Major Proposed Change: Ratings Cap Above 2400 for Unverified Players

again, thank you all for your input. let me clarify a few points:

1. we know how many people this will affect in each rating pool. 

2. we know bullet is inflated, and we're working on figuring out how to fix that (either a one-time 200 point CHOP or slower deflation). the cap doesn't make sense for the current bullet pool. 

3. we care TREMENDOUSLY about Chess.com and every member (well, not about cheaters). we are working this out to try and address all angles, needs, and ideas. we can't please everyone, but we're trying :) if we didn't care we'd just let the cheating go on and not go through all of this turmoil. 

4. this idea is on hold until we can figure out the bullet rating pool and how to address the extremely small number of 2400+ players who are not titled and cannot afford a membership. (ideas include: personal evaluation, help them find a way to earn $29 (if they are 2400+ they could become a coach and earn the $? http://www.chess.com/coaches , or write an article and get paid, or some other kind of service to the website to earn a 1 year gold membership, or get photocopy of ID along with OTB rating even if it isn't "titled"). the point is there needs to be a barrier of SOME kind. 

5. we are also working toward being able to detect cheating earlier and have found some promising ideas. 

6. noted that our graphs don't reflect the same data as the list (bug entered into the sytem to be fixed - thanks guys!)

anyway, thank you all for your helpful feedback! 

@erik: Not that I'm anywhere near the rating in question (2400), but I am not even a member of any chess organization (none exist in my area), and am extremely unlikely to ever have an official rating of any kind. So I would never be able to prove by external means that I'm legit. I am probably not alone in this situation.

...um.

Mouse's saying that you would need to pay to see people's real ratings.

If you detonated a 10 megaton bomb in the Pacific Ocean it wouldn't have a measurable effect, but that would be little consolation to anyone at ground zero.  The further one's rating is from 2400 the less would be the ripple effect, but at the rarefied atmosphere of 2400+ it probably would have a measurable effect.  Certainly it would have a measurable  effect on any individual (whose strenght is 2400) who has to play a 2700 strength player capped at 2400.  It will be tougher for premium members to break the 2400 barier.

I like the "show me some ID and your rating" idea. Wouldn't work for guys like me, of course, but it should work for a lot of people.

Bumping for bookmark purposes;

I think this change is a terrific initiative, capping at 2400 chess.com unless master grade which would indicate 2000+ otb for the introductory titles candidate / national.

This will increase the credibility of chess.com and net chess generally.

Maybe in future it will be applied to all here.

>:)

May I ask a very simple question? You care tremendously about cheating, but I know one example of a player who has been cheating here for two years. Obvious cheating, Rybka for each and every move. Of course, high rating, always around the top 10 players. For two years, I repeat. He is premium member, so the measure discussed here does not affect him in any way.

Can you explain the paradox?

Actually, you are still in database (in Game Explorer). Just check the Explore option on the game.

No - they are comparable to the skills of other bullet players rated 2500 on chess.com

I wish this wasn't true. :P My goal was to hit 2500, but now I can't get there. I guess people will quit after getting to 2400. But then again, it is hard to get there. I'm far... very far. :) I think that if you play like 1000+ games, you should be able to go as high as the sky.

that's me trying to keep this friendlier, RR+LN. thanks in advance.

Regrettably, I am not aware of a single example. Please reference your post.

>:)

I think the 2400 rating cap for non-members is a great idea. In fact, I say we use the rating cap for non-members, and INFLATE all of the ratings.

That way, the 2400-rated players won't feel special (since more people will reach 2400), and the non-cheaters will be more motivated to remove the rating cap by purchasing a lovely chess.com membership.

Win-win, right? You can thank me later.

After 268 posts, I think I got one major point down. Chess.com is not wanting to make this policy change to get rich. A handful of entry level gold membership fees collected won't even pay for one of their own kid's college tuition for one semester. It might cover the expenses of textbooks and some protein bar brain snacks for one child for one semester...maybe. This is all pretty pathetic. 

I want to see real ratings, not cheater ratings, so suck it up and pay the fee, or play elsewhere. What a bunch of whiners. It's just loose change for most...chump change.

I got a bad cold today, so I'm a bit more grumpy than usual, and have a band concert to direct in a few hours, so I need to pull some happiness out from deep down inside. Good riddance.

Yes. People will continue to cheat. I believe the intent of the policy is to discourage people from buying the gold membership over and over again in some sick and twisted way just to keep opening up even more dummy accounts, wreaking havoc on the ratings system. This reminds me of some sinister plot to take over the world. I read too many silly thriller books where doctor doom is at large menacing all of humanity.

I agree that some benchmark testing would be necessary, but it's certainly not "very difficult". chess.com already has timed tactics training. We're just talking about figuring out which problems are solved in a few seconds by a sufficiently large number of people, and then doing some simple analysis of the ratings of the players that solve them and how long they take on those problems and seeing if there are some really obvious patterns that emerge.

I strongly suspect that there will be obvious patterns in the data that correspond to the well-established scientific belief that as players improve their chess skills, they invariably learn more and more patterns whether they try to or not, whether they play blitz or do tactics training or not, and that their skill consists partly in their greatly improved pattern recognition abilities as a result of all the learned patterns.

Chess.com already has all the data that would be necessary, assuming they do actually keep the times for each solved problem in the database rather than just use the time to determine a score delta and then throw the time away. If they don't keep them already, we're talking about adding a column to a certain table in the database and changing a few lines of code to store that extra value after the problem is solved. That would suffice to be able to generate enough data from the existing tactics training software to be able to later do the simple statistical analysis necessary to determine the feasibility of such a project.

Anyway, that's all I have to say on the topic, which I realized today I should have referred to as a pattern recognition test in order to avoid the many off-topic remarks that apply to tactics training in general but not to the very specific proposal I made.

Might work. You'd need a pretty broad margin of error to account for slow brains and slow mice.

If you do this, you should only make them buy 1 month's worth.  I don't see a good reason to make them continue to pay.

You'd get their real name (from the credit card) making it very difficult for them to continue to make accounts if they were later caught cheating (even if they somehow were willing to continue to pay the fee.)