Originally posted by VarenkaWhat an idiotic assertion, that CC chess is more engine-like because the "human" players take more time making each move. No CC chess is more like engine-like precisely because the players use engines and openly cheat. Occams razor.
Top CC play is more engine-like than top OTB
- Joined
- 26 May 02
- Moves
- 72546
3 edits
Originally posted by Dragon FireDragon Fire: Q > What moves won't a human make?
We constantly seem to be giving too much importance to match ups.
Of course they are important and of course they are a major indicator of possible engine use but I have read enough in this thread and others that taking the raw statistics proves nothing as there could be a multitude of good reasons why in any given game or any series of games a high m so hard?
OK no1maruder and eldragonfly come along and tell me I'm talking c*** again.[/b]
A > A few examples
1. A human will take the simple and safe line that wins in 15 moves rather than the dangerous tactical line that wins in 8 but where the slightest slip will result in defeat. The engine will take the quick dangerous line;
2. A human will take the line than loses quicker because it creates counter play and the possibility of cheapos (eg back rank mates) whereas an engine will delay defeat as long as possible (eg by pointlessly giving away pieces).
These are 2 examples and I am sure there are others.
I agree with most of your post. It's very important to look for typical moves that an engine would make and a human wouldn't. But it's not so easy to find them. As to your two examples:
1. Most humans would choose a safer longer win than a dangerous shorter one. However not every human player would do so. I'm sure Tal and Shirov in their careers many times would choose the flashier, more spectacular win, even though they sometimes overlooked something and suffered embarrassing losses. Not everyone is sensible and rational. 😉
2. Many humans would prefer to take the risk of losing more quickly by playing a line that gives counterplay rather than suffer for a long time without counterplay. However many humans will choose to hang on for as long as possible, either out of stubbornness or in the hope of a miracle! The weaker the player, the more likely they are to do this. But strong players sometimes don't know when to resign either. There are many GMs who are notoriously bad losers. They don't bother to set traps such as play for back rank mates, because they know that their opponents won't fall them (they wouldn't in an OTB game and are even less likely to in a correspondence game). They might also give up pieces to stave off mate for a few moves.
Identifying moves that only an engine would make is actually extremely difficult. I think it's highly subjective in that various strong and experienced players will have completely different ideas about this.
Originally posted by Dragon Firei agree with most of what you have said here, makes perfect sense.
1. Identify suspect players, eg steeply rising graphs, sudden improvements, no defeats, etc.;
2. Analyse their games automatically for high match ups (I don't know what should be considered high, perhaps 70% of 1st choice and 90% of choices 1 - 3;
3. Review candidate games for signs of typical engine moves;
4. If a single typical engine move is found that is enough for an immediate ban.
Originally posted by eldragonflyThey don't; just check CC games played in '70s, '80s and early '90s, particurlary by Blokh and other CCIMs and CCGMs
What an idiotic assertion, that CC chess is more engine-like because the "human" players take more time making each move. No CC chess is more like engine-like precisely because the players use engines and openly cheat. Occams razor.
Originally posted by gambit05Clubs have a great, unmoderated private forum functionality.
Is it within the rules of this site to post game analysis of players, but without commenting them?
Would be nice to have a thread where such analysis would be shown from the top 50 or so players. Just out of interest of course and without any accusations.
D
2 edits
Originally posted by black beetleAll right now.
They don't; just check CC games played in '70s, '80s and early '90s, particurlary by Blokh and other CCIMs and CCGMs
fairies wear boots and you gotta believe me...
you are cherry picking your examples, even you would have to admit that most OTB and "old school" CC players don't play like that.
yeah i saw it i saw it i tell you no lies..
once again is not statistically significant; these are not well known examples/players, just propaganda and could even be an exercise in disinformation. You forgot to provide an example to describe your idea of more engine-like, besides which the ultimate test would be to identify said suspected engine user and have him play a number of OTB games with a human opponent. and it would be silly to generate a database based on these players, who may well have had access to computer programs anyway.
1 edit
It seems to me this methodology of testing for a match of the top 3 moves is flawed.
For example, in the previous thread, no1marauder showed a bunch of games with 90%+ matchups in the top 3 moves. When asked about his own games, he said he only runs a blundercheck, and winds up with about a 60% matchup of the #1 move. However, if you go back to those 90%+ matchup games and only look at the #1 matchup, it's only 60-70% there as well.
Look at the game at the end of page 7 of this thread. 88% of the top 3 moves, but only 62% for the #1 move.
How many positions are there where a very strong player would find 4 separate moves equally viable?
I don't see why the test would be done this way at all, as opposed to looking at the evaluation score differences.
Originally posted by incandenzaNow, I agree. Good point.
I don't see why the test would be done this way at all, as opposed to looking at the evaluation score differences.
Initially I was interested in seeing how the "top 3" matchup would compare for various players/games. But you're right; the method itself is questionable even before we consider any results.
One factor that influences the choice of cheat detection method is the degree to which it can be automated. e.g. is it supported by software such as Fritz analysis features. But of course, this by itself doesn't make the method correct/effective. Maybe more correct methods aren't yet automated, and it would be useful if they were.
When I say "more correct methods", I don't suggest there is any foolproof approach - as there isn't any.
Originally posted by eldragonflyIn a thread splattered with lots of wild conjecture and ignorance, I finally read a post that had some logic and thought put into it. So I recced it. I don't necessarily agree with all of it, but it was intelligent and thought-provoking. So much so, even David Tebb felt the need to enter into the debate and respond.
That is correct. But you did recc varenka's silly spastic nonsensical pseudo-drivel on page 5 of this thread.
With a little more style, balance and effort, you might also produce a post worthy of being recced one day.
2 edits
Originally posted by Dragon FireYou make the same mistakes over and over again. Specifically, you seem to think that stronger players NEVER err or make non-optimum moves. But that is ridiculous as you have been already told several times.
We constantly seem to be giving too much importance to match ups.
Of course they are important and of course they are a major indicator of possible engine use but I have read enough in this thread and others that taking the raw statistics proves nothing as there could be a multitude of good reasons why in any given game or any series of games a high m so hard?
OK no1maruder and eldragonfly come along and tell me I'm talking c*** again.[/b]
Over a series of games, 90% of the moves ARE NOT obvious and/or forced. Many moves are those where reasonable players might differ. But if the "reasonable mind" of a user name is almost invariably picking a move that a engine would, it is beyond question he's using an engine to "help" him. I don't know why this concept is so difficult for you to grasp; a human being looking at a chess board with his own eyes doesn't realize that Move A is .50 centipawns better than Move B - a human being looking at his handy dandy engine analysis does.
I analyze how the Game Mods told me to. Personally, I prefer an analysis which declares a "match up" to be the engine's first choice or a close equivalence (say .10 centipawns). However much it may shock you or others though, in the case of blatant engine cheats the results come out about the same either way.
EDIT: As an example, the 50 move draw of the suspect came out this way using the Game Mods way of looking at the data:
38 non-database moves
1st choice: 25
2nd choice: 10
3rd choice: 3
Total: 38/38 or 100%
Looking at the data the way I prefer, 36 of 38 moves (95😵were either Fritz's first choice or within .10 centipawns of Fritz's first choice. One second choice was .28 different; one third choice was .20 different.
Does the difference in looking at the data really change the conclusion here?
Originally posted by incandenzaYou are talking apples and oranges.
It seems to me this methodology of testing for a match of the top 3 moves is flawed.
For example, in the previous thread, no1marauder showed a bunch of games with 90%+ matchups in the top 3 moves. When asked about his own games, he said he only runs a blundercheck, and winds up with about a 60% matchup of the #1 move. However, if you go back to those ...[text shortened]... test would be done this way at all, as opposed to looking at the evaluation score differences.
When I check my games, I considered a match up to be Fritz's 1st choice + ANY MOVE THAT WAS EVALUATED WITHIN .10 CENTIPAWNS OF FRITZ"S 1ST CHOICE. That includes move which are not only 2nd choices but might not even be in Fritz's first five choices!
Someone who 2/3 of the time in a large number of moves is picking Fritz's 1st choice and then picking Fritz's 2nd choice 20-25% is using engine assistance. Period.
A strong human player doesn't evaluate a position like Fritz does. A strong human player sometimes misses tactically strong moves. I don't know why these fairly obvious points are so hard for some people to understand.