Chess.com FAQs and Discussion on Cheating

Ok, let's all keep the tone nice and friendly in this thread please.

There seem to be some unusual/bizarre assumptions here about ratings. The first is that a 2500 or 2600 rating on chess.com is somehow a GM rating. It's not equivalent IMO. I don't think we have any GMs regularly and seriously competing at cc on chess.com do we? I wonder how they'd rate if they did. On ICC there are IMs with Blitz ratings in the mid 3000s! Secondly, the idea that untitled players can't be _really_ good cc players is historically absurd. There have been successful unrated/untitled cc players going way back -- way before there was good chess software.

There are a lot of assumptions being made in this discussion backed up by little more than bluster and a bit of name-calling/name-dropping.

You make a very good point here.

2500-2600 chess.com rating is basically GM strength and above, judging only after the quality of the moves. The guys I reported (Chessnut and MirceaH) would have been by far the greatest chess players in the history (OTB or correspondence - no computer assistance). If they wouldn't have cheated.

Since you talked about ICC, how come the level of correspondence chess on ICC is much lower than that on chess.com? Why all the tough players, all the GM-s who don't want to play OTB, come to chess.com?

And the main argument I have: the playing strength displayed here is well above of the best CC players before the computers. Could you provide a decent explanation for this? Yes, you could argue about better developed theory, but what about tactical errors? There were plenty of tactical errors in the past, even at the highest level, there are NO tactical errors for the best players here.

2500-2600 chess.com rating is NOT GM strength. This is a major error in your reasoning, Costelus.

If we had more GMs who had played 200+ games on this site then we might have some valid statistical evidence to show what the equivalent chess.com rating would be to FIDE 2500 or so (although there would still be lots of holes in the stats and problems of comparison), but currently we don't.

I am currently rated 2375 and am way, way, way off being IM or FM strength. Most of the games I have won recently were littered with minor errors, but I still won them and so my rating continued to go up. Had I made the same mistakes against real masters then they would have seriously punished me and my rating would have gone down.

This brings up another comparison problem - chess.com ratings change after every game, but FIDE ratings are not updated live, they average out over a much longer timeframe. My rating is very likely to plummet in the next month or two now I am playing much stronger opposition (none of whom I yet suspect of cheating, by the way). This means that lots of players will be like me and peak at 2300 or even 2400+ and then go back down to their realistic chess.com rating of 2200 or so which is not necessarily equivalent to master strength in OTB.

Ratings have context -- this is the point being made.  You can't just compare Chess.com's ratings to FIDE's as though they are the same thing.

Johny, I said very clearly: "2500-2600 chess.com rating is basically GM strength and above, judging only after the quality of the moves."

Take a game of a top player here. There is one person for instance who clearly proved he is better than Becerra. Take then a game of a low rated GM (2550-2600) let's say. Compare the moves, and the errors they make. And then come here and post the results.

We don't have a large pool of titled players here, but we have several.  The highest rated titled players have chess.com ratings that are pretty close to their FIDE ratings.  NM s tend to have chess.com ratings that are a couple hundred points higher.

It looks like up around 2600 the 'real' ratings and the chess.com ratings converge, more or less.  The competition gets a LOT tougher, apparently.

http://www.chess.com/echess/players.html

FIDE does not recognise chess.com ratings. Chess.com tournaments do not count towards GM norms, nor could they because the pool of players does not contain enough players with the necessary titles and strength for a GM norm to be scored.

Any rating system is only valid within the environment for which it was created. The conditions for gaining rating points within the chess.com environment are significantly different to that in rated FIDE tournaments.

For example:

Players begin on 1200 points in this environment. ELO systems begin at 1000 although estimated grades can also be used to establish a starting point.

Many more people timeout in chess.com than would be the case in live tournaments, and after 5 games they stop losing points. The rest of their opponents then gain many points without actually having to play out their games - this causes ratings inflation.

The rules of turn-based chess here are different to OTB chess, allowing the use of books and DBs. Some people choose not to use them, thus making them easier to beat and inflating the ratings of the 'users' of books and DBs (including myself) even more.

Time controls are different, reflecting different approaches to the game and levels of analysis (not to mention analysis boards etc.).

All of the above, and many more points besides have been discussed in many other threads, so perhaps you need to do a little homework and read these before making comparisons between the ratings systems.

No, it is not true that a rating of 2500+ in any undefined system is GM strength. One needs to define the units of measurement before taking an accurate reading.

I Agree with your post as a whole but you know she could just be really good. I have an online chess rating going around 2250. I can assure you that with national ratings I'm somewhere around No20-25 in U14 category and I know that I don't cheat(you can believe me or you can believe I cheat, I don't mind). chess.com's ratings are just too high as said by jonnyjupiter.

@Karl_

First, let me say that I am very happy that you are contributing to this thread. Now down to business.

1. You said that you don't find the stats that I posted surprising. If I read your post correctly, you implied they are as meaningless as some sports stats. The sport I follow is baseball, and yes it is littered with meaningless stats. I would argue that they're meaningless because they don't have predictive value and because they have a sample size problem. But - you should know that I first created the stat and then I measured it. So it's not a case of finding a statistic to support my claim.
2. Getting back to the fact that you don't find such stats surprising, I would argue we don't have enough data. Let's look at other chess sites to find out some comparables.
3. Going to FICS, right now I don't see any titled players logged in. Of course the top players are basically all computers. Note that FICS only has FM and higher titles recognized. Now let's look at the hbest list. There are three of them (standard, blitz, lightning). Looking at the top 10 for blitz, none of them are titled. Top 10 in standard, one is an FM (listed in notes). Top 10 in lightning, none are titled. A comment on these lists, they are pretty low ratings all things considered, I think that most titled players play at ICC, comparing between those two only.
4. Going to ICC. Currently logged in, with rating from 2600+, 33 are titled and 13 are not. Going to the best list. Top 10 in standard, 2/10 are titled. In blitz, 9/10 are titled. In Bullet, also 9/10 are titled.

Some more data. Perhaps chess.com is more similar to FICS? Perhaps it is more similar to ICC standard - where the titled players just don't seem inclined to play? (The ratings there are much lower compared to blitz, and just a bit lower than bullet. It may have to do with how titled players are compensated by the site.)

Comments welcome.

No problem! Glad to be of service.

However, I need to clarify several of your points which are, again, inaccurate.

1) I didn't state anywhere that nobody is cheating. If you check through my previous posts, I'm sure this wil become clear. I'm absolutely convinced there is some cheating. Especially valid is Costelus' point about 'part-time' cheaters, who are very difficult to detect.

2) You seem to be somewhat inconsistent in your application of the word 'facts'. In previous posts you have outlined legal requirements for what is fact and what is not and appear to be contradicting youself here.

3) Your assertion that I have a passion for ignorance does not appear to have any basis. It rather seems like a playground trick of lashing out when one's argument has been refuted.

Guys, I'm really not liking the tone of some of today's posts. If this heated debate with all the name calling persists, I may have to start deleting posts or simply lock this thread. My suggestion is that everyone step back, calm down, and feel free to continue the debate when you are settled. I can't let the mean spirited arguments continue.

The stats I quoted are all live games. ICC does have correpondence but I just learned that today. :-)  It would be interesting to compare some statistics for chess.com correspondence with ICC and/or other comparable sites.

I think that many of your points are valid/possible. For example, I agree - why would a strong titled player play here just to get a high rating? One quick comment - I don't think that even half of titled players make their living through chess. But other than that, what do they have to learn from you, or from me? I guess that's sort of rhetorical, but the point is that you assume that the answer is no, and so therefore you wouldn't even expect to see many titled players here at all. I guess the next step would be to measure similar stats for other sites and see if the theory holds up.

The point that costelus keeps making is that plenty of the players here (myself NOT included) seem to make no errors. They seem to have a standard of play well above any standard reached by non-assisted players in the past. I haven't done such research myself, so it's secondhand, but this is much more relevant (and damning) if it's true than my stats, which as you point out have a supposition that there are at least _some_ titled players here. (Hey, that's another question - how many are there here?).

I think it's important to approach the issue fairly and not to let your preconceived notions and expectations determine the course of the argument. Of course we all have our own opinions, and we all have our own beliefs about what the data will show, but it's important to do the real work with an unbiased mindset.

@Karl_ FICGS is turn-based. ICC as far as I know has live chess, but I'm not sure whether it also has turn-based. I believe that titled players play on ICC just because they know they'll find more titled players to compete with. In fact in the home page it says "Internet Chess Club: Where the Grandmasters play online chess"

Edit: When I posted ozzie's post wasn't there, sorry.

Karl_, I think you're right about chess.com being in its infancy. Erik and co. are making a real effort to attract top players and have started doing so with a couple of GMs, IMs and FMs gradually making their way up the list. The more they can attract then the more others will come to play here - virtuous circle.

Ozzie's point about non-titled players making no mistakes - I haven't researched this either, and don't have the time or inclination to do so. I think that over 3 move per day games a really strong player is likely to play several games with no obvious tactical errors. Note that this is not necessarily the same as a high match with engine choices. Some games I have successfully carried out what I thought to be a great tactical combination only to find later that the computer disagreed with the way I went about it.

Yeah it would be nice if such research could be carried out with the push of a simple button, but life is not a Staples commercial, right?

I agree with some of what you are saying here - GM strength is more than just a number, it is reflected in the choice of moves and lack of errors.

The only bit I'm not so sure about is assuming that after one game this guy is better than Becerra. He might be better at turn-based chess than GM Becerra (it is possible!), but he would need to win several games before I would believe it.

You know how these things go - you play someone, don't take them seriously enough from the start and find yourself in a dodgy position from which you can't recover. Next game you take them a lot more seriously.

Agreed - one game doesn't show much.

I certainly agree with the idea that higher levels of titled players wouldn't typically spend a TON of time here on chess.com playing turn-based, just because they have other ways in which to spend their time besides playing against people from which they cannot learn much. In other words, the training regimen for an IM and higher level of player wouldn't typically include a lot of time here, or on any of these sites for that matter.

I disagree with the idea that some players here can play so well because they just spend more time double, triple checking their moves. If this were the case, then you would expect a similar level (or higher) of play for the correspondence champions of years past. After all, nobody should be spending more time than a correspondence champion, right? But (and still according to costelus) this is not the case.

I also disagree with the idea that we somehow aren't understanding the errors that they are making, because it's not us that are doing the analysis. In this discussion, we're talking about some definition of blunder as it relates to a computer. I of course am not going through any games by hand, and neither is costelus. He's running it in an engine, and even the lesser (non-Rybka 3) engines can tell you when you've blundered.

I share your sense of optimism, that someday chess.com will produce a high titled player! And I can say I'm a part of it!