Originally posted by no1marauderQuoting from my last post "There is a disparity in the strength of the human group and engine group; the engine group is stronger."
Contrary to yet another one of your baseless assertions, there's zero evidence to suggest that GMs are stronger than the engines mentioned.
We're not on the same page. I'm simply stating that I think one reason some human match-ups are higher is because the engines are stronger, so much stronger than the test engine, in fact, the humans have a greater match up with the test engine.
I repeat: I think that the strength of the test engine is an important factor in match-ups; if you disagree [and I hope that someone does because I like knowing how I am mistaken], then please explain why based on the above statistics.
^^I'm willing to debate this point given that you give me something to work with.
Somehow you are disagreeing with me even when we agree. Which is quite confusing to me. 🙂
Originally posted by YugaDespite the fact that there were 30 humans and only 10 engines (9 if you discount the Nokia Mobile) in the data posted, 5 engines were in the top 10 for 1st choice match-up (match-up variation is much higher for 1st choice alone), 8 engines in the top 10 for 2nd choice match-ups, 7 in the top 10 for 3rd move match-ups, 8 engines in the top 10 for 4th choice match-ups. The odds against this concentration happening randomly are extreme. While there were some notable exceptions, the computer match-ups rates do, on the whole, stick out like sore thumbs.
Quoting from my last post "There is a disparity in the strength of the human group and engine group; the engine group is stronger."
We're not on the same page. I'm simply stating that I think one reason some human match-ups are higher is because the engines are stronger, so much stronger than the test engine, in fact, the humans have a greater match up with ...[text shortened]...
Somehow you are disagreeing with me even when we agree. Which is quite confusing to me. 🙂
The strength of the test engine (or more likely the CPU strength) no doubt has an influence when pitted against against the electronic beasts of the WCCC, but I don't think there are too many 40 core clusters lurking at RHP.
Perhaps if the test engine were given more time for analysis, a clearer distinction would have resulted. The marginal benefit of calculation time reduces quite rapidly.
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.
I'm not sure if it is still done, but control data was always obtained for every engine used for cheat detection. However, among the most popular engines, there was surprisingly little variation in average match-up rates.
1 edit
I'm not sure if it is still done, but control data was always obtained for every engine used for cheat detection.
This, I think, is what I was after. In other words, when a user with moves matching suspiciously those of CheatSoft was pointed out to the game mods, a bunch of master games were matched agains CheatSoft in order to get "control data," the prior probabilities of an unassisted player achieving matches. You weren't just importing P's from your Fritz 10 comparisons, say.
This might be a daft question, but why hasn't someone put together a database of say 100 anonymous random games and include a smattering of games played by an engine, then send the database to someone to see if they can pick out which games is played by the engine?
Surely that would help answer this question?
Originally posted by BigMickIt 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).
This might be a daft question, but why hasn't someone put together a database of say 100 anonymous random games and include a smattering of games played by an engine, then send the database to someone to see if they can pick out which games is played by the engine?
Surely that would help answer this question?
Originally posted by KorchAhhhhh, of course. Yes, I failed to consider that they might not be using the engine for all their moves. Good point. I'll go back to lurking again 🙂.
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).
Originally posted by BigMickIt wouldn't work. Consider that when I analysed Berliner's games some of them had a 100% match up for top 3 engine choices. Suppose you had some high match up games in that 100 game sample. How would you know if the 90% match up was a human or an engine? It is consistently higher than likely match up that shows an individual is using an engine not a one off event. The more games that are analysed the better the accuracy of the result.
This might be a daft question, but why hasn't someone put together a database of say 100 anonymous random games and include a smattering of games played by an engine, then send the database to someone to see if they can pick out which games is played by the engine?
Surely that would help answer this question?
If you want to pick the engine games out of a random sample of games there are better methods. Asking an Italian GM if he can pick them out is one good method. A more practical method may be looking for blunders rather than match ups. Apart from doing it because you were asked to (my reason), why would you want to do this?
- Joined
- 07 Jun 05
- Moves
- 5301
Originally posted by no1marauderNo1
You provided NO evidence that players on RHP can achieve, without cheating, in a large sample the match up rates that Fischer did in a small sample of his games when he was playing at his absolute best. No, Fischer isn't a "very good player" like some players on RHP might be "very good players"; he was a super elite GM who is, at the very least, in the d ...[text shortened]... r than that in a large sample of games by non-cheating means. That is absolute nonsense.
What you are asking is 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 specific numbers: 83%. Top 3, Fischer had 88%. You say "playing at his absolute best" - so if he had even more time to think (like CC), would his match up have gone higher still?? How do you tell him apart from a computer, based on match up alone?
I still maintain, that Fischer was a "a very good player". You saying he was not will not diminish his ability in the slightest.
G
- Joined
- 07 Jun 05
- Moves
- 5301
Originally posted by greenpawn34Hi Greenpawn
Hi Gezza,
Quote:
"Fischer being "the best of all time" is a matter of opinion.
I happen to have other players in mind for the position."
Not looking for a squabble just curious. Who?
Think Fischer is certainly amongst top three along with Kasparov and Karpov.
But remembering that these players stand on the shoulders of
Alekhine, Capa, La ...[text shortened]... ames
out of chess history and we would be back in the 50's regarding chess
understanding.
I go with the same guys as you. The fact that they have a theory base which comes from previous players is clear. My comprehension of chess is that the theory base makes it *harder* to reach the top. The best players are now not only brilliant strategically and tactically, but they have to put in an enormous amount of work to learn theory - anyone though good, who does not put in the work, has no chance, as they are likely come out of the opening in a weaker position.
And yes, I base the above on reading others work - I'd have to be a Top-GM to know for certain that the above is the case.
- Joined
- 07 Jun 05
- Moves
- 5301
Originally posted by no1marauderSo do you prefer 10 ?
5 hours per move?? LMAO!
Get real.
http://www.redhotpawn.com/board/showthread.php?threadid=96203&page=1
wormwood:"the maximum time per move probably goes up to dozens of hours,"
I also recall, but can no longer find David Tebb saying that he used to put hours of effort into every move, to play at his best possible.
I certainly average several hours per move.
Originally posted by gezzaIt really depends on the move. A lot of my games here are against weaker players who make dreadful mistakes (stonger players probably think that of me). There really isn´t much point in spending large amounts of time thinking about obvious moves in those cases. If that means I walk into a trap then my opponent deserves a win for trapping me.
So do you prefer 10 ?
http://www.redhotpawn.com/board/showthread.php?threadid=96203&page=1
wormwood:"the maximum time per move probably goes up to dozens of hours,"
I also recall, but can no longer find David Tebb saying that he used to put hours of effort into every move, to play at his best possible.
I certainly average several hours per move.
Obviously, against people who aren´t going to give me an easy time I have to be more careful, but even then the number of moves where it´s really worth spending huge amounts of time is limited. Ideally I´ll look at a position several times before committing myself to a move, so that cumulatively the amount of time might add up to an hour. I found it was impossible to do this with more than about 50 games as you just lose track. With 50 you´re relying on most of the games being non-critical. With several hundred games in progress it is simply not possible to give each game more than a few minutes per move.
Not all players have a ludicrous game load, but I´ve got to agree with no1 about this. Except in a few cases players at RHP are not going to play brilliant correspondence chess because they have far too many games to be able to do so.
The games mods will be aware of how many concurrent games a player had at the time of the games they are analysing and will be able to take that into account.
1 edit
Originally posted by gezzaSee Gatecrasher's post above.
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 ry good player". You saying he was not will not diminish his ability in the slightest.
G[/b]
You're just yet another poster suggesting that's it is impossible to determine if anybody on RHP uses an engine. I tire of responding to such idiocy.
Originally posted by gezzaDave 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.
So do you prefer 10 ?
http://www.redhotpawn.com/board/showthread.php?threadid=96203&page=1
wormwood:"the maximum time per move probably goes up to dozens of hours,"
I also recall, but can no longer find David Tebb saying that he used to put hours of effort into every move, to play at his best possible.
I certainly average several hours per move.
You're ridiculous.
- Joined
- 07 Jun 05
- Moves
- 5301
Originally posted by GatecrasherGate
Despite the fact that there were 30 humans and only 10 engines (9 if you discount the Nokia Mobile) in the data posted, 5 engines were in the top 10 for 1st choice match-up (match-up variation is much higher for 1st choice alone), 8 engines in the top 10 for 2nd choice match-ups, 7 in the top 10 for 3rd move match-ups, 8 engines in the top 10 for 4th cho ...[text shortened]... otable exceptions, the computer match-ups rates do, on the whole, stick out like sore thumbs.
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 stage. You have to do more.
So you could raise the bar, to for example 80%. But then, according to numbers mentioned here, you don't catch all the engines.
By the various hints in the thread, getting a much higher match up rate (than 60😵 actually means you have identified the actual engine and time control used, which may be enough for you, but if you have to run every game through N different engines at M different time controls to find the one used, you might be "some time".
So you might be able to pick out a specific engine, but you miss any other, which throws into question the generic idea of using match-up against an engine as a test that someone is or is not using an engine.
So I repeat my conclusion 2 from a couple of pages back:
Conclusion 2: High match up, as defined above, alone cannot distinguish a very good player from an engine.
- Joined
- 07 Jun 05
- Moves
- 5301
Originally posted by no1marauderIf you are tired, maybe you will stop hunting witches.
See Gatecrasher's post above.
You're just yet another poster suggesting that's it is impossible to determine if anybody on RHP uses an engine. I tire of responding to such idiocy.
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.