Cheating menace worsens.

The website assumes that sub- 2000 players don't cheat as if they did, their rating would be much higher. However, this naive assumption does not take into account the following type of players who are in fact the ones most likely to cheat:

1) Part-time cheaters: These are the people who believe their rating should be higher that it actually is. Hence they use engines to bring their rating up to a certain level. Once they reach this level, they play without using engines, lose, and their rating goes down. Then they start using engines to up their rating again.

2) Part - game cheaters: These are the people who start out playing normally and without engine assistance. Then they hang a piece or overlook a fork. Rather than resigning they turn on an engine and try to salvage the situation.

I have found maximum no. of such cheaters in the rating range 1350 - 1450. Because of this, a lot of good players are stuck in this band and are not able to progress ahead.

So, what to do about it? 

The first step is detection and confirmation. 

For this, I have the following suggestions for the website:

1) Run an engine at an elo level of 500 points more than the rating band which is being tested. For eg. if you are testing in the rating band 1350 - 1450, the engine elo must be 1950.

2) This engine should target only 10% of the users to avoid inconvenience. The users whom the engine plays against should be randomly selected. There should be no indication from the username that the user is actually playing against an engine. So the user would actually believe that he is playing a regular player. The engine should keep changing its username to avoid detection.

3) If a person loses to this engine, his rating points should not be deducted. 

4) The engine can be expected to win 96% of the games. If its win percentage is less than this, then it indicates that a portion of the users sampled are cheating.

5) The actual percentage of users in the sampled rating band who are cheating can be calculated using the engine's actual win percentage and the relevant formulas for the rating system used on the site. This number should be calculated and the statistics made public so that users know how many cheaters are there in each rating band.

[Please limit all discussions on cheating to the Cheating Forum group or Off-Topic Forum - mod]

That's not right...

I like your first three points, but the fourth and fifth are not all right.

Some people play/fight for ELO and saying: "This game was a test, no ELO changed" is deceiving them.

And you'd only confirm that there are people in that band that cheat without being able to individually getting them. And to check if someone is using an engine, you'd need more than one game with them (after all, this is a statistical system).

Your idea is good, but your method and goal is not completely viable.

This is my opinion of course. There may be other people that see this diferently.

Anyway, keep studying and playing and having fun, because that's what will make you evolve and keep interested in the game ^^

Okay, well I have only put out the broad skeleton for the purpose. The individual details can be tweaked as required. And yes, people can be given some bonus elo points if they are put up against an engine. I'm sure honest people would appreciate the effort and not mind the inconvenience.

The process would not only confirm that people in a rating band are cheating, but it would also give the exact percentage (estimated from the sample) of people in a rating band which are cheating. This can be easily done using the formula for rating, and basic math and statistics. Right now, we only know some people are cheating, but we don't have a number. For eg., if it turned out that the no. of cheaters in a rating band is as high as 50%, then it indicates that serious action needs to be taken. Like I said, this is only the first step - detection and confirmation. Once these numbers are available, a similar procedure can be carried out to actually nab the cheaters.

I say wrong because what if the "blunder" was intended or part of a plan?

What if it was not? As far as I know, there are no piece gambits. Even the best gambits involve loss of only one pawn.

So what if they cheat? It's just playing for fun. Who cares.

I've got 10000 games under my belt. I can hold my own at 1200 over the board, but can't break out of 500 in 5min on chess.com. The obvious general cheating tell is the divergence of quality down here. You've got someone who doesn't have an opening and hangs a queen in five moves (like you're more or less supposed to at 400) followed by someone playing theory in an incredible London mid-game @ 88% accuracy with no mistakes. Right...

i've several profiles on my watchlist of players who i think are cheating. very often they played for several months on a certain level and then suddendly within days they move up several hundred points and they are still going.

the only player in the last few months that got banned was one who very obviously cheated with accuracy over 90 every time and who reached a rating over 2000. i wonder if chess.com is really able to catch cheaters which are playing enough inaccuracies to stay below 2000.

and i also Wonder how much you really can believe them when they say they caught someone like Niemann cheating in several games. i think their algorithms are just bs.