Cheaters plague

82% is over 22% above the human line...

82% is a calculated fact that can be verified independently.

I'll let everybody draw his own conclusions from the presented information.

Wow - we've got a player better then Fisher or Spasky on the sight?!!?!?!!

We should feel honoured!!!!! :-o

well taht is one way to interpret it...

Updated Summary

Chess Championship League 15.1.1

Comparing player's move to engine (Rybka 2.3.2a, 30 secs, no ply limit), starting with move 6 (first five moves of both sides are ignored);

% of moves that match the engine's best move:

Johubber: 82% (231 of 291)

KARPOV: 75% (139 of 201) (1)

(1) not counting an obviously 'agreed draw' in lost position (-9.4 according to Rybka); otherwise it would be 69%

...to be continued...

http://www.littlegolem.net/jsp/game/game.jsp?gid=1212341

Strange point to agree a draw!

@Fatphil

Maybe because they can SEE further with extra time available lol :P!

http://www.littlegolem.net/jsp/game/game.jsp?gid=1212341

Strange point to agree a draw!

@Fatphil

Maybe because they can SEE further with extra time available lol :P!

Updated Summary

Chess Championship League 15.1.1

Comparing player's move to engine (Rybka 2.3.2a, 30 secs, no ply limit), starting with move 6 (first five moves of both sides are considered 'opening' and are ignored);

% of moves that match the engine's best move:

Johubber: 82% (231 of 291)

KARPOV: 75% (139 of 201) [1]

jano_wol: 81% (221 of 273)

Alexander: 72% (189 of 262)

joerg: 69% (169 of 246)

YHW: 72% (113 of 156)

[1] not counting an obviously 'agreed draw' in lost position (-9.4 according to Rybka); otherwise it would be 69%

...to be continued...

KARPOV, Alexander, YHW are all of those mahmod, MM etc. accounts, so of course cheaters. The rest seems as blatant as them.

Chess Championship League 15.1.1 (Complete)

Comparing player's move to engine (Rybka 2.3.2a, 30 secs, no ply limit), starting with move 6 (first five moves of both sides are considered 'opening' and are ignored);

% of moves that match the engine's best move:

Johubber: 82% (231 of 291)

KARPOV: 75% (139 of 201) [1]

jano_wol: 81% (221 of 273)

Alexander: 72% (189 of 262)

Henrik Sjol: 81% (124 of 154)

joerg: 69% (169 of 246)

YHW: 72% (113 of 156)

bones: 67% (28 of 42)

dimitris: never made a move

[1] not counting an obviously 'agreed draw' in lost position (-9.4 according to Rybka); otherwise it would be 69%.

So, if somar's ranges are correct, we would have a computer championship. No humans.

Comments? What now?

IS COMPUTER CHEATING CONTROL NECESSARY AT LG CHAMPIONSHIP?

TOURNAMENTS?

There was a case when I suspected cheating of my opponent. He played too fast in very difficult position. It is possible but it was suspicious because he used to come out after some quick moves. I wanted also to play quickly in order not him to go away and get the help from outside. We should cope with the problem of the computer help. I see the boy that got caught cheating is young. This is even worse. This shows that since we are young we try to use Busch devices. When the dude grows up he will be more creative and find new cheating methods. So a ban here is the most logical thing.

Certainly, we need to learn how to cope with computer cheating till it hasn’’t get the big resonance in the chess world. The most cheating in chess nowadays goes in the online chess servers. And they cannot stop it.What now?

cheating has no place in chess

Cheating in chess is not a good thing and should be battled against. You sit down behind the board to prove your abilities there, and what do you get? You can prove against a computer at home. Chess is a battle of minds. So it should be played behind the board without computers.To prevent this it is necessary to search the players before entering the playing hall. I believe that if a player uses computer assistance, it will be difficult to play with him at the board. That is why a grandmaster shouldn’’t treat the searching before the game as something humiliating.

Moreover, there should be all kinds of jammers and the things like that. Only under all these conditions we can approach the situation when cheating, not only with the help of a computer, will become impossible. It is clear that in this case the medicine could be worse that the illness. Anyway as Sasha told if a man is using the computer help it would be not possible to play with him and all our attempts to avoid the cheating would be ineffective.

Thus, a simple way of checking to make sure that the next move did not involve a major blunder would give any computer a big advantage. What would be the basis of your protest? Errors are supposed to be corrected when they are found. You might have requested that the game be backed up to the Na3 move again and restarted from there with the castling status set up correctly, but none of this suggests 'cheating'. A computer might seem intelligent, but it is not aware of what it is doing, with no sense of self or consciousness. This raises the question of whether he might have been running two computers simultaneously and using one to check the other.

I assume you have _other_ data. But since you need game scores to show that cheating occurred, I smell sour grapes and baseless accusations once again.

You know they succeed to make a really good game. Why don’t we give a chance to them some creating virtual worlds? I don’t think that they are planning to make astronomic prices for subscriptions. And it looks very democratic for their own site " They say, it will like a real world, you cannot start a management with Premiership, etc. "

Finally, I think that they have some chances to create a really good dream. And I support them.

I'm really interested in your numbers, Aganju. Are you and somar96 saying that all of those people are chess cheaters? Maybe you should send a PM to all of them to ask for a comment on this thread.

I refrained from explicitly drawing the conclusion, as it is partly based on data I couldn't (didn't) verify.

The numbers (percentages) I posted are reproducible by anyone who cares, so they are facts.

somar96 gave some thresholds that supposedly are an established standard, but I did not verify that. Assuming that's true, yes, the whole first league would consist of computers/cheaters (except dimitris, who did not show up).

I would not be surprised, as I talked personally to some players that struggle to get even close to a 2000 rating, but are in the FIDE master area in the real world (which I verified on the FIDE page).

So we have >40 active chess players on LG that are either cheaters or play stronger than a typical FIDE master.

So far the facts, draw your own conclusions.

I'm getting some doubt on the mentioned thresholds, as for example Henrik Sjol is listed as player on the FIDE pages, with a FIDE rating of 1856. This makes him #383 in Norway, certainly an accomplishment and proof of playing strength. I calculated 81% for him.

So is he a cheater? Why would somebody like him cheat? Also, it could be someone different who just chose that LG name.

After finding that information, I'm not sure if the data is conclusive in any way.

:-/

Henrik is certainly not a cheater.

I would very much like to know where that:

> 82% is over 22% above the human line...

comes from. Somar?

So far the facts, draw your own conclusions. I think that when "official" ratings are at stake (or say online prize money), there must be a very high priority to keep the playing field as "level" as possible. Chessbase I think had to retract prizes from some events prior to "freestyle" because of evidence of cheating. So they created the freestyle mode, which doesn't have this concern.

So basically a no-holds barred format ("freestyle") where it is not against the rules to use engines or even assistance from others, is probably the only way the game can be "levelled" when the stakes are high - e.g. prize money/ official ratings. This also forms part of the criticism aimed at independent servers for having a "token" anti-engine policy - because when you are playing someone very high rated and get crushed, you probably didn't have much of a chance no matter what you did. So we have >40 active chess players on LG that are either cheaters or play stronger than a typical FIDE master.In ICCF you could feel free to ramp up your Brute-force power - e.g. buy a super-computer, hire a team to investigate each candidate move, get a few GM's working for you, etc, etc.

The term for this kind of chess when Chessbase organises it is "freestyle" - it can lead to very high quality games. However it does create some distance in the relationship between a person's intuitive understanding of chess, and their rating. The rating becomes more a measure of resources, than talent. But then there is always the choice of playing Over-the-board chess under more controlled conditions.

I have to slightly disagree with you about just being about computer power. Playing on the Engine room of the Playchess server as an experiment, I once reached into the top rankings (within top 20) on a friends Quad-core machine, but found it very difficult to progress higher. Even with the latest version of Rybka, and some basic tablebase installed, many of the games were decided by the *opening*. Getting an advantage in the opening is critical - independent of the amouint of "brute force" you have at your disposal. So there is still room for creativity, and perhaps the battle shifts significantly in the engine arena for research around the opening sequences. Also of course you need to have your tablebases installed - and keep up if it goes to 6 or 7 piece. So you need masses of disc space for this.

I have another view though for independent servers which is essentially this: When you play correspondence chess, you might assume you are playing a correspondence chess "entity" in any case, because in the opening a person can quite legally refer to gigantic databases. It is not essentially their own "talent" but actually the amount of research they put the effort for doing in the opening stage - whether this is ICCF or independt server [LG].

What if the view is taken that essentially you have a broad spectrum of opponents across the rating range from say 600 to 3000, and you really are given options to just not play people say over 2500. Then those people that feel they must use every resource at their disposal in the "no holds barred" context can happily play each other. Your "view" of the server is just opponents who you have a good time playing with. I think this "restricted view" model, with as many options given to say organise your own tournaments, etc, may be better than say the model of one server which is in the extreme to publically ban people for evidence of computer abuse.

The notion of publically banning engine users would seem to act as a deterrent, but given that there are more than 100 engines on the market, and in the opening phase you could legally refer to IECG or ICCF games which were computer assisted, it means that if a person wanted to not be easily detected as an engine user, perhaps they could keep throwing in the odd blunders here and there, using different engines, etc. Just being more subtle about it. So essentially the public banning model might just encourage cheaters to be more subtle?!

The "restricted view" model also has the advantage that if you fancied your chances one day to play a very strong correspondence entity, then the opportunity is there - just invite some 2500 to a game, and make it a different kind of game called "How many moves can I last before getting mated". So there might be fun to be had from playing these super-charged resource guys... do they have to be publically banned from a server?!

Feedback welcome from all

My favorite Chess Playing Web Site is right here at LittleGolem. LG is a very highly secure WEB private data, and information is protected by international laws. This is a great excellent site for improving your chess game overall.

Love of the game is not enough to make up for lack of competence. Here I am not refering to anyone named by you; however there are several servers out there where competence simply does not exist. I think we should be redirecting our flames that way instead of doing harakiri to each other... The argument would be how to determine competence. Well, this could serve as starting point in a real discussion!

You are true. I personaly know some people on this site. They have FIDE rating 2100-2300, but on this site they have 1600 - 2050.

1. Henrik Sjol is certainly a cheater. 450 point lower than me OTB, and 500 point higher here. 950 points difference. Impossible.

old rating: 1734

newrating: 1724

rating position: 196th

So I am around 200 on LG, and possibly the best (and certainly in the top 5) OTB player. (2303 FIDE).

Wakai Yushi. You ahve ebaten and drawn many cheaters, mostly all thise mahmod accounts. You are rated 200 points higher than me. You have 85% wins. It would be really interesting to analyse you!

Ypercube, the 82% being 22% over comes from an estabilished line of 60% found after analysing 30,000+ moves of World Champions OTB and CC pre-computer era.

Somar, I don't know why you're rated only 1750 here. But it's not simply cheaters. Maybe you just make moves as you would in speed chess or something. I suppose my rating might be a little higher because I resign against cheaters as soon as I recognize them, usually before the first move, so the game isn't rated. But I still play the pool of players who do play the cheaters, so my rating should be depressed like everybody else's. But I have, without cheating, about the same rating you do here and I'm so far from being a FIDE Master I couldn't even see that goal on the horizon on beautiful clear day. Furthermore, though MiM might not get you as high, anyone who consistently used an engine should be at least around 2000 here, from what I think I've seen. So something with your accusations doesn't add up.

Maybe part of what doesn't add up is that styles of play may have become more computer-like, even where that is only as good as older moves not made by the engines; I don't know. If that's true, it would affect your percentage comparisons. A better baseline than Fischer-Spassky would be a random collection of 2300-level play from today. In any case, stop signing up for monthly cups against Mahmoud, resign when paired against him, and if your rating doesn't rise, it's probably because your'e not giving correspondence chess the attention it deserves,

I probably will start doing it after I'm back from World Championships, when probably I will time out all my games and go down to 1500. I haven't met many fair players over 1800 from whom I could get rating points. And of course some timeouts. But I doubt I will be able to go over 1850-1900.

And the cheating lines are absolutely sure to be correct.

You say And the cheating lines are absolutely sure to be correct. I don't know what you mean by "lines" in this case. Nor do I know how seriously to take your absolute assertions. But I do know that my last USCF rating was 1555. My rating here is 200 points higher. Your rating here is 600 points lower than OTB. An 800 point difference. So am I supposed to "certainly" a cheater, like Henrik Sjol, or is there a world of difference between 800 points and 900 points?

By "lines" I mean the thresholds for fair players. And with resigning against cheaters, thinking a lot etc. you can gain 1700 here. However, with the LG rating system which is much different than FIDE, the 300 points FIDE (or USCF) difference should mean maybe 150 points here, and with you and Henrik Sjol it's 500. And I have seen his wins against other cheaters - not something a human would do.

LG uses Elo, same as FIDE, right?

I have no idea why, but I'm sure KARPOV and perhaps Alexander have blipped very briefly on my radar, but without any concrete proof of any cheating. I still think there's no evidence that they're mahmods, even if they do use engines to assist their play. YHW is certainly not a mahmod sockpuppet.

It would not surprise me to see engines used by good players as a 'defence' against the more notorious engine users. The use of an engine to follow what-ifs can assist the learning process. However, it's externally indistinguishable from just cheating to win.

Why do you say YHW is not Mahmod? I'm quite sure it is.

About ratings: on LG the difference between lowest and highest rated active players isn't do big. The lowest active (playing in MC, tournaments, not just championships once per a few months) I have seen is 1300, the highest 2100, so it's 800 points. In FIDE, the lowest is 1200, and the highest is 2800. So my estimate would be:

FIDE -> LG

1200 -> 1300

1400 -> 1400

1600 -> 1500

1800 -> 1600

2000 -> 1700

2200 -> 1800

2400 -> 1900

2600 -> 2000

2800 -> 2100

I may be underrated because of timeouts and fast play, but not too much. And long thinking and not playing rated against cheates like Dvd Avins makes him a bit overrated. But it seems to work.

If you're good enough to take 10 game points out of 11 against someone, you'll be rated 400 rating points higher than them. Here or OTB.

I should be able to et 10 out of 11 against you OTB. Not sure here, we can try when I'm back.

somar96

> with the LG rating system which is much different than FIDE

Not really much of a difference. Only in the number of players.

no, the difference between lowest and highest rating

somar, your "translation" assumes that the same quality of players are present at LG as they are in the World Chess population. I'm pretty sure that's just not true. The top chess players in the world probably don't play here. Also, I would think that the "worst" players don't play here either - or only play very rarely, and thus are not "active" by your definition.

I'm not saying that the LG rating equals "real world" rating - I'm pretty sure it doesn't. However, I _AM_ pretty sure that the difference is close to a constant - that is, if a 1500 rated LG player is really a 1600 rated "real world" player, I'd expect a 2000 rated LG player to be a 2100 "real world" player. That is how the rating system is supposed to work.

Hi,

First, I haven't read the entire thread, but what I read seems like a very simplified discussion. The best players are using rybka or computational help and that should be the only reason.

I remember a player named "Time_Dilation" who said that he was playing with rybka. He was unbelievable strong and totally unbeatable. He lost only 2 games withn 2 years. He had more than 2450 points and the distance to the 2nd best player was more than 150 (as I remember). If everyone or the top ten players are using Rybka, why was/is he so unbelievable strong compared to all others? Did he use a supercomputer?!

I don't know much about chess and I didn't know what rybka was before this thread was started.

This is what I get from Wikipedia: "Rybka is a computer chess engine designed by International Master Vasik Rajlich. As of February 2010, Rybka is top-rated on chess engine rating lists and has won many official Computer Chess Tournaments including the 2007, 2008, 2009, and 2010 World Computer Chess Championships."

Questions:

1. Are you saying that all players in championship 1 (including yourself) use this program or another program?

2. How is that not cheating? I heard some interesting explanations from top LG word-games players, but I'm really interested in this one.

3. How far down the ladder is the use of software program in the LG chess world well established?

4. What is the purpose of playing using a software engine against someone else using a software engine?

5. Is playing at LG using just one's brain (in chess or any other game) considered stupid?

@MH - "That is how the rating system is supposed to work. " - yes, on the assumption that people play the same way in both environments. Weaker players may rely more on opening books here than in OTB play. Or, a little "assistance".

From my perspective - apart from one small subset of the players ratings here seem to be pretty uncorrelated to OTB ratings. That subset being me and those like me - crap and can't be bothered do obtain even the slightest edge through any means at all. Don't mock - we're happy there.

I will spent some time to analyze another league, if there are people who think it helps. What about a 3.rd league? To see what the ration of matches there is?

I assume that enforced moves like exchanges or chess or other 'zugzwang' situations are included in these stats. That could be a reason why joergs matches are also close to 70% without beeing a cheater. But how many moves without alternative are coming up within a typical game? That depends on the variant surely. So all players prefering exchange variants should automatically get a higher percentage matching value independent of the chess engine used for the analyzes. So you should maybe better analyze the endgames.

If you have the time and energy I would be interested in seeing the results of that. From talking with one of the players involved in league one I get the impression that the use of software chess engines is a lot more widespread that we believe. And, more importantly, the players who use them during a game don't see anything wrong with that. Maybe LG should become an engine-only gaming site and players who use only their brain should be banned. Maybe Ja_ma_ica was right after all.

A 400 point difference here means that you play well enough here to beat the way someone else plays here 10 times out of 11. A 400 point difference OTB means that you play well enough OTB to beat the way someone else plays OTB 10 times out of 11.

YHW, the forced moves like checkmate, caputures etc. are 2,8% of all moves.

From http://dcook.org/compgo/human_computer_team.pdf :

"The experiment began, in 2002, with a couple of questions: How much stronger

would using computers, and lots of thinking time, make the author. And secondly,

what is the strongest opening move on 9x9? We have answered the first

question, and gained some good evidence to suggest an answer for the second.

2.2 Experiment Set-up

The experiment has been conducted solely on Little Golem, a turn based server,

where players get 240 hours initially and then 36 hours per move. Multiple games

are played simultaneously. There are various types of tournaments, including a

championship where players are organized into divisions. Japanese rules, and

5.5pt komi, are used. There are currently 1,400 registered go players, with a

little over 260 of those playing in the championship. An alias ("sm9") was used

for our player, and no hint that computers were being used was given. For more

information see [7]."

Cheating in the name of science. Still cheating, I think. I don't see much difference between this and Jamaica and Rororo at Dvonn... different degrees of openness, but it all comes down to using an AI that is as strong or stronger than good human players to be near or at the top of a particular game. (Havannah, I think is a different case, as the AIs are still not that good at it...)

I agree - still cheating, and very bad behaviour! Using the LG population for experiments without our consent. That experiment was an extremely bad idea.

Unbelievable.

Ingo Althöfer is named as a professor in there, so he would have known of it too.

Maybe we can sue them? Doesn't that violate our rights, being put under psychological pressure in 'the name of science'?

Remember that other guy that killed himself in 2007 because he lost?

People like that give science a bad name, treating humans like lab rats.

Worst of all, I see very little "science" in that. There was no control. Even the software being tested changed during the period of the test.

And "computer and human together play better than either individually" is about the least revolutionary discovery in the history of mankind. Breaking news, water is wet.

So much for "if you don't like playing reversi because of the cheating why don't you play go?"

What this person has done has no excuses and I'd like to know too why Ingo Althofer said nothing if he knew about the paper.

I don't think players can sue, but the owner of the site probably could. The only problem is: is there actually any rule written anywhere at LG about cheating?

Shocked.

he wouldn't have to sue because of cheating per se, but simply because this "experiment" damages the reputation of the site.

Probably it's ethically wrong, but I don't think any explicit rule has been violated. I mean, I dislike playing "hidden" bots, for sure, but is it written anywhere that bots have to declare themselves?