Chess.Com getting sued.

[Paul Cooksey]

I registered for the anti-cheating forum on Chess.com and there is indeed a thread on Justin's case. I suppose someone is updating him.

Interesting that the person who reported him did so on the basis of a statistical analysis, and from context I think there are people actively searching for statistical anomalies independently of the site management. It had not occurred to me that this kind of analysis would be anyone's idea of fun.

For what it is worth, in my opinion the case against Justin clearly not 100% proven. Some games where he played very strongly with excellent preparation. If he played blitz like this, it would be implausible. But these are correspondence games.

Ironically, I stopped playing on chess.com and moved to lichess because of the high level of computer cheating in blitz.

[JustinHorton]

I registered for the anti-cheating forum on Chess.com and there is indeed a thread on Justin's case. I suppose someone is updating him

For what it's worth, I'm aware of the existence of the forum and the existence of the thread, but no more than that.

[Roger de Coverly]

Interesting that the person who reported him did so on the basis of a statistical analysis

As described in the blog, many of the games made use of the move analysis contained in recent books. It's quite likely that the authors of these books consulted engines during the writing of the books if only to avoid recommending outright blunders.

You should not use the same methods of statistical analysis on correspondence play as for standard, rapid and Blitz. In particular "book" will go much deeper.

[NickFaulks]

It had not occurred to me that this kind of analysis would be anyone's idea of fun.

That was very naive of you, if you don't mind me saying so.

[Alex Holowczak]

I sent a PGN file of Justin's games to Ken Regan to pass through his anti-cheating software. Ken found equal parts corroboration and doubt. I have emailed a full copy of the report to Justin with Ken's permission.

[Roger de Coverly]

I sent a PGN file of Justin's games to Ken Regan to pass through his anti-cheating software. Ken found equal parts corroboration and doubt.

Does that not suggest the anti-cheating software is likely to make false accusations when, playing under the correspondence rule set, "open book" is completely legal, and for that matter the lines featured in the books may well have been engine checked at the very least?

[JustinHorton]

It might do, but it's difficult to discuss what it actually means without making reference to the report itself, which Ken has quite properly asked to be kept confidential!

I don't follow all the maths in it, I have to say, having given that subject up at the age of 16.

[Geoff Chandler]

Using the same technique/software to catch an OTB chancer is not really viable in this case.

The game load is different, correspondence players usually have numerous games on the go.

The safest way to catch someone at RHP was to look at their game load. If they were playing 30-40 games
at once and the middle game produced a high number of match up's, no errors etc.
Also time of making the moves. Recall they caught someone once whose very high standard of play
was established over 20-30 games where they played one half hour session a day.
The good 100% moves were getting sent approx every two minutes, about as long as it takes to copy and
paste in a position, get one minute engine analysis, play it, open up the next game and do it again.

If the correspondence player is only playing one or two games and they are a known good player
then one must apply the benefit of doubt. The good player, unlike in OTB games can analyse a few games
without an engine for days. Something obviously an OTB player cannot. There is bound to be a high match up.

Also It's not really opening match up's one should look for, that area in correspondence chess is a grey area.
I reach a position after 16 moves in game A where my book stopped and I took over and lose.
Two months later I get the same position in correspondence chess so I look at Game A the finished game
with an engine and it guides me. I've not looked at the active game (though in effect I have).
Or I could have ran over an engine over Game A right after I lost and noted the engine improvement.

Justin's game load was very light, there were short games. (as I said middle games is where engine play is best detectable)
Justin is a known good player. (here I add that most cheats we caught at RHP were unknowns, no forum post, no profile and
if caught they would simply join again and do it again like naughty schoolboys) The case v Justin appears to be very flimsy
I'd not entertain it.

[JustinHorton]

Justin is a known good player.

Well I guess the question is, how good? I mean while I strongly insist on common sense here, I imagine it's a lot easier to believe there might be something to this if you think my true Chess.com Daily Chess strength is around 2000 than if you think it's likely two or three hundred points higher than that.

Now one thing that occurred to me (just yesterday, though I was grasping at it here) is that we're working with a pretty unsatisfactory dataset, not just because it's small, which it is, but because it never comes to a point where you have a reliable marker of strength. It's not like the situation where somebody with a longstanding Elo of say around 2100 turns up to an Open and starts playing games in the 2600 bracket: there's no longstanding anything, we never got to that point.

So how can you properly, confidently and reliably evaluate these moves and say they're too strong for this player? You can't.

[Roger Lancaster]

Speaking as someone who doesn't play online, this whole thing confuses me. I think we're all aware of someone [no need to identify him but he's not unfamiliar to members of this forum] whose FIDE ratings - standard, rapid and blitz - are all around the 2250 mark yet advertises himself as having a 3000+ online rating. I don't remotely suggest any impropriety on his part but, in that case, we're left with the situation that online and OTB ratings can genuinely be 700+ points apart. Go figure!

[Geoff Chandler]

Hi Justin,

This was posted on The RHP site in 2009. (i tried it recall it) the criteria is 20 game.

'How to detect engine cheats - a guide'

https://www.redhotpawn.com/forum/only-c ... ide.114715

A few of your games would not even qualify to be examined.

After showing the method the poster then gave these stats where we know computers did not exist or could be of little value.

https://www.redhotpawn.com/forum/only-c ... 715/page-3
(it is the 7th post down . Squelchbelch)

First he gives Alekhine - Capablanca and Fischer-Spassky results as a rough guide line.
But more to point he covers correspondence play.

7th CC World Championship 1972-1975
Top 3 finishers

1st Place Result:
Estrin (9 games)
Top 1 Match: 153/256 (59,8% )
Top 2 Match: 191/256 (74,6% )
Top 3 Match: 209/256 (81,6% )

2nd Place Result:
Boey (13 games)
Top 1 Match: 268/449 (59,7% )
Top 2 Match: 342/449 (76,2% )
Top 3 Match: 376/449 (83,7% )

3rd Place Result:
Zagorovsky V (10 games)
Top 1 Match: 153/252 (60,7% )
Top 2 Match: 190/252 (75,4% )
Top 3 Match: 208/252 (82,5% )

10th CC World Championship 1978-1981
Top 3 finishers

1st Place Result:
Palciauskas, V (11 games)
Top 1 Match: 209/355 (58,9% )
Top 2 Match: 268/355 (75,5% )
Top 3 Match: 291/355 (82,0% )

2nd Place Result:
Morgado, J (9 games)
Top 1 Match: 131/247 (53,0% )
Top 2 Match: 189/247 (76,5% )
Top 3 Match: 208/247 (84,2% )

3rd Place Result:
Richardson, K (10 games)
Top 1 Match: 204/361 (56,5% )
Top 2 Match: 258/361 (71,5% )
Top 3 Match: 284/361 (78,7% )

In the next November 2009 post he added.

For instance, Erik on chess.com has kicked their former #1 player a few weeks after I sent in analysis for him.
The results were

ouachita (20 games)
Top 1 Match: 495/719 (68,8% )
Top 2 Match: 603/719 (83,9% )
Top 3 Match: 650/719 (90,4% )

I've no idea what method Chess,com use these days.

As the poster says:

Top 3 matchup will of course only catch blatant engine users, but as I said, these are the most annoying.
The already strong player who only uses Fritz to check the odd line or validity of a few moves in a few games
won't really be caught by any detection method. Top 3 matchup is just fine for 99% of the obvious cheats.

That last bit confirms your sentence:

"So how can you properly, confidently and reliably evaluate these moves and say they're too strong for this player? You can't."

I call these one or two move users, Stealth-Cheats and are nigh Impossible to nail down.

I think you have a valid case and are correct in asking how and why you got banned.
They have picked on the wrong man. Demand to see the evidence.
They might claim to be withholding their method as this would aid cheats in avoiding detection.

I try not play anyone over 2200 anymore. The average rating on my opponent is 1500 (I'm currently 1918 )
I trick, trap and unsound sac my way through these games. I was once examined by the above method and
came in at a dismal 40% which I found a wee bit disappointing. Am I really that bad? I must try harder.

[JustinHorton]

Demand to see the evidence.

It may surprise you to learn that I have tried that tack. (This saga has been going on some time, by the way.)

They might claim to be withholding their method as this would aid cheats in avoiding detection..

Yes they do, and my view of this is that it isn't my problem, it's not a dilemma to which, in the circumstances, I can be expected to be sympathetic.

There may be a bit more going on than this though: it seems that after a couple of fiascos in mid-decade, they decided that they would no longer seek to explain themselves, but would instead basically shut up shop unless (at least implicitly) they got sued again.

Fine, but that does mean they risk persistently refusing to examine questionable and downright bad decisions. There's quite a lot in this case, for instance (obviously this chap's a lot stronger than me) that feels very familiar.

[Joseph Conlon]

Roger - I think that 3000+ rating involves an online server where you also gain rating points based on the absolute number of games played rather than just on the results of the games.

On chess.com/lichess the blitz ratings correlate reasonably well with OTB strength (they might be out by a couple of hundred points, but the ordering is what you would expect.)

[Angus French]

Re Geoff's post of 11:17am today:
I wonder what constitutes a top 3 match? If the move played was the engine's fourth move and had the same valuation as the the third move - would that count as a match? And what if the fourth move had a valuation within a margin of the third move's? Could that count as a match?

Of course, it matters what engine is used and how deeply it assesses.

[Roger de Coverly]

Fine, but that does mean they risk persistently refusing to examine questionable and downright bad decisions.

FIDE, who were threatening briefly to go down the witchfinder path, do at least have the safeguard that the Ethics Commission will sanction false accusations and indeed have done so.

Regarding OTB cheaters though, there was a player caught in the act at a tournament this time last year and was sanctioned by the organisers who cancelled his score in the tournament. There doesn't seem to have been any comment from either the ECF or FIDE about a ban. That might be self imposed anyway as I don't think the player in question has played any graded or rated chess since.