Go back
A thought, to catch out machine cheats...

A thought, to catch out machine cheats...

Only Chess

Paul Leggett
Chess Librarian

The Stacks

Joined
21 Aug 09
Moves
114058
Clock
04 Jan 12
1 edit
Vote Up
Vote Down

It's the first game of the 1997 Kasparov-Deep Blue match. Kasparov was white.

Edit: I wonder what the matchup rates are for each one!

e

Jerusalem

Joined
20 Sep 04
Moves
37178
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Proper Knob
He's been banned. For what we don't know.
Good desition. What about Koops & Poops?

thaughbaer
Duckfinder General

223b Baker Street

Joined
25 Apr 06
Moves
33101
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Proper Knob
He's been banned. For what we don't know.
My money's on 3c....

V

Joined
21 Sep 05
Moves
27507
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Paul Leggett
It's the first game of the 1997 Kasparov-Deep Blue match. Kasparov was white.

Edit: I wonder what the matchup rates are for each one!
For Kasparov:

Move: 10. e2e3 MultiPV: 1 d19 a2a3 0.0 MultiPV: 2 d19 c2c4 -0.02 MultiPV: 3 d19 d1e1 -0.02 *** NO MATCH ***
Move: 11. d1e1 MultiPV: 1 d20 d1e1 -0.07 MultiPV: 2 d20 a2a3 -0.08 MultiPV: 3 d20 d1e2 -0.09 *** MATCH choice: 1 ***
Move: 12. a2a3 MultiPV: 1 d21 a2a3 0.05 MultiPV: 2 d21 e3e4 -0.02 MultiPV: 3 d21 d2e4 -0.06 *** MATCH choice: 1 ***
Move: 13. f3h4 MultiPV: 1 d20 e3e4 0.06 MultiPV: 2 d20 f3h4 0.05 MultiPV: 3 d20 b3b4 -0.01 *** MATCH choice: 2 ***
Move: 14. h4f3 MultiPV: 1 d20 h4f3 0.02 MultiPV: 2 d20 g3g4 -0.02 MultiPV: 3 d20 b2f6 -0.45 *** MATCH choice: 1 ***
Move: 15. e3e4 MultiPV: 1 d19 e3e4 0.05 MultiPV: 2 d19 b3b4 0.0 MultiPV: 3 d19 e1c1 -0.02 *** MATCH choice: 1 ***
Move: 16. f3h2 MultiPV: 1 d19 e1e2 0.0 MultiPV: 2 d19 c2c4 0.0 MultiPV: 3 d19 e1e3 -0.03 *** NO MATCH ***
Move: 17. e1c1 MultiPV: 1 d19 h2g4 0.18 MultiPV: 2 d19 a1c1 0.0 MultiPV: 3 d19 g2f3 -0.02 *** NO MATCH ***
Move: 18. f1e1 MultiPV: 1 d20 h2g4 0.04 MultiPV: 2 d20 f1e1 0.02 MultiPV: 3 d20 a3a4 -0.01 *** MATCH choice: 2 ***
Move: 19. d2f1 MultiPV: 1 d19 h2f1 0.02 MultiPV: 2 d19 d2f1 -0.09 MultiPV: 3 d19 g2f3 -0.12 *** MATCH choice: 2 ***
Move: 20. d3e4 MultiPV: 1 d20 d3e4 -0.06 MultiPV: 2 d20 g2e4 -0.22 MultiPV: 3 d20 f1e3 -0.52 *** MATCH choice: 1 ***
Move: 21. f1e3 MultiPV: 1 d20 f1e3 0.0 MultiPV: 2 d20 c1d2 -0.06 MultiPV: 3 d20 e1e3 -0.79 *** MATCH choice: 1 ***
Move: 22. h2f1 MultiPV: 1 d20 b2c3 -0.04 MultiPV: 2 d20 h2f1 -0.07 MultiPV: 3 d20 c1d2 -0.08 *** MATCH choice: 2 ***
Move: 23. h3g4 MultiPV: 1 d18 f1d2 0.2 MultiPV: 2 d18 h3g4 0.16 MultiPV: 3 d18 b2c3 0.02 *** MATCH choice: 2 ***
Move: 24. f2f3 MultiPV: 1 d20 b2c3 0.31 MultiPV: 2 d20 f2f3 0.29 MultiPV: 3 d20 e3g4 0.18 *** MATCH choice: 2 ***
Move: 25. f1e3 MultiPV: 1 d19 f1e3 0.26 MultiPV: 2 d19 e1e3 -0.33 MultiPV: 3 d19 g1h2 -3.67 *** MATCH choice: 1 ***
Move: 26. g1h1 MultiPV: 1 d21 a1b1 0.32 MultiPV: 2 d21 g1h2 0.27 MultiPV: 3 d21 a3a4 0.2 *** NO MATCH ***
Move: 27. e1e2 MultiPV: 1 d20 c1d2 0.16 MultiPV: 2 d20 e1e2 0.07 MultiPV: 3 d20 b2c3 0.03 *** MATCH choice: 2 ***
Move: 28. b3b4 MultiPV: 1 d21 b3b4 0.16 MultiPV: 2 d21 b3a4 -0.22 MultiPV: 3 d21 a1b1 -0.34 *** MATCH choice: 1 ***
Move: 29. e4f5 MultiPV: 1 d20 e4f5 0.29 MultiPV: 2 d20 c1e1 0.1 MultiPV: 3 d20 c1g1 -0.64 *** MATCH choice: 1 ***
Move: 30. f3f4 MultiPV: 1 d20 f3f4 0.26 MultiPV: 2 d20 g3g4 -3.02 MultiPV: 3 d20 c1f1 -3.41 *** MATCH choice: 1 ***
Move: 31. f4g5 MultiPV: 1 d20 f4g5 0.06 MultiPV: 2 d20 c1d2 -1.57 MultiPV: 3 d20 c2c4 -5.0 *** MATCH choice: 1 ***
Move: 32. g5g6 MultiPV: 1 d18 g5g6 0.06 MultiPV: 2 d18 b2e5 -0.2 MultiPV: 3 d18 g5h6 -0.89 *** MATCH choice: 1 ***
Move: 33. b2c3 MultiPV: 1 d21 b2c3 0.0 MultiPV: 2 d21 c1e1 0.0 MultiPV: 3 d21 a1b1 0.0 *** MATCH choice: 1 ***
Move: 34. c1f1 MultiPV: 1 d20 c1e1 0.27 MultiPV: 2 d20 c1f1 0.0 MultiPV: 3 d20 g2f3 -0.12 *** MATCH choice: 2 ***
Move: 35. a1f1 MultiPV: 1 d22 a1f1 0.14 MultiPV: 2 d22 e3f1 -1.23 MultiPV: 3 d22 h1h2 -0.1 *** MATCH choice: 1 ***
Move: 36. h1g1 MultiPV: 1 d23 h1g1 0.28 MultiPV: 2 d23 f1e1 0.0 MultiPV: 3 d23 g2f3 -0.16 *** MATCH choice: 1 ***
Move: 37. g2h3 MultiPV: 1 d22 g2h3 0.38 MultiPV: 2 d22 b4b5 0.0 MultiPV: 3 d22 f1b1 0.0 *** MATCH choice: 1 ***
Move: 38. g1f2 MultiPV: 1 d21 g1f2 0.56 MultiPV: 2 d21 f1a1 0.45 MultiPV: 3 d21 f1e1 0.39 *** MATCH choice: 1 ***
Move: 39. g3g4 MultiPV: 1 d19 g3g4 1.45 MultiPV: 2 d19 f1g1 0.76 MultiPV: 3 d19 e3g2 0.0 *** MATCH choice: 1 ***
Move: 40. f1g1 MultiPV: 1 d21 g4h5 2.46 MultiPV: 2 d21 f1g1 1.8 MultiPV: 3 d21 f2g3 1.16 *** MATCH choice: 2 ***
Move: 41. h3g4 MultiPV: 1 d21 h3g4 1.91 MultiPV: 2 d21 h3g2 1.67 MultiPV: 3 d21 e3g4 0.33 *** MATCH choice: 1 ***
Move: 42. e3g4 MultiPV: 1 d21 c3e5 2.81 MultiPV: 2 d21 e3g4 2.12 MultiPV: 3 d21 g6g7 0.56 *** MATCH choice: 2 ***
Move: 43. g1g4 MultiPV: 1 d21 g1g4 2.12 MultiPV: 2 d21 f2e2 -5.3 MultiPV: 3 d21 f2e1 -5.37 *** MATCH choice: 1 ***
Move: 44. f5f6 MultiPV: 1 d24 f5f6 2.54 MultiPV: 2 d24 g6g7 -0.28 MultiPV: 3 d24 f2e1 -0.46 *** MATCH choice: 1 ***
Move: 45. g6g7 MultiPV: 1 d21 g6g7 11.85 MultiPV: 2 d21 f6f7 9.06 MultiPV: 3 d21 f2e2 5.73 *** MATCH choice: 1 ***

Overall Match top 1: 22/36 = 61.1% ### Overall Match top 2: 32/36 = 88.9% ### Overall Match top 3: 32/36 = 88.9%

Starting at move 10 / engine: Houdini 1.5 / 30s per move / i7-980 CPU

S

Joined
27 Apr 07
Moves
122435
Clock
04 Jan 12
Vote Up
Vote Down

So, if that game were played here, would Kasparov's match up rate draw accusations of engine use on these boards?

Are there folks here who usually match up at rates higher than Kasparov? If so, it would be interesting to hear the explanation from such folks.

V

Joined
21 Sep 05
Moves
27507
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Paul Leggett
It's the first game of the 1997 Kasparov-Deep Blue match. Kasparov was white.

Edit: I wonder what the matchup rates are for each one!
For Deep Blue:

Move: 10... h7h6 MultiPV: 1 d20 h7h6 0.05 MultiPV: 2 d20 d8e7 0.0 MultiPV: 3 d20 d8a5 0.0 *** MATCH choice: 1 ***
Move: 11... d8a5 MultiPV: 1 d20 d8e7 0.09 MultiPV: 2 d20 h5g6 0.0 MultiPV: 3 d20 d8c7 0.0 *** NO MATCH ***
Move: 12... d6c7 MultiPV: 1 d19 d6e7 -0.03 MultiPV: 2 d19 a5c5 -0.03 MultiPV: 3 d19 h5g6 -0.03 *** NO MATCH ***
Move: 13... g7g5 MultiPV: 1 d20 g7g5 -0.02 MultiPV: 2 d20 a5c5 -0.03 MultiPV: 3 d20 a5b6 -0.13 *** MATCH choice: 1 ***
Move: 14... e6e5 MultiPV: 1 d20 e6e5 -0.06 MultiPV: 2 d20 f8e8 -0.06 MultiPV: 3 d20 a5c5 -0.08 *** MATCH choice: 1 ***
Move: 15... f8e8 MultiPV: 1 d20 f8e8 0.0 MultiPV: 2 d20 a5b6 -0.06 MultiPV: 3 d20 a8e8 -0.08 *** MATCH choice: 1 ***
Move: 16... a5b6 MultiPV: 1 d20 a5c5 0.08 MultiPV: 2 d20 a8d8 0.02 MultiPV: 3 d20 a8b8 -0.05 *** NO MATCH ***
Move: 17... a7a5 MultiPV: 1 d19 a8d8 0.01 MultiPV: 2 d19 a7a5 -0.02 MultiPV: 3 d19 h5g6 -0.03 *** MATCH choice: 2 ***
Move: 18... c7d6 MultiPV: 1 d20 c7d6 -0.01 MultiPV: 2 d20 h5g6 -0.02 MultiPV: 3 d20 b6a7 -0.1 *** MATCH choice: 1 ***
Move: 19... d5e4 MultiPV: 1 d19 d5e4 0.08 MultiPV: 2 d19 d6c5 0.08 MultiPV: 3 d19 d5d4 0.03 *** MATCH choice: 1 ***
Move: 20... d6c5 MultiPV: 1 d21 d6f8 0.11 MultiPV: 2 d21 a5a4 0.06 MultiPV: 3 d21 d6c5 0.0 *** MATCH choice: 3 ***
Move: 21... a8d8 MultiPV: 1 d20 a8d8 0.0 MultiPV: 2 d20 g8h7 0.0 MultiPV: 3 d20 a8c8 0.0 *** MATCH choice: 1 ***
Move: 22... g5g4 MultiPV: 1 d19 h5g6 0.07 MultiPV: 2 d19 g8h7 -0.02 MultiPV: 3 d19 g8f8 -0.13 *** NO MATCH ***
Move: 23... f6g4 MultiPV: 1 d20 h5g4 -0.15 MultiPV: 2 d20 f6g4 -0.27 MultiPV: 3 d20 h5g6 -0.63 *** MATCH choice: 2 ***
Move: 24... g4e3 MultiPV: 1 d20 g4e3 -0.36 MultiPV: 2 d20 g4f6 -0.43 MultiPV: 3 d20 c5e3 -0.54 *** MATCH choice: 1 ***
Move: 25... c5e7 MultiPV: 1 d20 c5e7 -0.28 MultiPV: 2 d20 g8g7 -0.3 MultiPV: 3 d20 h5g6 -0.41 *** MATCH choice: 1 ***
Move: 26... e7g5 MultiPV: 1 d20 e7g5 -0.21 MultiPV: 2 d20 h5g6 -0.39 MultiPV: 3 d20 g8h7 -0.41 *** MATCH choice: 1 ***
Move: 27... a5a4 MultiPV: 1 d21 a5a4 -0.15 MultiPV: 2 d21 d7c5 -0.2 MultiPV: 3 d21 b6a6 -0.23 *** MATCH choice: 1 ***
Move: 28... f7f5 MultiPV: 1 d20 b6a6 -0.13 MultiPV: 2 d20 d7f8 -0.18 MultiPV: 3 d20 b6c7 -0.22 *** NO MATCH ***
Move: 29... e5e4 MultiPV: 1 d19 e5e4 -0.25 MultiPV: 2 d19 b6c7 -0.48 MultiPV: 3 d19 g8f8 -0.55 *** MATCH choice: 1 ***
Move: 30... h5e2 MultiPV: 1 d18 h5e2 -0.23 MultiPV: 2 d18 g5f6 -0.91 MultiPV: 3 d18 b6c7 -2.28 *** MATCH choice: 1 ***
Move: 31... d7e5 MultiPV: 1 d18 d7e5 -0.08 MultiPV: 2 d18 h6h5 -1.82 MultiPV: 3 d18 d7f8 -2.19 *** MATCH choice: 1 ***
Move: 32... e2f3 MultiPV: 1 d18 d8d7 -0.15 MultiPV: 2 d18 e2f3 -0.24 MultiPV: 3 d18 b6a7 -0.38 *** MATCH choice: 2 ***
Move: 33... b6b5 MultiPV: 1 d20 d8d7 0.0 MultiPV: 2 d20 d8d6 -0.06 MultiPV: 3 d20 b6a7 -0.17 *** NO MATCH ***
Move: 34... b5f1 MultiPV: 1 d21 h6h5 0.02 MultiPV: 2 d21 b5f1 -0.25 MultiPV: 3 d21 e5g4 -0.27 *** MATCH choice: 2 ***
Move: 35... h6h5 MultiPV: 1 d21 g8f8 -0.16 MultiPV: 2 d21 h6h5 -0.18 MultiPV: 3 d21 f3e2 -1.27 *** MATCH choice: 2 ***
Move: 36... g8f8 MultiPV: 1 d21 g8f8 -0.16 MultiPV: 2 d21 e5g4 -0.21 MultiPV: 3 d21 e8e7 -1.02 *** MATCH choice: 1 ***
Move: 37... b7b5 MultiPV: 1 d21 b7b5 -0.15 MultiPV: 2 d21 b7b6 -0.27 MultiPV: 3 d21 d8a8 -0.4 *** MATCH choice: 1 ***
Move: 38... f8g7 MultiPV: 1 d20 d8d6 -0.17 MultiPV: 2 d20 d8d7 -0.39 MultiPV: 3 d20 e8e7 -0.83 *** NO MATCH ***
Move: 39... g7h6 MultiPV: 1 d19 g7f8 -1.87 MultiPV: 2 d19 d8d6 -1.94 MultiPV: 3 d19 g7f6 -2.02 *** NO MATCH ***
Move: 40... h5g4 MultiPV: 1 d20 e5g4 -1.91 MultiPV: 2 d20 h5g4 -2.04 MultiPV: 3 d20 h5h4 -2.47 *** MATCH choice: 2 ***
Move: 41... f3g4 MultiPV: 1 d20 e5g4 -1.92 MultiPV: 2 d20 f3g4 -3.28 MultiPV: 3 d20 c6c5 -4.56 *** MATCH choice: 2 ***
Move: 42... e5g4 MultiPV: 1 d21 e5g4 -2.28 MultiPV: 2 d21 h6g5 -6.57 MultiPV: 3 d21 h6g7 -13.47 *** MATCH choice: 1 ***
Move: 43... d8d5 MultiPV: 1 d24 d8d5 -2.59 MultiPV: 2 d24 h6h5 -3.36 MultiPV: 3 d24 e4e3 -3.36 *** MATCH choice: 1 ***
Move: 44... d5d1 MultiPV: 1 d22 e8g8 -2.55 MultiPV: 2 d22 d5f5 -2.59 MultiPV: 3 d22 h6h5 -4.86 *** NO MATCH ***

Overall Match top 1: 18/35 = 51.4% ### Overall Match top 2: 25/35 = 71.4% ### Overall Match top 3: 26/35 = 74.3%

Starting at move 10 / engine: Houdini 1.5 / 30s per move / i7-980 CPU

thaughbaer
Duckfinder General

223b Baker Street

Joined
25 Apr 06
Moves
33101
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by SmittyTime
So, if that game were played here, would Kasparov's match up rate draw accusations of engine use on these boards?

Are there folks here who usually match up at rates higher than Kasparov? If so, it would be interesting to hear the explanation from such folks.
Zygalski's method says to be suspected you need:

Top 1 match = 60%

Top 2 match = 75%

Top 3 match = 85%

so it would appear so. This just makes you a suspect though, it doesn't get you banned. I'm assuming this was Gazza in his prime and there are no verified GMs, IMs or FMs on here prime or not. We need an extra column on the player tables. Matchup rate.

V

Joined
21 Sep 05
Moves
27507
Clock
04 Jan 12

Originally posted by SmittyTime
So, if that game were played here, would Kasparov's match up rate draw accusations of engine use on these boards?
Remember that a single game is not a big enough sample. Many games need to be analysed and averaged. See Zygalski's post for full details.

thaughbaer
Duckfinder General

223b Baker Street

Joined
25 Apr 06
Moves
33101
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Varenka
Remember that a single game is not a big enough sample. Many games need to be analysed and averaged. See Zygalski's post for full details.
No.. but I doubt this is a one-off. As World Champ he should be able to sustain his highest level of play over a long stint.

Z

Joined
24 May 08
Moves
717
Clock
04 Jan 12
Vote Up
Vote Down

The match rate methodology thresholds aren't applicable to a single game with a minuscule sample size.
Kasparov has been analysed properly by several people over the last few years & obviously found to be below the thresholds, or else they'd have to be raised!

Z

Joined
24 May 08
Moves
717
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by thaughbaer
No.. but I doubt this is a one-off. As World Champ he should be able to sustain his highest level of play over a long stint.
You're making the mistake of confusing strong human play with engine-like chess.
The method tests for engine choice correlation, not sheer strength of play.

The strongest centaurs (like those over on ICCF) often have very much lower match rates than idiotic engine users that you find at the top of the player tables.

The bannings shouldn't even come into question below 65/80/90% top 1/2/3, but the idiots often return figures above 70/85/90% from a 20 game sample with over 600 non-database moves.

w
If Theres Hell Below

We're All Gonna Go!

Joined
10 Sep 05
Moves
10228
Clock
04 Jan 12
Vote Up
Vote Down

Originally posted by Paul Leggett
It's the first game of the 1997 Kasparov-Deep Blue match. Kasparov was white.

Edit: I wonder what the matchup rates are for each one!
hah, the circle is complete: I first learned chess when I saw the documentary, and began wondering what was going on in the games they showed. I had no idea, I didn't know how pieces moved or anything. so I went online and looked up the rules. then I ran into 'basic opening principles', and was into training tactics within the first two weeks if I remember correctly. it was like heroin right from the beginning. never got back to the deep blue games.

Paul Leggett
Chess Librarian

The Stacks

Joined
21 Aug 09
Moves
114058
Clock
04 Jan 12
1 edit

Originally posted by Varenka
Remember that a single game is not a big enough sample. Many games need to be analysed and averaged. See Zygalski's post for full details.
Hi Varenka,

I am just using your post as a reply because I think it is important to remember your point, as I don't want to make too much of what I say:

Since Kasparov and Deep Blue predate Houdini, it might be reasonable for us to say, in the context of this one game, that Houdini would play more like Kasparov than Deep Blue.

I have a feeling that this has much to do with the fact that modern engines are more than just brute force calculators. Kasparov intentionally chose a strategically-driven opening as his counter to a computer back in 1997, and now a strong home cpmputer running Houdini finds his moves.

I think sometimes we give all the credit to the machinery and fail to fully appreciate the programmers.

V

Joined
21 Sep 05
Moves
27507
Clock
05 Jan 12

Originally posted by Zygalski
The match rate methodology thresholds aren't applicable to a single game with a minuscule sample size.
To help emphasise this point I ran the exact same analysis again for the same single game:

Kasparov first run:
Overall Match top 1: 22/36 = 61.1% ### Overall Match top 2: 32/36 = 88.9% ### Overall Match top 3: 32/36 = 88.9%

Kasparov second run:
Overall Match top 1: 19/36 = 52.8% ### Overall Match top 2: 29/36 = 80.6% ### Overall Match top 3: 32/36 = 88.9%

The reason for the difference is that multi-processor engines are not deterministic in their analysis, i.e. you will not get the same results everytime. This is yet another reason to regard small samples as curiousity only; they can prove nothing statistically.

V

Joined
21 Sep 05
Moves
27507
Clock
05 Jan 12

Originally posted by Paul Leggett
that Houdini would play more like Kasparov than Deep Blue.
I think you have a point. I believe that modern engines are closer to Kasparov positionally than they are to Deep Blue. But tactically, I think Houdini will play more like Deep Blue. So, as you suggest, it may depend on the nature of the game in question.

Cookies help us deliver our Services. By using our Services or clicking I agree, you agree to our use of cookies. Learn More.