Computer cheaters

Originally posted by belgianfreak
could you run a few test games that we´re sure aren´t engines. How about david tebb v´s ironman31, or my latest game against bbarr (which I suspect will match up >90😵. Isn´t simply matching moves to an engine suseptable to a good player making good moves that a good engine would have picked? What if the first 20 moves were recognised opening, an ...[text shortened]... stem for rooting out cheats, but is using >90% engine match as foolproof as you seem to suggest?

Ok let me just state, desipte my play on RHP im not chess stupid and my OTB rating is probably in the top 5 on the site. I know and understand forced moves, opening endgame tactics and id expect everyone to match up to those moves as iv said time and time again.


Iv already dropped one of Bbarrs games somewhere, it came up with a 30% match I think. Anyway it was clear it wasnt an engine.

You dont match the 3rd move, its jut not worth it, the top 2 or none, that is the only way to be sure.

Look the chances of anyone matching an engine for 90% time and time again is the same chance of you matching my moves for 90% in most games it just doesnt happen.

Its the same chance of every GM past and present matching all of Tals move.

Its the same chance of every 1200 playing the same game again and again.

When you have a group of them all match up to engines as we do on RHP, its the same chance that as going to your local club and all of you playing the same game of chess.

Or all the players in a tournament making the same moves.

It just doesnt happen!!!! *several deep breaths* :-)

Originally posted by Grayeyesofsorrow
Tejo, you are making me laugh soooooo much. :-)

If the mods didnt want this in the forums they would delete it. Frankly I think people who have been proven to cheat should be named. Otherwise how would anyone else know not to play them, at least till the chess mods start, right now we cant name everyone since it would probably start a site wide riot. But the chess mods start soon, so fun fun fun dont you think Tejo ? :-)

and what if the accusations are wrong? What will that do to that person's reputation?

Ok back to the "No human players like a computer game after game" statement. Its never happened, no one has ever heard of it happening, its like a chess camp fire story to scare the little chess kiddies at night :-)

The brilliant thing about chess is it doesnt lie, there is always chess truth. If someone has used an engine, the moves will still be the same today as when they used it.

Any false accusations are easy to prove false, I thought Trackhead accused Boby Fischer of being an engine and went and proved that he was a human player. I dont think anyone is looking for false accusations and if they are, well they are going to end up with egg on their face :-)

In the game Feviel is playing against my fritz8 since we got off book he has only had two moves that match up to fritz, its easy to tell human from computers as I keep saying :-)

I dont think any one would disagree with me about a few chess logic things.

Peoples games dont match up to other peoples games time after time.

No human has ever been known to play like a computer game after game.

Tal is god.

If those are given, then the only thing left is to work out who is cheating and who isnt and that can be done by detection and maths.


Just incase you did disagree with me :-)

Iv given the maths about 100 times, which is there are
1 000 000 000 000 (more than the population of earth here) 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 (more than all the atoms in the universe here!!) 000 000 000 000 000

variations of chess. This isnt up for debate its a fact. Each time you play chess you play 1 game this means you have a

1 in 1 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 chance of matching another chess players game.

Well maybe a few zeros less for good moves, but not many zeros off :-)


Computers and humans play in a totally different style, this is best seen at the top levels. You know the human cant workout the variations as well as a computer, but we have developed chess instincts which is what varies our chess from game to game.
Computers crunch numbers in the same way again and again, over and over.


All this is well known and I can post links to sites for anyone who wants to look it up.

Tal is god.

So is their anyone out there who still thinks human players games will match up to fritz game after game?

If not then you'd have to agree anyone turning out stupidly high % of engine moves game after game is using an engine.

Im open to all arguements however :-)

Originally posted by Grayeyesofsorrow


Tal is god.

Im open to all arguements however :-)

I'm not going to argue, but rather look at it a different way. For my sins, probabilities have always fascinated me.

The following illustrates your point. The assumptions can fluctuate, but the the basic idea stands.

My assumptions:

A game lasts 60 moves (120 half moves)

The probability of a human player matching a computer on any move is 1 in 4 (0.25). A human player would therefore expect 15 matches throughout the game.

The chance of 15 Matches ( 25% ) or more is 1 in 2.
The chance of 18 Matches ( 30% ) or more is 1 in 6.
The chance of 24 Matches ( 40% ) or more is 1 in 90.
The chance of 30 Matches ( 50% ) or more is 1 in 5000.
The chance of 36 Matches ( 60% ) or more is 1 in 840,000.
The chance of 42 Matches ( 70% ) or more is 1 in 350,800,000.
The chance of 48 Matches ( 80% ) or more is 1 in 328,500,000,000.
The chance of 54 Matches ( 90% ) or more is 1 in 643,400,000,000,000.

95% under these assumptions is a virtual impossibility.

And this is just for a single game. Game after game is just plain cheating.

Originally posted by Gatecrasher
I'm not going to argue, but rather look at it a different way. For my sins, probabilities have always fascinated me.

The following illustrates your point. The assumptions can fluctuate, but the the basic idea stands.

My assumptions:

A game lasts 60 moves (120 half moves)

The probability of a human player matching a computer on any move is 1 ...[text shortened]... impossibility.

And this is just for a single game. Game after game is just plain cheating.

Its good to have new ideas. I hadnt considered it this way. :-)

Originally posted by Gatecrasher
I'm not going to argue, but rather look at it a different way. For my sins, probabilities have always fascinated me.

The following illustrates your point. The assumptions can fluctuate, but the the basic idea stands.

My assumptions:

A game lasts 60 moves (120 half moves)

The probability of a human player matching a computer on any move is 1 ...[text shortened]... impossibility.

And this is just for a single game. Game after game is just plain cheating.

from where do you get the assumption that 1 in 4 moves will match an engine? Surely in an established opening 100% of moves would match fo rthe first 8-20 moves. In a forced endgame the last 5-8 moves are obvious so could again have near 100% match.

I'm not saying that you can't get to a stage where it's statistically provable that someone's move match an engines, but you have to be very carefull what assumptions you start with.

Originally posted by Grayeyesofsorrow
I dont think any one would disagree with me about a few chess logic things.

Peoples games dont match up to other peoples games time after time.

No human has ever been known to play like a computer game after game.

Tal is god.

If those are given, then the only thing left is to work out who is cheating and who isnt and that can be done ...[text shortened]... % of engine moves game after game is using an engine.

Im open to all arguements however :-)

as above, I don't disagree totally. But do your figures not assume that each move is random, where they are not. If for each move I have take maybe only have between 1 and 4 viable options how does this effect the figures?

Originally posted by belgianfreak
as above, I don't disagree totally. But do your figures not assume that each move is random, where they are not. If for each move I have take maybe only have between 1 and 4 viable options how does this effect the figures?

Exactly your point. Which is why I used 1 in 4, not 1 in say, 20 or 30.

Because rather than every possible move, we want to consider only "sound" moves. In my own experience, except when there is a forced combination in progress, I would typically consider 3-8 candidates moves.

I disagree that established openings would match for 8-20 moves. Most peter out by move 8-10. Even then, most openings lines have several variations branching at every point. So why should there be a 100% match?

It is true that in end games the number of candidates would reduce.

Overall, however, a 1 in 4 match seems reasonable. Even if reduced to 1 in 3 it would produce a 15.8 billion to 1 chance of a 90% match.

But the real purpose of my post to show how improbable a very close match is for an entire game. And then to have several similarly matched games, well, you just can't get better evidence of cheating.

Originally posted by Gatecrasher
Exactly your point. Which is why I used 1 in 4, not 1 in say, 20 or 30.

Because rather than every possible move, we want to consider only "sound" moves. In my own experience, except when there is a forced combination in progress, I would typically consider 3-8 candidates moves.

I disagree that established openings would match for 8-20 moves. Mos ...[text shortened]... ve several similarly matched games, well, you just can't get better evidence of cheating.

3-8 moves candidate moves per a person is pretty much spot on. Although in OTB I dont consider more than 3, but when I sit down and take a time over a correspondence game then the amount of variations I look at is closer to 8 than 3. In most given positions the possible candidate moves are probably as high as 10-15. Yet due to peoples style of play they might only consider 5 of those 15 playable moves. I know in some positions i'll sac a pawn for an attack where others will try to build up their position more etc.

Anyway at the end of the day, chess games dont match up. :-)

Anyway at the end of the day, chess games dont match up. :-)

[/b]

I really don't know, which is why I'm asking. I'll admit that in the mid game I may consider up to 8 moves (although I can't usually see that many because I'm not that good), but of those 8 maybe 1 or 2 are clearly better? Therefore the better the player the more likely they are to pick the best of the 8, which I would assume would be the same as an engine (although I do take your point that the best move for an engines way of thinking isn't the same as the best move from a human though process in all cases).

What I'm worried about is a blanket statement along the lines that "anyone who plays a game that matches any engine for more than 90% of moves (1st or 2nd engine choice) must be cheating". This could lead to many users analysing games and then accusing someone on the basis of 1 game matching >90%.

However, a general consensus that >90% match to an engine is pretty unlikely (if this is indeed a good figure to use) then a consensus that a player matching >90% might be worth further investigation by the game moderators might be better. I would of course expect the game moderators to investigate in much more depth, and bear in mind that higher % matches might be possible. For this maybe a study into many high level games that we are sure did not involve engines (eg. GM OTB games) to show what a realistic maximum % match is, and how likely any one person is to match as such a high % several times?

If we're going to officially brand people as using engines, thus wrecking their reputation, we've got to be sure beyond doubt before we do so. I'm sure that this is possible, but let's make sure we do it right.

Originally posted by belgianfreak
I really don't know, which is why I'm asking. I'll admit that in the mid game I may consider up to 8 moves (although I can't usually see that many because I'm not that good), but of those 8 maybe 1 or 2 are clearly better? Therefo ...[text shortened]... re that this is possible, but let's make sure we do it right.

I'll try to make this as clear as I can, not something im very good at sorry.

The chances of two 40 move chess games ever been the same is near impossible. There are databases online with 10 million+ games, out of all them no two 40 move games are the same.

There are exceptions where peoples moves will match up 90% to an engine and I assure you people know this. Such games include,

Short games

Quick draws

Theory games

One of the players making a number of blunders and the other player just takes advantage of them

Combinations/Tactical shots

Forced moves

You can tell games that match up on these points on games that match up to an engine move after move for 20 positional moves where the human style of play would show.

The 8 candidate moves in each position is best seen on move 1. There is a number of things considered good and a fair few considered playable

1. e4
2. d4
3. c4
4. b4
5. Nf3
6. Nc6
7. g3
8. b3
9. f4

People would debate if all those moves are good but all are playable and each move would suit some and they love/play that move over all the others listed. The same can be said for the responces
Take the responces to e4 for example

1. e5
2. d5
3. c5
4. c3
5. e6
6. d6
7. g6
8. b3
9.Nf6
10 Nc6

You can start to see why peoples games dont match up at all.

Basicly the 90% to an engine isnt the end of it, its only an indication. I believe in trying to prove people dont use engines, which means looking for reasons why they would match the engine. If you cant provide any reason why they match the engine move after move then, you have eliminated every possiblity apart from engine use.

Grayeyes,

can u post the control games of Exy and bbarr? I've had a look for them, but couldn't find them.

They're a good indication on the matchup percentage of a NON-CHEATER.

D

Originally posted by Grayeyesofsorrow
I'll try to make this as clear as I can, not something im very good at sorry.

The chances of two 40 move chess games ever been the same is near impossible. There are databases online with 10 million+ games, out of all them no two 40 move games are the same.

There are exceptions where peoples moves will match up 90% to an engine and I assure ...[text shortened]... engine move after move then, you have eliminated every possiblity apart from engine use.

Im going to follow the moves down a little using Chessgamers online database to show how quickly no games match up. The site has a database of 300,000 games. I'll play from the top responce each time which is the most popular move.

1...c5

1 2. Nf3 46,044
2 2. Nc3 4,869
3 2. c3 3,479
4 2. d4 656
5 2. f4 545
6 2. Ne2 353
7 2. d3 317
8 2. g3 231
9 2. b3 228
10 2. c4 176
11 2. b4 109
12 2. Bc4 70
13 2. a3 20
14 2. Qe2 3
15 2. Bb5 T Christiansen vs I Sajid, 2004 0-1
16 2. Qh5 E Treger vs S Agaian, 2003 0-1
17 2. Nh3 Brian Wall vs Popovych, 2001 0-1
18 2. e5 A Mongiello vs E Yeo Min Yang, 2004 0-1
19 2. Be2 Larsen vs L Schandorff, 1997 1/2-1/2