Contemptuous

[overlord]

You must be joking It is not about your strength...you can be absolutly average player (2000-2100 FIDE ELO), but you need to know how to play against engines. See following game...did white played even one single move that would not play very average chess player???

[pgn][Event "Blitz 5m"]
[Site "?"]
[Date "2013.12.19"]
[Round "?"]
[White "Kvicala, Miroslav"]
[Black "Stockfish DD 64 SSE4.2"]
[Result "1/2-1/2"]
[PlyCount "278"]
[TimeControl "300"]

{509MB, Mirekbook.ctg, BobaBobek-PC} 1. d4 {0} e6 {0} 2. c4 {1} Bb4+ {0} 3. Bd2
{1} Qe7 {0} 4. Nf3 {2} Bxd2+ {6} 5. Qxd2 {2} Nf6 {3} 6. Nc3 {1} O-O {5} 7. e3 {
1} d6 {6} 8. Rc1 {3} b6 {9} 9. Be2 {1} Bb7 {9} 10. O-O {0} Ne4 {6} 11. Nxe4 {5}
Bxe4 {1} 12. Qc3 {2} c5 {12} 13. a3 {2} Nd7 {5} 14. dxc5 {8} dxc5 {5} 15. Rfd1
{1} Bc6 {9} 16. Ne5 {8} Nxe5 {4} 17. Qxe5 {1} Ba4 {3} 18. Rd2 {6} Rad8 {5} 19.
Qc3 {6} Rxd2 {5} 20. Qxd2 {0} Rd8 {6} 21. Qc3 {1} f6 {5} 22. b3 {3} Bc6 {3} 23.
Rd1 {6} Rxd1+ {0} 24. Bxd1 {1} Qd6 {5} 25. Be2 {10} a5 {2} 26. a4 {6} e5 {6}
27. Qd3 {6} Qxd3 {0} 28. Bxd3 {1} e4 {6} 29. Bc2 {3} Kf7 {4} 30. h4 {4} Ke6 {4}
31. g3 {2} Ke5 {4} 32. Bd1 {4} Be8 {4} 33. Kg2 {5} Kf5 {4} 34. Be2 {6} g6 {0}
35. Bd1 {5} Bc6 {4} 36. Be2 {1} Bd7 {5} 37. Bd1 {1} Ke5 {9} 38. Be2 {4} g5 {6}
39. Bd1 {7} g4 {4} 40. Be2 {3} f5 {3} 41. Bd1 {1} h6 {4} 42. Bc2 {1} Bc6 {3}
43. Bd1 {2} Be8 {3} 44. Bc2 {1} Bh5 {2} 45. Bd1 {5} Bf7 {3} 46. Bc2 {1} Be6 {3}
47. Kf1 {1} Bc8 {3} 48. Ke2 {0} Bd7 {3} 49. Kf1 {0} Ke6 {3} 50. Ke2 {0} Be8 {3}
51. Kf1 {0} Bf7 {3} 52. Ke2 {0} Ke5 {2} 53. Kf1 {0} Bh5 {3} 54. Ke2 {0} Bg6 {3}
55. Kf1 {0} Kd6 {2} 56. Ke2 {0} Bh5 {2} 57. Kf1 {0} Bf7 {2} 58. Ke2 {0} Be6 {3}
59. Kf1 {0} Bd7 {2} 60. Ke2 {0} Bc6 {2} 61. Kf1 {0} Ke5 {2} 62. Ke2 {0} Be8 {2}
63. Kf1 {0} Bg6 {2} 64. Ke2 {0} Bh5 {2} 65. Kf1 {0} Kd6 {2} 66. Ke2 {0} Be8 {2}
67. Kf1 {0} Bc6 {2} 68. Ke2 {0} Bd7 {1} 69. Kf1 {0} Be8 {2} 70. Ke1 {0} Kc7 {1}
71. Kd1 {0} Kc6 {1} 72. Kc1 {0} Kd6 {2} 73. Kb1 {0} Bc6 {1} 74. Ka1 {0} Bd7 {3}
75. Kb1 {0} Ke5 {1} 76. Kc1 {0} Bc8 {2} 77. Kd1 {0} Kf6 {1} 78. Ke1 {0} Bd7 {1}
79. Kf1 {0} Bc6 {1} 80. Kg1 {0} Ke6 {1} 81. Kh1 {0} Be8 {1} 82. Kg1 {0} Bf7 {1}
83. Kf1 {0} Bg6 {6} 84. Ke1 {0} Bh5 {1} 85. Kd1 {0} Be8 {1} 86. Kc1 {0} Ke5 {1}
87. Kb1 {0} Bc6 {1} 88. Ka1 {0} Ke6 {1} 89. Kb1 {0} h5 {1} 90. Kb2 {0} Ke5 {1}
91. Kc1 {0} Bb7 {1} 92. Kd1 {0} Bc8 {1} 93. Ke1 {0} Ba6 {1} 94. Kf1 {0} Bb7 {1}
95. Kg1 {0} Bc8 {1} 96. Kh1 {0} Be6 {1} 97. Kg2 {0} Bd7 {0} 98. Kg1 {0} Be8 {0}
99. Kg2 {0} Bc6 {0} 100. Kg1 {0} Bb7 {1} 101. Kg2 {0} Kd6 {1} 102. Kg1 {0} Bc6
{1} 103. Kg2 {0} Bd7 {1} 104. Kg1 {0} Be8 {0} 105. Kg2 {0} Ke6 {0} 106. Kg1 {0}
Bc6 {0} 107. Kg2 {0} Bd7 {0} 108. Kg1 {0} Bc8 {1} 109. Kg2 {0} Kf6 {0} 110. Kg1
{0} Be6 {0} 111. Kg2 {0} Ke5 {0} 112. Kg1 {0} Bf7 {0} 113. Kg2 {0} Bg6 {0} 114.
Kg1 {0} Bh7 {0} 115. Kg2 {0} Bg8 {0} 116. Kg1 {0} Kd6 {0} 117. Kg2 {0} Be6 {0}
118. Kg1 {0} Bc8 {0} 119. Kg2 {0} Kc6 {0} 120. Kg1 {0} Bd7 {0} 121. Kg2 {0} Be6
{0} 122. Kg1 {0} Bg8 {0} 123. Kg2 {0} Bf7 {0} 124. Kg1 {1} Kd6 {0} 125. Kf1 {0}
Bg6 {0} 126. Ke1 {1} Ke5 {0} 127. Kd2 {1} Be8 {0} 128. Kd1 {0} Bd7 {0} 129. Kd2
{0} Be6 {0} 130. Kd1 {0} Bf7 {0} 131. Kd2 {0} Kd6 {0} 132. Kd1 {0} Be6 {0} 133.
Kd2 {0} Bd7 {0} 134. Kd1 {0} Bc6 {0} 135. Kd2 {0} Bb7 {1} 136. Kd1 {0} Ke6 {0}
137. Kd2 {0} Ba6 {0} 138. Kd1 {0} Ke5 {0} 139. Kd2 {0} Bb5 {0} 1/2-1/2[/pgn][/list][/list]

[Lyudmil Tsvetkov]

You don´t have to be strong player at all. You have to know to play against engines. See next game...is there even one move which would not play avarage player (e.g.2000 - 2100 ELO)?

[pgn][Event "Blitz 5m"]
[Site "?"]
[Date "2013.12.19"]
[Round "?"]
[White "Kvicala, Miroslav"]
[Black "Stockfish DD 64 SSE4.2"]
[Result "1/2-1/2"]
[PlyCount "278"]
[TimeControl "300"]

{509MB, Mirekbook.ctg, BobaBobek-PC} 1. d4 {0} e6 {0} 2. c4 {1} Bb4+ {0} 3. Bd2
{1} Qe7 {0} 4. Nf3 {2} Bxd2+ {6} 5. Qxd2 {2} Nf6 {3} 6. Nc3 {1} O-O {5} 7. e3 {
1} d6 {6} 8. Rc1 {3} b6 {9} 9. Be2 {1} Bb7 {9} 10. O-O {0} Ne4 {6} 11. Nxe4 {5}
Bxe4 {1} 12. Qc3 {2} c5 {12} 13. a3 {2} Nd7 {5} 14. dxc5 {8} dxc5 {5} 15. Rfd1
{1} Bc6 {9} 16. Ne5 {8} Nxe5 {4} 17. Qxe5 {1} Ba4 {3} 18. Rd2 {6} Rad8 {5} 19.
Qc3 {6} Rxd2 {5} 20. Qxd2 {0} Rd8 {6} 21. Qc3 {1} f6 {5} 22. b3 {3} Bc6 {3} 23.
Rd1 {6} Rxd1+ {0} 24. Bxd1 {1} Qd6 {5} 25. Be2 {10} a5 {2} 26. a4 {6} e5 {6}
27. Qd3 {6} Qxd3 {0} 28. Bxd3 {1} e4 {6} 29. Bc2 {3} Kf7 {4} 30. h4 {4} Ke6 {4}
31. g3 {2} Ke5 {4} 32. Bd1 {4} Be8 {4} 33. Kg2 {5} Kf5 {4} 34. Be2 {6} g6 {0}
35. Bd1 {5} Bc6 {4} 36. Be2 {1} Bd7 {5} 37. Bd1 {1} Ke5 {9} 38. Be2 {4} g5 {6}
39. Bd1 {7} g4 {4} 40. Be2 {3} f5 {3} 41. Bd1 {1} h6 {4} 42. Bc2 {1} Bc6 {3}
43. Bd1 {2} Be8 {3} 44. Bc2 {1} Bh5 {2} 45. Bd1 {5} Bf7 {3} 46. Bc2 {1} Be6 {3}
47. Kf1 {1} Bc8 {3} 48. Ke2 {0} Bd7 {3} 49. Kf1 {0} Ke6 {3} 50. Ke2 {0} Be8 {3}
51. Kf1 {0} Bf7 {3} 52. Ke2 {0} Ke5 {2} 53. Kf1 {0} Bh5 {3} 54. Ke2 {0} Bg6 {3}
55. Kf1 {0} Kd6 {2} 56. Ke2 {0} Bh5 {2} 57. Kf1 {0} Bf7 {2} 58. Ke2 {0} Be6 {3}
59. Kf1 {0} Bd7 {2} 60. Ke2 {0} Bc6 {2} 61. Kf1 {0} Ke5 {2} 62. Ke2 {0} Be8 {2}
63. Kf1 {0} Bg6 {2} 64. Ke2 {0} Bh5 {2} 65. Kf1 {0} Kd6 {2} 66. Ke2 {0} Be8 {2}
67. Kf1 {0} Bc6 {2} 68. Ke2 {0} Bd7 {1} 69. Kf1 {0} Be8 {2} 70. Ke1 {0} Kc7 {1}
71. Kd1 {0} Kc6 {1} 72. Kc1 {0} Kd6 {2} 73. Kb1 {0} Bc6 {1} 74. Ka1 {0} Bd7 {3}
75. Kb1 {0} Ke5 {1} 76. Kc1 {0} Bc8 {2} 77. Kd1 {0} Kf6 {1} 78. Ke1 {0} Bd7 {1}
79. Kf1 {0} Bc6 {1} 80. Kg1 {0} Ke6 {1} 81. Kh1 {0} Be8 {1} 82. Kg1 {0} Bf7 {1}
83. Kf1 {0} Bg6 {6} 84. Ke1 {0} Bh5 {1} 85. Kd1 {0} Be8 {1} 86. Kc1 {0} Ke5 {1}
87. Kb1 {0} Bc6 {1} 88. Ka1 {0} Ke6 {1} 89. Kb1 {0} h5 {1} 90. Kb2 {0} Ke5 {1}
91. Kc1 {0} Bb7 {1} 92. Kd1 {0} Bc8 {1} 93. Ke1 {0} Ba6 {1} 94. Kf1 {0} Bb7 {1}
95. Kg1 {0} Bc8 {1} 96. Kh1 {0} Be6 {1} 97. Kg2 {0} Bd7 {0} 98. Kg1 {0} Be8 {0}
99. Kg2 {0} Bc6 {0} 100. Kg1 {0} Bb7 {1} 101. Kg2 {0} Kd6 {1} 102. Kg1 {0} Bc6
{1} 103. Kg2 {0} Bd7 {1} 104. Kg1 {0} Be8 {0} 105. Kg2 {0} Ke6 {0} 106. Kg1 {0}
Bc6 {0} 107. Kg2 {0} Bd7 {0} 108. Kg1 {0} Bc8 {1} 109. Kg2 {0} Kf6 {0} 110. Kg1
{0} Be6 {0} 111. Kg2 {0} Ke5 {0} 112. Kg1 {0} Bf7 {0} 113. Kg2 {0} Bg6 {0} 114.
Kg1 {0} Bh7 {0} 115. Kg2 {0} Bg8 {0} 116. Kg1 {0} Kd6 {0} 117. Kg2 {0} Be6 {0}
118. Kg1 {0} Bc8 {0} 119. Kg2 {0} Kc6 {0} 120. Kg1 {0} Bd7 {0} 121. Kg2 {0} Be6
{0} 122. Kg1 {0} Bg8 {0} 123. Kg2 {0} Bf7 {0} 124. Kg1 {1} Kd6 {0} 125. Kf1 {0}
Bg6 {0} 126. Ke1 {1} Ke5 {0} 127. Kd2 {1} Be8 {0} 128. Kd1 {0} Bd7 {0} 129. Kd2
{0} Be6 {0} 130. Kd1 {0} Bf7 {0} 131. Kd2 {0} Kd6 {0} 132. Kd1 {0} Be6 {0} 133.
Kd2 {0} Bd7 {0} 134. Kd1 {0} Bc6 {0} 135. Kd2 {0} Bb7 {1} 136. Kd1 {0} Ke6 {0}
137. Kd2 {0} Ba6 {0} 138. Kd1 {0} Ke5 {0} 139. Kd2 {0} Bb5 {0} 1/2-1/2[/pgn][/url]

Hi Miroslav.

I can confirm that I get against Stockfish not only similar positions, but regarding the particular game, an absolutely identical position.

When you reach that level of play, it is not necessary to play the game to the end when it is an obvious draw. You might have adjudicated a draw already at move 40, but you might have been training speedy movements.
Nice drawing finish by Stockfish at move 139.

[mohzus]

Hey Miroslav, was the last game you posted here with SF with contempt =50 cp?
In all cases what you have shown guys is that drawing against SF with contempt=50 cp does not seem much harder than without contempt. Beating it seems another story though. Interesting...

[overlord]

Hello Robert, yes, make some draw is much easier than win. The last game was played with contempt=0. In this case I wanted to show that it is not necessary to play some extraordinary moves. There are some kinds of positions where it is not so difficult at all. I believe that last game demonstrated it properly. Everybody is speaking about inceasing engines strength and improving hardware. Well, it is true. Engines like Stockfish or Houdini are amazing pieces of software. I remember the time when most people believed that Fritz 6 is almost unbeatable. Try to play against it today...it has so many weaknesses...We, human are very adaptable beings, so we can learn also from engines. We can investigate their weak points and benefit from them. Also simple trial and error method is progressive way how to play against machines. There are so many ways...

[carldaman]

I do not know how Stockfish contempt works, might not be implemented in the most sophisticated way, but what I managed to ascertain, as expected, is that engines with contempt set on in no way play any better against humans.

Stockfish simply uses the contempt value to score draws. These are only the draws by chess rules (not by chess knowledge): 3-repetition, 50 move, stalemate, insufficient material. Obviously, in most cases, it's the 3-repetition draw that contempt helps avoid. For example, if the engine is given the choice between accepting a position worth -20cp and going for a draw by repetition, it will not accept the draw if contempt > 20cp, and will accept the draw when contempt <= 20cp.

This is the standard and most basic way of implementing a contempt feature. SF also does some dubious rescaling of the contempt score, but that's a detail.

Even with large contempt, SF will not avoid blocked positions. Often you manage to draw or even win by creating a deadly bind, which prevents the computer from exploiting its tactical superiority. A nice improvement of SF contempt would be to make it avoid closed positions, by introducing some asymmetric eval term. I can easily code it, and submit it to Marco. The real difficulty is to get Marco to commit the code...

Otherwise I can put an anti human feature in DiscoCheck. Interested ?

Thanks, Lucas! Highly-interested

As for SF, there is nothing wrong with Marco adding another parameter to SF, so such a feature can be activated by the user if so desired.

Regards,
CL

[carldaman]

Otherwise I can put an anti human feature in DiscoCheck. Interested ?

I am very much interested in anything that would not allow me even a single win.

The problem is if this is achievable...
I think DiscoCheck with most sophisticated contempt still would not be able to surpass the level of play against humans of Stockfish with whatever contempt. But of course, very nice idea, I would like to see it implemented.

There are engines that *may* be able to play better against humans, despite being hundred of points below SF in rating, because of their playing style. It remains an educated conjecture until established by actual further tests, of course.

Among the free engines, Thinker 5.4D Active, intended to only play against humans, is available for download somewhere. Hannibal is also quite interesting among the stronger free non-derivatives, and then there's its older cousin, called Twisted Logic. Two engines developed by Pawel Koziol, Rodent and Glass, are worth a look at. SmarThink is now free (v1.40), and very interesting, too.

Rebel by Ed Schroeder was known for its anti-GM functionality, but I'm not sure if the feature is implemented in ProDeo, the free but stronger version of Rebel.

You could also experiment with many of the free 'weaker' engines such as LittleThought (can play closed positions and flank attacks reasonably well in tests), and even Delphil, both around the 2500 CCRL Elo mark.

There are of course commercial engines, but some are hard to get these days -- Deep Fritz 10 is one of my favorite engines, especially in closed positions, and Zappa Mexico II with its crazy tweaked parameters (settings available thanks to Robert Flesher). I've had it play online in the past, and it destroyed every human it played at 5m blitz, despite playing wildly unsound chess half the time.

[Lyudmil Tsvetkov]

There are engines that *may* be able to play better against humans, despite being hundred of points below SF in rating, because of their playing style. It remains an educated conjecture until established by actual further tests, of course.

Among the free engines, Thinker 5.4D Active, intended to only play against humans, is available for download somewhere. Hannibal is also quite interesting among the stronger free non-derivatives, and then there's its older cousin, called Twisted Logic. Two engines developed by Pawel Koziol, Rodent and Glass, are worth a look at. SmarThink is now free (v1.40), and very interesting, too.

Rebel by Ed Schroeder was known for its anti-GM functionality, but I'm not sure if the feature is implemented in ProDeo, the free but stronger version of Rebel.

You could also experiment with many of the free 'weaker' engines such as LittleThought (can play closed positions and flank attacks reasonably well in tests), and even Delphil, both around the 2500 CCRL Elo mark.

There are of course commercial engines, but some are hard to get these days -- Deep Fritz 10 is one of my favorite engines, especially in closed positions, and Zappa Mexico II with its crazy tweaked parameters (settings available thanks to Robert Flesher). I've had it play online in the past, and it destroyed every human it played at 5m blitz, despite playing wildly unsound chess half the time.

I fully agree with you, Carl, that many weaker engines provide a useful variety to the end user. I also agree that most of those engines are very interesting by themselves. I could also share to a point the view that some of those engines would be a bit difficult opponents for a human to play against because of their style, some preferring open positions, and other playing too many random moves so that one can not get accustomed to the style of the engine.

However, I do not believe that such engines could be in any way stronger than top engines against humans. Maybe in blitz, when speculative play often pays off, but definitely not in normal LTC, where weak moves are simply weak moves.

I have played almost all of the engines you mention (once I was a geek almost as you and Dr. D, playing every conceivable engine, but I now stick mostly to the top ) , and what I remember is:

- 17 move win against Thinker (but maybe it was the 53b version after all)
- easy wins against Fritz 10, Fritz was in the habit to get losing positions in almost every single game, but would still win the bigger share
- more successful performance against Zappa than against any of the current top engines (is Zappa a gardener's tool?)

Usually, and this is my conclusion from the 50 000 engine games I have played in the past 15 years, the stronger the engine, the more difficult for a human to play against. So that I really do not believe in contempt, that would make an engine weaker by definition.

PS. But I would certainly try Stockfish_Cannibal or DiscoCheck_Maneater when available...

[carldaman]

There are engines that *may* be able to play better against humans, despite being hundred of points below SF in rating, because of their playing style. It remains an educated conjecture until established by actual further tests, of course.

Among the free engines, Thinker 5.4D Active, intended to only play against humans, is available for download somewhere. Hannibal is also quite interesting among the stronger free non-derivatives, and then there's its older cousin, called Twisted Logic. Two engines developed by Pawel Koziol, Rodent and Glass, are worth a look at. SmarThink is now free (v1.40), and very interesting, too.

Rebel by Ed Schroeder was known for its anti-GM functionality, but I'm not sure if the feature is implemented in ProDeo, the free but stronger version of Rebel.

You could also experiment with many of the free 'weaker' engines such as LittleThought (can play closed positions and flank attacks reasonably well in tests), and even Delphil, both around the 2500 CCRL Elo mark.

There are of course commercial engines, but some are hard to get these days -- Deep Fritz 10 is one of my favorite engines, especially in closed positions, and Zappa Mexico II with its crazy tweaked parameters (settings available thanks to Robert Flesher). I've had it play online in the past, and it destroyed every human it played at 5m blitz, despite playing wildly unsound chess half the time.

I fully agree with you, Carl, that many weaker engines provide a useful variety to the end user. I also agree that most of those engines are very interesting by themselves. I could also share to a point the view that some of those engines would be a bit difficult opponents for a human to play against because of their style, some preferring open positions, and other playing too many random moves so that one can not get accustomed to the style of the engine.

However, I do not believe that such engines could be in any way stronger than top engines against humans. Maybe in blitz, when speculative play often pays off, but definitely not in normal LTC, where weak moves are simply weak moves.

I have played almost all of the engines you mention (once I was a geek almost as you and Dr. D, playing every conceivable engine, but I now stick mostly to the top ) , and what I remember is:

- 17 move win against Thinker (but maybe it was the 53b version after all)
- easy wins against Fritz 10, Fritz was in the habit to get losing positions in almost every single game, but would still win the bigger share
- more successful performance against Zappa than against any of the current top engines (is Zappa a gardener's tool?)

Usually, and this is my conclusion from the 50 000 engine games I have played in the past 15 years, the stronger the engine, the more difficult for a human to play against. So that I really do not believe in contempt, that would make an engine weaker by definition.

PS. But I would certainly try Stockfish_Cannibal or DiscoCheck_Maneater when available...

Thought-provoking stuff, but I'd still like to see the paradigm tested at slower time control (non-blitz). Very surprised at your comments about Deep Fritz 10, as that is the version that defeated Kramnik in their famous match in 2006!

A couple of more points -- it seems that engines lose a lot of their edge when playing without a book; not initially maybe, but after a few games the human begins to learn the engine's opening preferences, and by trial and error he proceeds to fine tune his play whereas the engine becomes more and more predictable. Note that Fritz did play with a book in that Kramnik match. Without a book, I highly doubt it would've won.

Then, playing without contempt, the engines have no way of knowing who they're playing, and in effect 'assume' the opponent is of comparable rating, absent any way of determining the strength or type (human or engine) of the opposition. Besides contempt, some engines have anti-human settings, or they allow you to specify if you're a human before playing. For example, ZChess/Pharaon has an opponent_type parameter.

I only wish other developers would add such settings into their engines. In my opinion, they should also 'mandate' that a positive contempt setting be used for fair testing vs humans. Otherwise, by default the engine will foolishly assume it's playing a roughly equal opponent.

On this last point, our opinions probably differ. I'm not convinced SF will play weaker relative to humans with max contempt on, regardless of time control. It would likely hurt its performance vs other engines, however.

CL

[Robert Flesher]

[size=24]If you want to be taken seriously, why not have a controlled match against one of the top engines in controlled conditions.[/size] Sorry to say, but your claim that you can regularly beat these engines smells a little ... especially seeing how doggedly you defended Ivanov in his obvious cheating scandal where he was using computer help in his games. Maybe you both have something in common ... not the cheating part of course ... but you know ... the mysterious,yet incredible ability to play the game of chess at a level not yet understood by other mortal human beings. Please let us setup a match between you and any engine ... no need big hardware ... any desktop will do. Put the conditions up and I am sure a lot of people would be interested in observing how you crush these digital monsters. Kramnik, Carlsen, Anand, Aronian, Nakaumura ... all these guys will probably want a front row seat! I don't count Kasparov as he is retired and your immense talent would probably go right over his head I think it is about time that the world discovered your incredible abilities!

+1

But this will never occur, we all know that.

Amused regards,

The guy who still loses to the top engines.

My advice to you would be to play 20-30 games against engines per night instead of 5-10, and you will see with time it is not that impossible to play decently against even the top.

My controlled conditions would be airtight playing room (so no noise there), just me and the computer, you check the human for whatever cheating devices you suppose she might be using, you install some cameras that inspect the room and the behaviour of the human player, but none of the available technical equipment should not be noisy (for example loud computer fans), and then you start playing.

You can not even imagine the amount of mental energy humans are gaining if able to play in optimal quiet, undisturbed playing conditions. It is like tripling, quadrupling, etc. the power of a computer, and you know there is some effect with this in terms of strength. It is a known fact that humans use 10% of their mental energy, and so do chess players when facing each other or engines. Where do the other 90% go? I will tell you. For example, you play a tournament somewhere. 10% go to not able to concentrate deep enough because the noise level is higher than necessary. 10% go to the toll of getting accustomed to all kinds of external disturbances, for example the referee indicating something or 2 players on the next table quarreling about something. 10% go to putting up with the psychological pressure of facing your opponent: instead of thinking only about the game position, you also think if it would be possible to win the game and how difficult that would be. Another 10% go to the fight with your inner self, when you are uncertain of your own capabilitities, and that happens sometimes. And of course, another 10% go to taking the occasional look at beautiful ladies in the playing hall (if present).

When you do the maths, 90% of your mental energy is gone to doing something else different from thinking on the game itself. I.e., human chess players are using in general only about 10% of the time allocated to the game for thinking strictly on the game position. Of course, they underperform. Now, when you are not distracted by the noise, opponents, internal fights, beautiful ladies, etc., you have suddenly 90% more thinking time! That certainly makes a difference. That is why a human player to be able to successfully play against the engines, needs airtight room. But he will also have to learn to take advantage of the available more quiet, move to the background any psychological tensions, etc. Nothing surprising there, it is as if a computer had only one or multiple cores. Do you notice the difference?

For myself, I can tell you one thing: I am relatively very weak, from time to time have my psychological problems, like everyone else, so that I am unable to play good chess whenever and wherever, but, I have observed that in quiet conditions with a minimum number of disturbances (usually it is difficult to eradicate them all) I play unimaginably better than when playing in noisy conditions with many external disturbances (which unfortunately happens quite frequently).

So that my point was that I myself am unable to do well against chess opponents under normal conditions, but there certainly are human players that are much better than me in terms of the level of chess, are more psychologically stable, have more stamina, etc., and the only thing they would need in order to perform successfully against top engines is learn to use their mental energy in quiet conditions, prepare well with some anticomputer strategy, and find a way to dissipate any existing psychological tensions. It is certainly pssible to do this, I am trying (mostly unsuccessfully, as there is always something in life that will prevent you to do so) to do this occasionally, but there certainly are humans that are much more talented than me, much more psychologically stable that would be able to compete even against the top engines.

I mean no offense, but after reading what you posted. I surmise that the prose of which I enlarged above, is all I can believe to be true. However, if you were using subtle humor, lol, then I retract my previous statement.

[Lyudmil Tsvetkov]

A couple of more points -- it seems that engines lose a lot of their edge when playing without a book; not initially maybe, but after a few games the human begins to learn the engine's opening preferences, and by trial and error he proceeds to fine tune his play whereas the engine becomes more and more predictable. Note that Fritz did play with a book in that Kramnik match. Without a book, I highly doubt it would've won.
CL

Absolutely true: players like me and others who do not use books very likely gain a lot of elo based on the fact that they gradually start getting accustomed to their opponent. However, this could happen also when playing with a book, as you simply start learning how to play the lines there. Besides, I would very much like to spare myself the trouble of playing familiar openings and improvise a bit - and that is why I play without a book.

Still, I do not believe the influence of surprise is so big, it definitely plays a role, but not a primordial one. Same for contempt.