31 36. Ra7 h5 37. f4 Rb6 38. Rd7 Rb5 39. g3 Ra5 }
fxg4 37. hxg4 h6 38. f4 { +2.77 38. f3 Rb8 39. Ra5 Nc7 40. Kf2 Rb7 41. e4 } Ke7
39. f5 gxf5 40. gxf5 Ng5 41. Kg2 { +3.48 41. Rxh6 Kf7 42. f6 Ne4 43. Nxe4 dxe4
44. b4 Rxb4 45. d5 Rb1 46. Kg2 Rb2 47. Kg3 } Ne4 42. Nxe4 dxe4 43. Rxh6 { +3.13
43. Re6 Kf7 44. Rxe4 Rxf5 45. Rf4 Rxf4 46. exf4 h5 47. b4 Ke6 48. b5 Kd5 49. b6
} Rxb3 44. Re6+ Kf7 45. Rxe4 Kf6 46. Re5 Rd3 47. Kf3 Rd1 48. Ke4 Rh1 49. Re6+
Kf7 50. Ke5 Rh5 51. e4 { +5.62 51. Rb6 Ke8 52. d5 Kd7 53. d6 Kc8 54. e4 Rh2 55.
f6 } Rh8 52. d5 Re8 53. Rxe8 *

Originally posted by Mahout
Looking for an easy quick game against a 1200 player I foolishly thought the first few moves were weak enough to allow me some early tactics so I jumped in with my knight to see what I could make of it...and before I could fully formulate all the sage advice I would happily share after my victory the game turned around. Within a few moves their position lo 4 { +5.62 51. Rb6 Ke8 52. d5 Kd7 53. d6 Kc8 54. e4 Rh2 55.
f6 } Rh8 52. d5 Re8 53. Rxe8 *

I stopped looking at the game after 9...Na6?. Moving the knight there means that it is out of the game because it has to go backwards to advance and that is just a waste of time! didn't you notice that it wouldn't be able to move anywhere besides its original square! and then move 20...Nc7 ...b4 was better because it was more aggressive. In short you played too passively which is why the computer didn't get it because it doesn't understand positional concepts!

well you dropped a pawn, that's +1. a lost game. then you just randomly shuffled pieces around while white developed and improved his pieces. allowing the queen exchange burned your queenside to the ground. that's +2. then you dropped the bishop, +5, which he missed, dropped another pawn +6 (the bishop is still hanging), he exchanged the rooks which meant he couldn't take the bishop anymore, back to +2...

mostly you were doing just random, passive, pointless moves. almost all of the evaluation is strictly material. pay attention to material, it's extremely important that you don't drop anything. and have a purpose when you move a piece.

mostly you were doing just random, passive, pointless moves.

...don't be afraid to tell it like it is! So given the random and pointless nature of the moves why isn't the engine finding better ones?

Originally posted by Mahout
...don't be afraid to tell it like it is! So given the random and pointless nature of the moves why isn't the engine finding better ones?

Maybe the best moves are actually crap? Maybe it is just Sigma being a tad dim. See if you can find Fruit 2.1 for Mac. It is free and works with Sigma and may give you a more sensible answer.

Originally posted by Mahout
...don't be afraid to tell it like it is! So given the random and pointless nature of the moves why isn't the engine finding better ones?

are you kidding me? you lost huge amount of material (because you did random, pointless moves instead of defending the material), wrecked your queenside, allowed white to penetrate, allowed him to create a passed pawn...

find the moves to stop these things (without allowing further advantage) to begin with, and you'll have your absolutely necessary basic improvements.

Originally posted by wormwood
are you kidding me? you lost huge amount of material (because you did random, pointless moves instead of defending the material), wrecked your queenside, allowed white to penetrate, allowed him to create a passed pawn...

find the moves to stop these things (without allowing further advantage) to begin with, and you'll have your absolutely necessary basic improvements.

I don't think he wants advice on how to improve. He is asking why his engine fails to find fault with his poor moves but suggests so many improvements for his opponent.

Originally posted by Kepler
I don't think he wants advice on how to improve. He is asking why his engine fails to find fault with his poor moves but suggests so many improvements for his opponent.

ah, okay, now I got it.

it does look weird that the engine doesn't give any improvements on blunders like 23...Bd6. it comments only after white moves the rook. it does comment on black on move 6 though, so it comments on some black moves... can't be the blunder threshold either as it said nothing about dropping a piece. weird...

I tried running the game through Sigma and got a totally different result. I used 5 seconds per move and I am willing to bet my Mac is more ancient than yours! This time it seems the improvements are more evenly distributed. Sigma even catches 52. ... Re8 and is very rude about it. I have some thoughts on why you got a dodgy analysis. You may not have set Sigma on its full strength although generally Sigma squeals about that when set to analyse. It is possible the extra time you allowed per move confused Sigma. I have noticed that Sigma seems to play worse when allowed a long time in which to think and two minutes is a long time on a modern Mac.

For purposes of comparison here is the output I got:

[Event "Open invite"]
[Site "http://www.playtheimmortalgame.com"]
[Date "2007.10.22"]
[Round "?"]
[White "half cut hero"]
[Black "Mahout"]
[Result "1-0"]

1. d4 Nf6 2. e3 e6 { (-0.30) book 2... Nc6 } 3. a3 Be7 4. h3 { (-0.04) +0.33 4.
Nf3 b6 5. Nbd2 d5 6. Bd3 Bb7 } Ne4 { (-0.26) +0.04 4... d5 5. Nf3 O-O 6. Bd3
Nc6 7. Nbd2 Bd7 8. O-O } 5. Bd3 d5 6. Qg4 { (-0.15) +0.27 6. Nd2 f5 7. Ne2 Bd7
8. O-O O-O 9. f3 Bc6 10. Bxe4 fxe4 11. fxe4 Rxf1 12. Kxf1 dxe4 } g6 { (-0.39)
+0.15 6... Ng5 7. Nf3 f5 8. Qg3 Nxf3 9. Qxf3 O-O 10. Nd2 Bd7 } 7. Bxe4 f5 8.
Bxf5 exf5 9. Qd1 Na6 { (-1.00) -0.64 9... O-O 10. Nf3 Be6 11. Nbd2 Bd6 12. b3
f4 13. e4 } 10. Nf3 c5 11. c3 { (+0.75) +1.03 11. O-O Be6 12. b3 Qc7 13. Bb2
O-O 14. Nbd2 } cxd4 { (-1.24) -0.75 11... Be6 12. b4 c4 13. Bb2 O-O 14. Nbd2 b6
} 12. cxd4 Qd6 13. Bd2 Bd7 { (-1.29) -1.01 13... O-O 14. Qb3 Be6 15. Nc3 } 14.
Qb3 Qc7? { (-2.19) -1.09 14... Bc6 15. Nc3 Nc7 16. Ne5 O-O 17. e4 Rab8 } 15.
Nc3 Be6 16. Rc1 Qb6 { (-2.55) -2.16 16... Rd8 17. Qb5 Kf8 18. Ne2 Qb8 19. Nf4
Bf7 20. Ba5 Rc8 21. O-O } 17. Qxb6 axb6 18. O-O { (+1.65) +2.09 18. Ne2 Bd6 19.
Nf4 Bg8 20. O-O Bf7 21. g4 Nc7 22. gxf5 gxf5 } O-O 19. Na4 { (+1.61) +2.01 19.
Nb5 Rfd8 20. Ne5 Rdb8 21. b3 Bf6 22. f4 Rc8 23. Rxc8 Rxc8 } b5 20. Nc3 {
(+1.47) +1.73 20. Nc5 Nxc5 21. dxc5 Bd7 22. Ne5 Be6 23. Bb4 Bf6 24. f4 } Nc7 {
(-2.04) -1.47 20... b4 21. Nb5 bxa3 22. bxa3 Rfd8 23. Ne5 Re8 24. Kh2 Rec8 25.
Rxc8 Rxc8 } 21. Ne2 Ne8 { (-2.17) -1.91 21... Rfc8 22. Nf4 Bf7 23. Ne5 Bd6 24.
Nxf7 Kxf7 25. h4 Be7 } 22. Nf4 Ng7 23. Rc7 Bd6?? { (-5.15) -2.68 23... Rfe8 24.
Rfc1 Rab8 25. Nxe6 Nxe6 26. Rd7 Rbd8 27. Rxb7 Rb8 28. Rxb8 Rxb8 } 24. Rxb7?? {
(+3.27) +5.15 24. Rxg7 Kxg7 25. Nxe6 Kg8 26. Nxf8 Rxf8 27. Rc1 Ra8 28. g3 b6 }
Rab8?? { (-5.95) -3.27 24... Bc8 25. Rxb5 Bxf4 26. exf4 Ba6 27. Rxd5 Bxf1 28.
Kxf1 } 25. Rxb8?? { (+2.89) +5.95 25. Rxg7 Kxg7 26. Nxe6 Kg8 27. Nxf8 Rxf8 28.
Rc1 Re8 29. Rc6 Rd8 } Rxb8 26. Nxe6 Nxe6 27. Rc1 Rb6 28. Bc3 { (+2.20) +2.51
28. Ne5 Kg7 29. Nc6 h6 30. Bb4 Bxb4 31. Nxb4 Rd6 } Kf7 { (-2.48) -2.20 28...
Rc6 29. Ne5 Ra6 30. f4 Kg7 31. g4 Ra8 32. gxf5 gxf5 } 29. Nd2 { (+2.14) +2.48
29. Ne5 Kg7 30. Bd2 Ra6 31. Rc8 Kf6 32. Nd7 Kf7 33. Nc5 Ra7 34. Nxe6 Kxe6 } b4
{ (-2.54) -2.14 29... Rb8 30. Nf3 Re8 31. Bd2 Rd8 32. Ne5 Kf6 33. f4 Ra8 } 30.
axb4 Bxb4 31. Bxb4 Rxb4 32. b3 Kf6 33. Ra1 Rb5 34. Ra4 Nd8 35. Ra6+ Ne6 36. g4
fxg4 37. hxg4 h6 38. f4 Ke7? { (-3.68) -2.87 38... h5 39. gxh5 gxh5 40. Kf2 Rb4
41. Ra5 Nxf4 42. exf4 Rxd4 43. Ra6 Kf5 } 39. f5 gxf5 40. gxf5 Ng5 41. Kg2 {
(+3.19) +3.77 41. Rxh6 Ra5 42. Rg6 Nf7 43. Kf2 Nd6 44. Re6 Kd7 45. Re5 } Ne4
42. Nxe4 dxe4 43. Rxh6 Rxb3? { (-4.41) -2.95 43... Rxf5 44. Rh4 Rg5 45. Kf2 Rf5
46. Rf4 Rb5 47. Rxe4 Kd6 48. Kf3 Rxb3 49. Re8 } 44. Re6+ Kf7 45. Rxe4 Kf6 46.
Re5 Rd3 47. Kf3 Rd1 48. Ke4 Rh1 49. Re6+ Kf7 50. Ke5 Rh5 51. e4 Rh8 52. d5
Re8?? { (-21.63) -6.12 52... Ra8 53. Rc6 Ra7 54. d6 Ra5 55. Kf4 Ra2 56. Rc7 Ke8
57. Ke5 Rd2 58. f6 } 53. Rxe8 1-0

Originally posted by Kepler
I tried running the game through Sigma and got a totally different result. I used 5 seconds per move and I am willing to bet my Mac is more ancient than yours! This time it seems the improvements are more evenly distributed. Sigma even catches 52. ... Re8 and is very rude about it. I have some thoughts on why you got a dodgy analysis. You may not have set Sig 53. Rc6 Ra7 54. d6 Ra5 55. Kf4 Ra2 56. Rc7 Ke8
57. Ke5 Rd2 58. f6 } 53. Rxe8 1-0

OK, thanks - very interesting, (it's a G4 with 1.67 GHZ processer with 1GB of RAM) I'll give it a try with the shorter time controls...would never have occurred to me to try that...and glancing through your analysis it makes more sense.

I haven't looked at the game, but don't rely on such weak engines like Sigma Chess. Sure, it will beat me most of the times, but I see its errors. Sometimes they are small positional errors, sometimes more. Use a good chess engine, HIARCS 11.2 is fantastic and Glaurung is free and much stronger than Sigma Chess, but much weaker than HIARCS.

OK, just looking at it briefly without engines.

4. ..Ne4? There is nothing to attack and you need to develop or play a move like c5. Sorry, but this is just bad.

6. ...g6? Just castle, there is no Bh6 and White is not well developed.

7. ...f5? Just gives a pawn and weakens the already weak kingside.

8. ...Na6!? is pretty bad, but at least it puts more force on c5 and doesn't block the bishop.

Ok, I think I had enough. 😉 It's a blitz game anyway and so it's not worth analyzing too deeply. If you had thought a long time on each move, then you really have to see why you would still make them moves.

Originally posted by exigentsky
I haven't looked at the game, but don't rely on such weak engines like Sigma Chess. Sure, it will beat me most of the times, but I see its errors. Sometimes they are small positional errors, sometimes more. Use a good chess engine, HIARCS 11.2 is fantastic and Glaurung is free and much stronger than Sigma Chess, but much weaker than HIARCS.

I don't think sigma would stand a chance against even me if it played like it analyzed this game.

Originally posted by tomtom232
I don't think sigma would stand a chance against even me if it played like it analyzed this game.

Sigma is OK if set at a reasonable level and given sensible time controls. It tends not to behave too well if given strict "move in x secs/mins" with no leeway. Tell it to play x moves in n minutes and allow it to decide how long to spend on each move and it tends to play better. I find it is mainly useful because it includes various database functions and acts as a GUI for UCI engines.

I notice that Kepler's version gives the difference in score between the actual move and the suggested move, whereas Mahout's version gives only one score. I suspect Mahout has some option set incorrectly in the software, but since I've never used that program I can't say what it might be. (Something other than just the time controls, I mean.)