Open letter to Russ re/engine use

Originally posted by gezza
Gate

The trouble is that it is a statistical thing. With a bunch of games with known good players, picking by match up rate alone gets you precisely nowhere.

Choosing two humans from ten when you want to catch the engines is simply too much.

Maybe you can use your first pass filter of over 60% to raise suspicions, but that is all they are at that sta ...[text shortened]... High match up, as defined above, alone cannot distinguish a very good player from an engine.

Reading is Fundamental.

Gate: However, among the most popular engines, there was surprisingly little variation in average match-up rates.

There is no necessity to know exactly what engine the cheater used. It is sufficient to know that the match up rate to the engine could not be achieved by a human player. That's proven by "statistical things" (which you should have learned in high school) like standard deviation.

Originally posted by gezza
No1

What you are asking is [b]impossible to provide evidence for.

You cannot *prove* that any player did not *ever* cheat, irrespective of their match up rate.

If you think this is possible, suggest how it could be done. Stick to a practical way, rather than the theoretical supervise them 24x7, without affecting their play.


You mention speci ...[text shortened]... ry good player". You saying he was not will not diminish his ability in the slightest.

G[/b]

Top 3, Fischer had 88%

You should also take into account that theory in 2008 is much more developed than theory in 1970/71 - many moves which are considered as established DB moves now were not DB moves then. so Fisher`s (like result of other players in 1970ties) can`t be considered as too precious.

Results of modern GM`s could be more valuable material for comparing (also taking into account that there were reached against more or less equal opponents).

Originally posted by gezza
If you are tired, maybe you will stop hunting witches.

I am not aware that I suggested it was "impossible to determine if anybody on RHP uses an engine."

If you are not too tired, maybe you can point out where I suggested that. My reference to impossible was on proving a negative.

With regard to idiocy, I'll not bother to respond. It is interesting that you resort to insults, rather than addressing the content.

The old "hunting witches" nonsense. Has has been pointed out to you and others, there were no witches (at least ones who could do any harm by witchcraft), but there are plenty of engine cheats here.

Your suggestion is that no matter what match up rate a user achieves here, it is not sufficient proof that he is using an engine. That argument is just another way of saying that it can never be proven that anyone here is a cheater. That's nonsense. The content of this "argument" has been addressed.

Only an idiot would assert that people on RHP take hours to consider what move they are going to make. Only a liar would claim they actually do.

Originally posted by Korch
[b]Top 3, Fischer had 88%

You should also take into account that theory in 2008 is much more developed than theory in 1970/71 - many moves which are considered as established DB moves now were not DB moves then. so Fisher`s (like result of other players in 1970ties) can`t be considered as too precious.

Results of modern GM`s could be more valuable ma ...[text shortened]... mparing (also taking into account that there were reached against more or less equal opponents).[/b]

Gate addressed the Fischer results:

Fischer's stunning performance in the 1971 candidate matches has few precedents and we certainly haven't seen anything like it since. A 17-3-1 record against the chess elite of the day, and wiping the floor with the Russian chess establishment was a phenomenal feat.

I included that batch, because it is also the highest verifiably human match-up I ever saw in control data, and significantly higher than anything produced by pre-computer era CC players or modern GMs.

no1: The Fischer sample is a small, abnormal one. Any player who consistenly acheives those match up rates here over a large sample is a cheat.

On page 6, there are "modern" GM results. They don't vary very much from pre-engine era top CC GM results. This suggests there is an upper limit that humans can achieve in match up rates without using an engine.

Originally posted by no1marauder
Dave Tebb has made 46 moves this month. By your figures, he has used every single moment of his life + another moment considering what moves he was going to make on RHP.

You're ridiculous.

More insults No1?

I said that I remember him writing it somewhere, but that I can no longer find it. Given what I recall, and the context, I am pretty sure I recall correctly.

I did say, he said "that he used to put hours of effort ". You'll note the totally appropriate use of Past Tense in that sentence. And interestingly enough, looking at David's rating graph shows a slight dip in March. Recalling precisely what was said, and why, that fits with the context.

How many hours David does or does not invest in each move is his business. I mentioned him in addition to Wormwood, because he did say it, as I recall. So we know of 3 people who mention spending significant time for at least some moves - which is why I ask if you prefer to spend 10 hours, rather than a mere five.

You on the other hand are merely amusing - Try giving me an example of how to prove the negative I mentioned earlier - I need to laugh some more.

edit: grammar

Originally posted by gezza
Gate

The trouble is that it is a statistical thing. With a bunch of games with known good players, picking by match up rate alone gets you precisely nowhere.

Choosing two humans from ten when you want to catch the engines is simply too much.

Maybe you can use your first pass filter of over 60% to raise suspicions, but that is all they are at that sta ...[text shortened]... High match up, as defined above, alone cannot distinguish a very good player from an engine.

Well I see what you are saying. There are two points to the statistical part of your argument.

They don´t just require a match up rate that is higher than 85% (for top 3 choices) they require a match up rate that is significantly higher than it. By significantly I don´t mean ´a lot´ I mean that it is highly improbable that it came about by chance. So you have a strong effect (that kind of match-up rate is very high) and it is significant. This is possible using the techniques stated. There are technical issues, but the games mods and former games mods aren´t going to discuss that sort of detail openly.

The second point is that the former games mods have stated that they never relied soley on match up rates. I do not know what further steps the mods take, but you can think of several things they could do such as try to identify the engine, look at the games to check that the match-ups are not dominated by obvious re-captures and other easy to see moves, and look for moves which an engine would make and a human wouldn´t.

But all this is hypothetical anyway. There are at most 2 or 3 players on RHP anywhere near strong enough for this to be a real worry, and there is clear blue water between them and the players in Gatecrasher´s control data.

Originally posted by no1marauder
The old "hunting witches" nonsense. Has has been pointed out to you and others, there were no witches (at least ones who could do any harm by witchcraft), but there are plenty of engine cheats here.

Your suggestion is that no matter what match up rate a user achieves here, it is not sufficient proof that he is using an engine. That argument urs to consider what move they are going to make. Only a liar would claim they actually do.

Talking about time spend per move - it depends on opponent. Against opponents below 1800 (important part from my games in progress) I` usually play in blitz tempo. Against stronger opponents I tend to spend more time per move. Maximum time which I have ever spend per move was more than hour but it happens very rarely - in very complicated positions in the most important games.

Taking into account that average time per move in OTB games usually are 3-6 min then even spending 10 minutes per move (without time trouble, without so much stress as during OTB game) allows you to play better than in OTB.

P.S. Taking into account inconsistency of your OTB rating to your RHP rating I hope you are not going to deny that RHP allows you to reach better results than in OTB?

Originally posted by no1marauder
The old "hunting witches" nonsense. Has has been pointed out to you and others, there were no witches (at least ones who could do any harm by witchcraft), but there are plenty of engine cheats here.

Your suggestion is that no matter what match up rate a user achieves here, it is not sufficient proof that he is using an engine. That argument ...[text shortened]... urs to consider what move they are going to make. Only a liar would claim they actually do.

Ever had "witch hunter" in your location? Or maybe some variant thereof?

So now you call me a liar. Tell you what: prove to this forum that I do not consider some moves for hours.

And no, the content of the discussion has not been addressed - I maintain that using match up rate as sole criteria is not an adequate way of finding cheats.

Rather than just insulting me, why not do your homework: where is my suggestion that it is "impossible to determine if anybody on RHP uses an engine."

Originally posted by no1marauder
Gate addressed the Fischer results:

Fischer's stunning performance in the 1971 candidate matches has few precedents and we certainly haven't seen anything like it since. A 17-3-1 record against the chess elite of the day, and wiping the floor with the Russian chess establishment was a phenomenal feat.

I included that batch, because it ...[text shortened]... there is an upper limit that humans can achieve in match up rates without using an engine.

Actually if we would not include Fisher`s (which was phenomen in his time) results then average matchup in 1971 would be lower than in modern GM games. Which is not surprising taking into account that engines have affected modern chess so much....

Also top CC players in 1968-1971 tend to have slightly higher matchup than top OTB players in 1971.

Btw. Main reason why analysis of these 2 tournaments were made was to compare matchups between top OTB players and CC players in pre-computer era.

Originally posted by Korch
Talking about time spend per move - it depends on opponent. Against opponents below 1800 (in most of my games in progress) I` usually play in blitz tempo. Against stronger opponents I tend to spend more time per move. Maximum time which I have ever spend per move was more than hour but it happens very rarely - in very complicated positions in the most importa ...[text shortened]... time trouble, without so much stress as during OTB game) allows you to play better than in OTB.

Thanks Korch.
Given that you are a rather stronger player than I am, you probably spot things in your blitz tempo that take me hours.

Given that my in progress games are in tournaments, and against stronger opponents, together with the reason I play correspondence chess at all, then taking longer is not so unreasonable.

Originally posted by gezza
Ever had "witch hunter" in your location? Or maybe some variant thereof?

So now you call me a liar. Tell you what: prove to this forum that I do not consider some moves for hours.

And no, the content of the discussion has not been addressed - I maintain that using match up rate as sole criteria is not an adequate way of finding cheats.

Rather than j ...[text shortened]... here is my suggestion that it is "impossible to determine if anybody on RHP uses an engine."

Your initial claim was:

geeza: I certainly average several hours per move.

This was in support of another poster who theorized that Anand would spend five hours a move on average in a CC game.

You've now changed your tune:

geeza: prove to this forum that I do not consider some moves for hours.

Even someone like you who seems to lack even a basic knowledge of statistics should know the difference between "average" and "some".

Originally posted by Korch
Actually if we would not include Fisher`s (which was phenomen in his time) results then average matchup in 1971 would be lower than in modern GM games. Which is not surprising taking into account that engines have affected modern chess so much....

Also top CC players in 1968-1971 tend to have slightly higher matchup than top OTB players in 1971.

Btw. Ma ...[text shortened]... ts were made was to compare matchups between top OTB players and CC players in pre-computer era.

Agreed, but that is the trouble - you can't take out Fisher's results without biasing the stats. But Anand's results for top move match up were not so different either.

And top CC players have higher match up results still?

You appear to be saying that match up rates alone are not enough. Top GMs and CC players have similar rates to the computers in the tournaments.

Though yes, as I said earlier, I do understand that extremely high match up rates are an indication that a particular engine is in use.

Originally posted by no1marauder
Your initial claim was:

geeza: I certainly [b]average several hours per move.

This was in support of another poster who theorized that Anand would spend five hours a move on average in a CC game.

You've now changed your tune:

geeza: prove to this forum that I do ...[text shortened]... en a basic knowledge of statistics should know the difference between "average" and "some".[/b]

No1

I asked you to *prove* something.
I assumed that you would find it equally easy to prove the other task.
I could also ask you to *prove* something else, such as that I average several hours per move.
I didn't change what I do.

But if past history is any record, you will not prove anything. I'll just get another bunch of insults about "someone like me", and some more theories about what I do or don't know.

I vaguely recall wishing you something rather polite. Found it:

Have a nice life.

Originally posted by gezza
Agreed, but that is the trouble - you can't take out Fisher's results without biasing the stats. But Anand's results for top move match up were not so different either.

And top CC players have higher match up results still?

You appear to be saying that match up rates alone are not enough. Top GMs and CC players have similar rates to the computers in th ...[text shortened]... tand that extremely high match up rates are an indication that a particular engine is in use.

You appear to be saying that match up rates alone are not enough.

Not exactly - if suspect matchup is much higher than any human player have ever reached (like player which is the reason of why this thread was started) then in my opinion it should be enough.

Additional evidence is needed if matchup is similar or lower than best results of human players.

Originally posted by Korch
[b]You appear to be saying that match up rates alone are not enough.

Not exactly - if suspect matchup is much higher than any human player have ever reached (like player which is the reason of why this thread was started) then in my opinion it should be enough.

Additional evidence is needed if matchup is similar or lower than best results of human players.[/b]

originally posted by korch "It does not matter which moves/games exactly were played by engine. It does matter to find out that engine was used to assist at least one move (does not matter which one exactly)."


so what is correct...one move? alot of moves? one game? alot of games?

i think i see where everyone is going in this thread and all you have done is completely confuse them by saying one thing then saying another. i know i am confused, because if one move matches and that is enough then i should use that same logic and know that anyone that has won the lottery had inside information. anyone who makes that tough call or lay down in poker and is correct knows exactly what the other person is holding because they are cheating with x-ray vision. when that blind squirrel finds a nut it was because he had assistance to get there.

one move, that is great.