Cheating case-my quick analysis: Opinions/analysis requested

[CRoberson]

Quick Game Investigation
By Charles Roberson 2/15/2017

Code: Select all

 Game
1. e4 c6  2. d4 d5  3. e5 Bf5  4. h4 h6  5. g4 Bh7  6. e6 fxe6  7. Bd3 c5
8. Bxh7 Rxh7  9. Qd3 Nf6  10. Nf3 Nc6  11. Qg6+ Kd7  12. Ne5+ Kc8  13. c3 cxd4
14. cxd4 Qc7  15. Bf4 Nxd4  16. Na3 Nf3+  17. Nxf3 Qxf4  18. Ke2 Kd7  19. Rac1 a6
20. Rhd1 Qxg4  21. Qc2 Ke8  22. Rd4 Qf5  23. Qc7 e5  24. Nxe5 g5  25. Nac4 e6
26. Nd6+ Bxd6  27. Qxd6 Rd8  28. Nc4 Ne4  29. Qb6 Rf7  30. Rxe4 Qxe4+  31. Ne3 d4
32. Rd1 Qf3+  White resigns 0--1

Program used for testing is Stocfish 8 for Windows
Move is considered a computer move if it is the top choice
Move is considered a computer move if is the 2nd choice and score is close to best
Human/Computer -> both human and computer move
Human/Computer F -> both human and computer move and Forced move
Scored = 0 means scored as not a computer move or it is a natural human move
Scored = 1 means scored as a computer move
Count = total moves to this point including this move scored as a computer move
Base Count = Total moves considered (opening and Forced not considered)
Move     Commments      Computer Move     Scored     Count    Base Count  Run %age
1. e4    Opening                             0         0           0           0
c6       Opening                             0         0           0           0
2. d4    Opening                             0         0           0           0
d5       Opening                             0         0           0           0
3. e5    Opening                             0         0           0           0
Bf5      Opening                             0         0           0           0
4. h4    Opening                             0         0           0           0
h6       Opening                             0         0           0           0
5. g4    Opening                             0         0           0           0
Bh7      Opening                             0         0           0           0
6. e6    Opening                             0         0           0           0
fxe6     Opening                             0         0           0           0
7. Bd3   Opening                             0         0           0           0
c5       Human/Computer      Yes             0         0           1           0
8. Bxh7  Human/Computer      Yes             0         0           1           0
Rxh7     Human/Computer  F   Yes             0         0           1           0
9. Qd3   Human/Computer      Yes             0         0           2           0               
Nf6      Human/Computer      Yes             0         0           2           0
10. Nf3  Human/Computer      Yes             0         0           3           0
Nc6      Human/Computer      Yes             0         0           3           0
11. Qg6+ Human               No              0         0           4           0
Kd7      Human/Computer  F   Yes             0         0           4           0
12. Ne5+ Human/Computer      Yes             0         0           4           0
Kc8      Questionable        Yes             0         0           5           0
13. c3   Human               No              0         0           5           0
cxd4     Questionable        Yes             0         0           6           0
14. cxd4 Human/Computer      Yes             0         0           6           0
Qc7      Human/Computer      Yes             0         0           7           0
15. Bf4  Human/Computer      Yes             0         0           7           0
Nxd4     Computer = Nxe5     No              0         0           8           0
16. Na3  Human mistake       No              0         0           8           0
Nf3+     Human/Computer      Yes             0         0           9           0
17. Nxf3 Human/Computer  F   Yes             0         0           9           0
Qxf4     Human/Computer  F   Yes             0         0           9           0
18. Ke2  Human/Computer  F   Yes             0         0           9           0
Kd7      Human/Computer      Yes             0         0          10           0
19. Rac1 Human/Computer      Yes             0         0          10           0
a6       Human/Computer      Yes             0         0          11           0
20. Rhd1 Human/Computer      Yes             0         0          11           0
Qxg4     Human/Computer      Yes             0         0          12           0
21. Qc2  Human mistake       No              0         0          12           0
Ke8      Human/Computer      Yes             0         0          13           0
22. Rd4  Human               No              0         0          13           0
Qf5      Human/Computer      Yes             0         0          14           0
23. Qc7  Human/Computer      Yes             0         0          14           0
e5       Computer            Yes             1         1          15           6.67
24. Nxe5 Human               No              0         0          15           0
g5       Computer            Yes             1         2          16          12.50
25. Nac4 Human               No              0         0          16           0
e6       Human/Computer      Yes             0         2          17          11.76
26. Nd6+ Human/Computer      Yes             0         0          18           0
Bxd6     Human/Computer  F   Yes             0         2          17          11.76
27. Qxd6 Human/Computer  F   Yes             0         0          18           0
Rd8      Human/Computer      Yes             0         2          18          11.11
28. Nc4  Human/Computer      Yes             0         0          19           0
Ne4      Human/Computer      Yes             0         2          19          11.11
29. Qb6  Human               No              0         0          20           0
Rf7      Computer            Yes             1         3          20          15.00
30. Rxe4 Human/Computer      Yes             0         0          21           0
Qxe4+    Human/Computer      Yes             0         3          21          14.28
31. Ne3  Human mistake       No              0         0          22           0
d4       Human?/Computer     Yes             0         3/4        22          13.64 - 18.18
32. Rd1  Human mistake       No              0         0          23           0
Qf3+     Human?/Computer     Yes             0         3/5        23          13.04 - 21.74
White resigns 0--1

-- White's USCF rating was (before start of game) 1995
-- Black's USCF rating was (before start of game) 1940
-- On move 31 for Black, I think many humans would have played Qf3+
-- On move 32 for Black, I think many humans would play gxh5 for a simple endgame
The above table represents a very conservative analysis. It consivatively shows that
white did not cheat at all and on multiple moves made human moves including mistakes.
OTOH, black made 3 moves that look like computer moves and 2 others that are possibly
computer moves.

From a simple nonconservative perspective:
Black only made one move that wasn't predicted by Stockfish 8 on my laptop and I was moving
through the game at a pace that may have been faster than those playing at the
40/120, SD/30;d5 Time Control. That move was 15 ... Nxd4 when the computer predicts
Nxe5 which could have been the human trying to make the computer move but making a mistake.

OTOH, White made several moves predicted by stockfish, but all look like human moves as
well. Also, white made several mistakes and several human moves that the Stockfish 8 program
would not have made.

I chose the Stockfish 8 program after a quick run through the game with Komodo 8 and
Stockfish 5. I noticed that Stockfish 5 had a greater hit rate on a quick run through.
Given that Stockfish is a very popular free program on the PC, laptops, iPhone and
Android, it made it the most likely choice for a human cheat.

[jdart]

Hard to say. I think 6. .. Bh7 is bad but it's been frequently played. A program with a good book would probably avoid it. 7. .. c5 is not in MegaBase 2017 but it is a move Stockfish likes. So maybe Stockfish with a bad book, or no book?

--Jon

[Leto]

12...Kc8 is not questionable, it's obvious that black would want to avoid placing the king on either d6 or c7 because of the Bf4 threat.

13...cxd4 doesn't look questionable to me either. c3 blocking black knight from the b4 square, unblocking it with cxd4 makes sense.

[yanquis1972]

exactly 3 moves "stood out" (went against the grain) to me using an engine (SF, then komodo 10.3) + database as benchmarks for proper play. 7...c5!? is the novelty; its not thematic & if you were intent on pushing a pawn e5 seems more sensible to me, but im far too weak for that to mean anything on my standing, i'm just throwing it out there. this move is significant particularly because its komodos first choice (depending, but its very computery -- not only unnatural, but probably bad -- a 1200 might play it, but a 1900 is odd).

only 2 more & they are back to back, which is noteworthy, but so are the moves themselves.

14...Qc7 SF & komodo label this an inaccuracy, but only with sufficient depth. engines have a bit of a fit here. when i switched K10 to mpv-4 it was tied first for some time, then when i narrowed back to spv it was the preferred move with a + for black for a bit. just did a 2nd pass w/ SF & the output is similarly confused, with Qc7 preferred to d19, then Qd6, then some time later Qc7 again but only for a few seconds. with sufficient depth this is NOT a computer move, but an old version of SF on an old phone...its still possible. it IS certainly a very natural human move (to my eyes) -- as is 15...Nxd4 , an outright blunder. in fact the only move that holds the game, per K & SF, is 15...Qb6. black is walking a tightrope, quite voluntarily so. no problem for a machine. but very tricky for a human... Nxe4 never showed on my radar -- rybka 3 dynamic is the oldest engine ive presently got installed & with one core it preferred Nxd4 for a couple seconds before moving off of it (also to Qb6). i cant imagine any top post-rybka 3 engine playing this move on any hardware if its given more than a few seconds to think on just about any hardware; the king is just wildly exposed & the tactic should be exposed at a very shallow depth...either a human mistake, or a foolish or very cocky deliberate misplay.

signs that point to computer -- odd opening; seemingly no concern for king safety that is factually very solid play; tossing out the first 8 moves, there is one blunder & one inaccuracy; the rest are machine moves. again, i'm very weak, but looking at the game with my own eyes as opposed to comparing against computer analysis, 19...a6 is first inscrutable, then with computer assistance, hella smart...he sure regained his ability to calculate after that Nxd4 slip up. and from here on out, with an ever-exposed king, he slices surgically to a crushing win.

to me i admit it does reek of a machine; the Nxd4 move is very curious indeed. but this is just really my confirmation that engines play or could play all but 1 move. i'd like to hear from tournament players who can actually put themselves in blacks seat.

[yanquis1972]

Hard to say. I think 6. .. Bh7 is bad but it's been frequently played. A program with a good book would probably avoid it. 7. .. c5 is not in MegaBase 2017 but it is a move Stockfish likes. So maybe Stockfish with a bad book, or no book?

--Jon

Bh7 is indeed a general computer choice, at lower depths anyway.

[MikeB]

Quick Game Investigation
By Charles Roberson 2/15/2017

Code: Select all

 Game
1. e4 c6  2. d4 d5  3. e5 Bf5  4. h4 h6  5. g4 Bh7  6. e6 fxe6  7. Bd3 c5
8. Bxh7 Rxh7  9. Qd3 Nf6  10. Nf3 Nc6  11. Qg6+ Kd7  12. Ne5+ Kc8  13. c3 cxd4
14. cxd4 Qc7  15. Bf4 Nxd4  16. Na3 Nf3+  17. Nxf3 Qxf4  18. Ke2 Kd7  19. Rac1 a6
20. Rhd1 Qxg4  21. Qc2 Ke8  22. Rd4 Qf5  23. Qc7 e5  24. Nxe5 g5  25. Nac4 e6
26. Nd6+ Bxd6  27. Qxd6 Rd8  28. Nc4 Ne4  29. Qb6 Rf7  30. Rxe4 Qxe4+  31. Ne3 d4
32. Rd1 Qf3+  White resigns 0--1

Program used for testing is Stocfish 8 for Windows
Move is considered a computer move if it is the top choice
Move is considered a computer move if is the 2nd choice and score is close to best
Human/Computer -> both human and computer move
Human/Computer F -> both human and computer move and Forced move
Scored = 0 means scored as not a computer move or it is a natural human move
Scored = 1 means scored as a computer move
Count = total moves to this point including this move scored as a computer move
Base Count = Total moves considered (opening and Forced not considered)
Move     Commments      Computer Move     Scored     Count    Base Count  Run %age
1. e4    Opening                             0         0           0           0
c6       Opening                             0         0           0           0
2. d4    Opening                             0         0           0           0
d5       Opening                             0         0           0           0
3. e5    Opening                             0         0           0           0
Bf5      Opening                             0         0           0           0
4. h4    Opening                             0         0           0           0
h6       Opening                             0         0           0           0
5. g4    Opening                             0         0           0           0
Bh7      Opening                             0         0           0           0
6. e6    Opening                             0         0           0           0
fxe6     Opening                             0         0           0           0
7. Bd3   Opening                             0         0           0           0
c5       Human/Computer      Yes             0         0           1           0
8. Bxh7  Human/Computer      Yes             0         0           1           0
Rxh7     Human/Computer  F   Yes             0         0           1           0
9. Qd3   Human/Computer      Yes             0         0           2           0               
Nf6      Human/Computer      Yes             0         0           2           0
10. Nf3  Human/Computer      Yes             0         0           3           0
Nc6      Human/Computer      Yes             0         0           3           0
11. Qg6+ Human               No              0         0           4           0
Kd7      Human/Computer  F   Yes             0         0           4           0
12. Ne5+ Human/Computer      Yes             0         0           4           0
Kc8      Questionable        Yes             0         0           5           0
13. c3   Human               No              0         0           5           0
cxd4     Questionable        Yes             0         0           6           0
14. cxd4 Human/Computer      Yes             0         0           6           0
Qc7      Human/Computer      Yes             0         0           7           0
15. Bf4  Human/Computer      Yes             0         0           7           0
Nxd4     Computer = Nxe5     No              0         0           8           0
16. Na3  Human mistake       No              0         0           8           0
Nf3+     Human/Computer      Yes             0         0           9           0
17. Nxf3 Human/Computer  F   Yes             0         0           9           0
Qxf4     Human/Computer  F   Yes             0         0           9           0
18. Ke2  Human/Computer  F   Yes             0         0           9           0
Kd7      Human/Computer      Yes             0         0          10           0
19. Rac1 Human/Computer      Yes             0         0          10           0
a6       Human/Computer      Yes             0         0          11           0
20. Rhd1 Human/Computer      Yes             0         0          11           0
Qxg4     Human/Computer      Yes             0         0          12           0
21. Qc2  Human mistake       No              0         0          12           0
Ke8      Human/Computer      Yes             0         0          13           0
22. Rd4  Human               No              0         0          13           0
Qf5      Human/Computer      Yes             0         0          14           0
23. Qc7  Human/Computer      Yes             0         0          14           0
e5       Computer            Yes             1         1          15           6.67
24. Nxe5 Human               No              0         0          15           0
g5       Computer            Yes             1         2          16          12.50
25. Nac4 Human               No              0         0          16           0
e6       Human/Computer      Yes             0         2          17          11.76
26. Nd6+ Human/Computer      Yes             0         0          18           0
Bxd6     Human/Computer  F   Yes             0         2          17          11.76
27. Qxd6 Human/Computer  F   Yes             0         0          18           0
Rd8      Human/Computer      Yes             0         2          18          11.11
28. Nc4  Human/Computer      Yes             0         0          19           0
Ne4      Human/Computer      Yes             0         2          19          11.11
29. Qb6  Human               No              0         0          20           0
Rf7      Computer            Yes             1         3          20          15.00
30. Rxe4 Human/Computer      Yes             0         0          21           0
Qxe4+    Human/Computer      Yes             0         3          21          14.28
31. Ne3  Human mistake       No              0         0          22           0
d4       Human?/Computer     Yes             0         3/4        22          13.64 - 18.18
32. Rd1  Human mistake       No              0         0          23           0
Qf3+     Human?/Computer     Yes             0         3/5        23          13.04 - 21.74
White resigns 0--1

-- White's USCF rating was (before start of game) 1995
-- Black's USCF rating was (before start of game) 1940
-- On move 31 for Black, I think many humans would have played Qf3+
-- On move 32 for Black, I think many humans would play gxh5 for a simple endgame
The above table represents a very conservative analysis. It consivatively shows that
white did not cheat at all and on multiple moves made human moves including mistakes.
OTOH, black made 3 moves that look like computer moves and 2 others that are possibly
computer moves.

From a simple nonconservative perspective:
Black only made one move that wasn't predicted by Stockfish 8 on my laptop and I was moving
through the game at a pace that may have been faster than those playing at the
40/120, SD/30;d5 Time Control. That move was 15 ... Nxd4 when the computer predicts
Nxe5 which could have been the human trying to make the computer move but making a mistake.

OTOH, White made several moves predicted by stockfish, but all look like human moves as
well. Also, white made several mistakes and several human moves that the Stockfish 8 program
would not have made.

I chose the Stockfish 8 program after a quick run through the game with Komodo 8 and
Stockfish 5. I noticed that Stockfish 5 had a greater hit rate on a quick run through.
Given that Stockfish is a very popular free program on the PC, laptops, iPhone and
Android, it made it the most likely choice for a human cheat.

From Move 16 on , it appears to be all computer. Up until that point, it could been human, but after making the human mistake on 15. Nxd4 he went to stockfish ( my version of SF he matched stockfish for every move after that - that's no coincidence). He's probably doing a human/comp, using the computer to keep his rating near 1900. who knows - why I avoid playing online -if I'm gonna play a computer I don't need to do it online - if you re going to play online just accept the fact that people are cheating and move on , it can eat you up and spit you out.

[yanquis1972]

i was under the impression it was an OTB tournament game, but i'm curious

[CRoberson]

Yes, it was an OTB tournament game.

[MikeB]

i was under the impression it was an OTB tournament game, but i'm curious

yes, I thought it was online , but with uscf ratings it must be presumed otb, how does one cheat otb these days - aren't cell phones restricted ?

[yanquis1972]

i'm quite sure, but if someone really really wants to its possible. possible even to be a complete lunatic about it & not -- technically -- ever have it proven against you.

last (quite) infamous case i recall was this but im sure thereve been incidents since -- http://www.independent.co.uk/news/world ... 20186.html