I was actually looking forward to the findings.
I think using a computer to catch a computer is cheat may be flawed.
The guy may be using Fritz 8 and you are looking
for him using RYBKA. To carry out a true test surely you
would have to know the engine he is using.
Also there are some good players on this site who would match a
box for perhaps 90% in a game of 30 moves.
But I am willing to be turned on this stance.
So can someone with a Rybka please do an experiment.
Put it on a reasonable setting - what ever you would put it on
if you were looking for a box cheat.
Then let it go through Morphy v Allies at the Opera.
(we know for sure a computer was not used there)
Be interesting to see the percentage.
Perhaps try it with another classic game - a Capablanca game,
he was often accussed of being a machine.
Give it an honest trial ( do not be biased ).
I think just by playing over a game you can smell a box move.
It's odd - it's deep - unexpected sting in tail possibly beyond
human calculation with no 'sign posts' in the position to guide you
towards playing that particuliar combination.
But, as I said before, I'm willing to change my mind.
There are people on here who have caught cheats.
Be interesting to let them tell us what alerted them and what
clinched it.
That way you guys who play a lot will know what to look out for.
Originally posted by greenpawn34That's surely not the best game to go through as Black will not have played anything like optimal moves.
...Then let it go through Morphy v Allies at the Opera.
(we know for sure a computer was not used there)
Be interesting to see the percentage.
Perhaps try it with another classic game - a Capablanca game,
he was often accussed of being a machine.
Give it an honest trial ( do not be biased ).
I think just by playing over a game you can smel and what
clinched it.
That way you guys who play a lot will know what to look out for.
Capablanca has the highest engine match-ups over all his games I think - someone suggest a game against highest quality opposition & then see the results in a few games.
Originally posted by greenpawn34Even more sinister - What if an engine user decides to use an engine known for its human style with tweakable personalities, like Rebel/Pro Deo? (I've been playing around with the personalities, and its quite fun.) You can even introduce a blunder factor. Yeah, it weakens it a bit, but does he really need a full-strength engine? I'd think with a setup like this, it would be much harder to identify an engine user because it would be harder to come up with clear match-up numbers.
I was actually looking forward to the findings.
I think using a computer to catch a computer is cheat may be flawed.
The guy may be using Fritz 8 and you are looking
for him using RYBKA. To carry out a true test surely you
would have to know the engine he is using.
Also there are some good players on this site who would match a
box for perh ...[text shortened]... and what
clinched it.
That way you guys who play a lot will know what to look out for.
Sort of a depressing thought, but I'm sure I'm not the only one to have thought of this.
Originally posted by Mad RookWeakening the engine isn't neccesary. If the player has the knowledge to audit the Engine personalities, he won't be adding blunders. Engine personalities can be made to follow "activity" guidelines. Teaching the engine to make more "active" or less "active" moves. Simple limitations such as this in separate parts of the game, would disprove standard analytical practice for discovery. Engine personalities can also be made to choose moves based on difficulty of positional calculation. The more complex, the more valuable the move to the engine, and once again, the engine may be choosing a less common move, that would not be found with standard techniques.
Even more sinister - What if an engine user decides to use an engine known for its human style with tweakable personalities, like Rebel/Pro Deo? (I've been playing around with the personalities, and its quite fun.) You can even introduce a blunder factor. Yeah, it weakens it a bit, but does he really need a full-strength engine? I'd think with a setup like ...[text shortened]...
Sort of a depressing thought, but I'm sure I'm not the only one to have thought of this.
OK Morphy at the opera. I didn't use Rybka (doesn't run on a Mac) but the latest HIARCS ought to be good enough. White scores 88% and black 71%. On that basis some people would be convinced Paul Morphy was an engine. However, this game is very short (17 moves) and there is a difficulty with which moves to include. My conclusion? Don't base statistical decisions on too small a data set and examine the data with a critical eye.
Windows Vista comes with a "Chess" programme. I hesitate to call it an engine or even chess as I scored 12/12 against it playing virtual blitz and then gave up it was so easy.
But lots of people will have this patzer of a programme and not knowing anything about chess will come and use it here.
What would their rating be and how do we catch this new generation of cheats?
Originally posted by Dragon FireMaybe Morphy's conglomerate opponent was the Vista engine?
Windows Vista comes with a "Chess" programme. I hesitate to call it an engine or even chess as I scored 12/12 against it playing virtual blitz and then gave up it was so easy.
But lots of people will have this patzer of a programme and not knowing anything about chess will come and use it here.
What would their rating be and how do we catch this new generation of cheats?
The Opera Game
Morphy - Karl/Isouard, Paris, 1858, 1-0
1.e4 e5 2.Nf3 d6 3.d4 Bg4 4.dxe5 Bxf3 5.Qxf3 dxe5 6.Bc4 Nf6 7.Qb3 Qe7 8.Nc3 c6 9.Bg5 b5 10.Nxb5 cxb5 11.Bxb5+ Nbd7 12.O-O-O Rd8 13.Rxd7 Rxd7 14.Rd1 Qe6 15.Bxd7- Nxd7 16.Qb8- Nxb8 17.Rd8#
This is the game in question correct?
I'll run analysis on it with Rybka if its the right game.
Originally posted by NowakowskiThat's the one. I was impressed with how early HIARCS saw black's awful fate.
The Opera Game
Morphy - Karl/Isouard, Paris, 1858, 1-0
1.e4 e5 2.Nf3 d6 3.d4 Bg4 4.dxe5 Bxf3 5.Qxf3 dxe5 6.Bc4 Nf6 7.Qb3 Qe7 8.Nc3 c6 9.Bg5 b5 10.Nxb5 cxb5 11.Bxb5+ Nbd7 12.O-O-O Rd8 13.Rxd7 Rxd7 14.Rd1 Qe6 15.Bxd7- Nxd7 16.Qb8- Nxb8 17.Rd8#
This is the game in question correct?
I'll run analysis on it with Rybka if its the right game.
Originally posted by KeplerRybka 2.1c Analysis - 45 Seconds Table 256 MB at 3.6Ghz on Core 1 (QX9775)
That's the one. I was impressed with how early HIARCS saw black's awful fate.
White - Moves 1-17
84%
White - Moves 6-17
79%
Black - Moves 1-17
63%
Black - Moves 6-17
60%
Deep Shredder 10 - 90 Seconds Table 512 MB at 3.6Ghz on Core 2 (QX97752)
White - Moves 1-17
81%
White - Moves 6-17
77%
Black - Moves 1-17
71%
Black - Moves 6-17
67%
EDIT: actually that was something else. NVM
Originally posted by NowakowskiSo Paul probably wasn't a dual core contraption running Shredder or Rybka. My G4 machine is fairly ancient but I doubt that HIARCS was available back then.
Rybka 2.1c Analysis - 45 Seconds Table 256 MB at 3.6Ghz on Core 1 (QX9775)
White - Moves 1-17
84%
White - Moves 6-17
79%
Black - Moves 1-17
63%
Black - Moves 6-17
60%
Deep Shredder 10 - 90 Seconds Table 512 MB at 3.6Ghz on Core 2 (QX97752)
White - Moves 1-17
81%
White - Moves 6-17
77%
Black - Moves 1-17
71%
Black - Moves 6-17
67%
EDIT: actually that was something else. NVM