Originally posted by greenpawn34No one game "clinched" it. A person can have a very high matchup in a single game and not be a cheat; I've had near 100% matchups in a few of my OTB ones though I don't think any of them were 20+ non-database. It's the cumulative data which is persuasive.
I'd much rather see what No1 thought was his top choice.
Or the game that clinched it for him.
No offence meant. I'll PM him. He is most likley asleep in NY Giants pyjamas.
Skeet had over 95% matchups for the first three choices in these games on SB/Z's list:
Game 7709081
Game 7696193
Game 7696189
Game 6452026
She had a 100% matchup rate in this one (with 55 non-database moves!):
Game 6646398
Originally posted by no1marauderThese games are all win and draws against high rated players. Wouldn't you expect these type of games to have higher numbers than a random selection of win draw lost?
No one game "clinched" it. A person can have a very high matchup in a single game and not be a cheat; I've had near 100% matchups in a few of my OTB ones though I don't think any of them were 20+ non-database. It's the cumulative data which is persuasive.
Skeet had over 95% matchups for the first three choices in these games on SB/Z's list ...[text shortened]... e had a 100% matchup rate in this one (with 55 non-database moves!):
Game 6646398
WELL?
Originally posted by PhlabibitI would be concerned about the inclusion of games like 6452026. White mates in 27 moves. Skeeter is white so obviously this game starts with the English opening 1. c4. The analysis gives 23 non-book moves. I would be amazed if Skeeter's database of English games stopped at move 4. Any game where skeeter is white and the game concludes quickly is likely because the opponent didn't know the English well and fell into a trap.
These games are all win and draws against high rated players. Wouldn't you expect these type of games to have higher numbers than a random selection of win draw lost?
WELL?
Originally posted by thaughbaerShe's probably got hundreds of variations of the English jotted on sticky pad paper clinging to her monitor.
I would be concerned about the inclusion of games like 6452026. White mates in 27 moves. Skeeter is white so obviously this game starts with the English opening 1. c4. The analysis gives 23 non-book moves. I would be amazed if Skeeter's database of English games stopped at move 4. Any game where skeeter is white and the game concludes quickly is likely because the opponent didn't know the English well and fell into a trap.
Originally posted by PhlabibitQuite the contrary.
These games are all win and draws against high rated players. Wouldn't you expect these type of games to have higher numbers than a random selection of win draw lost?
WELL?
The games against higher rateds (2000+RHP in the batch) which all have more than 20 non-database moves should ensure much more even positions, rather than a high rated player pummelling a much lower rated one with plenty of obvious top engine choice moves played in the process.
Random selection is no good.
If I randomly selected 20 games of skeeters against 2000+ rateds & got the results
Top 1 match = 80%
Top 2 match = 85%
Top 3 match = 90%
you'd likely say "clearly you've cherry-picked higher match rate games to try to prove guilt".
Likewise if the results from randomly chosen games didn't indicate engine use, I could be accused of deliberately picking low match rate games to protect someone.
How could I prove random selection after the event???
Games selection must be transparent & proven objective or else results lack credibility.
In November 2010 I chose skeeter's 20 then most recently completed games against 2000+ rateds which I found had at least 20 non-database moves.
Originally posted by thaughbaerAn 809 move sample size should eliminate some of these problems.
I would be concerned about the inclusion of games like 6452026. White mates in 27 moves. Skeeter is white so obviously this game starts with the English opening 1. c4. The analysis gives 23 non-book moves. I would be amazed if Skeeter's database of English games stopped at move 4. Any game where skeeter is white and the game concludes quickly is likely because the opponent didn't know the English well and fell into a trap.
You can always say "he shouldn't have chosen this game or that game" but as I say, the process is pretty much automated, so long as games meet very basic criteria.
By the way, I don't know how a player with no real-world credentials at all could possibly claim to be an expert in any opening, but that's another discussion.
Originally posted by ZygalskiCorrect.. this on my part is conjecture at why the recommendation to ban may have been discarded at the next phase of the process. Of course it may be fantasy on my part that the next phase happened at all.
An 809 move sample size should eliminate some of these problems.
You can always say "he shouldn't have chosen this game or that game" but as I say, the process is pretty much automated, so long as games meet very basic criteria.
By the way, I don't know how a player with no real-world credentials at all could possibly claim to be an expert in any opening, but that's another discussion.
Originally posted by PhlabibitGP asked for the games from the sample provided which had the highest % of matchups; that's what I gave him. As already stated, I did not reach my conclusion that she was an engine user based solely on those games but upon the cumulative evidence.
These games are all win and draws against high rated players. Wouldn't you expect these type of games to have higher numbers than a random selection of win draw lost?
WELL?
I will say that a 100% matchup in a game containing 55 non-database moves is pretty much unique among the games I have analyzed (and that's in the thousands).
Originally posted by no1marauderWhen you do your percent checking for engine use. What moves do you
GP asked for the games from the sample provided which had the highest % of matchups; that's what I gave him. As already stated, I did not reach my conclusion that she was an engine user based solely on those games but upon the cumulative evidence.
I will say that a 100% matchup in a game containing 55 non-database moves is pretty much unique among the games I have analyzed (and that's in the thousands).
check? That is, it would be misleading to check all moves, since some
moves are considered forced and nobody in his right mind would pick
a different move. I suppose you would not include established opening
moves either. So how can you tell which moves are outside of the
mormal moves that most anyone should see and those that may require
computer analysis and should be included for the study?
Originally posted by RJHindsYou're now on the list.
When you do your percent checking for engine use. What moves do you
check? That is, it would be misleading to check all moves, since some
moves are considered forced and nobody in his right mind would pick
a different move. I suppose you would not include established opening
moves either. So how can you tell which moves are outside of the
mormal mov ...[text shortened]... e should see and those that may require
computer analysis and should be included for the study?
Originally posted by no1marauderI thought there was a Weyerfritz vs Seadevil game which was 100% top 3 match for both sides over sixty odd moves?
I will say that a 100% matchup in a game containing 55 non-database moves is pretty much unique among the games I have analyzed (and that's in the thousands).