Major Proposed Change: Ratings Cap Above 2400 for Unverified Players

+1

I'm sure chess.com has looked at the numbers while we're left to guess.

I'm guessing the number of legitimate 2400+ untitled players is considerably smaller than a lot of users are suggesting.

I'm guessing the 2400 mark would only apply for live chess; It seems very low for correspondence.

The OP says all ratings (live, turn based, 960... everything) but they might change it.

FWIW 2400 turn based seems right on the money to me.

Something everyone should remember that makes our site (Chess.com) different from other playing sites like ICC: On ICC you CAN'T even play full time (meaning the amount of games we allow free members to play on our site) without a paying membership. Same on PlayChess.com! So, to make the statement that "we --chess.com -- can't control cheating" simply isn't fair.

We have over 3 million members and because of that number, we naturally have many repeat-cheaters being banned regularly. But because we offer such a great service that the other sites don't (namely that free non-paying members can play UNLIMITED games) there has to be SOME KIND of limitation.

I am not saying that there aren't some strong experts out there who can't play at a 2400+ level and be clean, but another thing to remember is that those players would likely soon be masters in over the board chess -- AND AGAIN, unlike OTHER sites, we give free DIAMOND level memebrships to anyone 2200+, while ICC and other sties don't offer this unlimited free access to everything to anyone but IM level+, which is a significantly smaller portion of the population.

Point being, even if there is a rare number of players who meet all the following criteria: wish to stay at a basic level; CAN play at a 2400+ level; and aren't cheating -- those people within that rare group have the opportunity to become lifetime Diamond members if they only gain a few more "official" (either USCF or FIDE rating points).

We love our members and want EVERYONE on our site, but the abuse and amount of people taking advantage of our "basic level access" -- that no other major site provides please remember -- is simply too high to not implement some kind of cap.

Danny

A suggestion has already been made (by somebody else), but I'm going to repeat it and elaborate a little, because it was such a good suggestion: tactics tests.

Every player, no matter what their strength, is capable of solving some tactics problems in 5 seconds of less. Players of greater strength can solve problems of greater difficulty.

If chess.com had a large database of tactics problems similar to those at http://chess.emrald.net/ but with slightly shorter time limits, problems that can be solved in seconds by players who are sufficiently strong, then players who approach 2400 and are not able to supply proof of an OTB rating could be given some of these problems now and again, a sufficient number of which they will be able to solve in 5 seconds or less if they are not cheaters.

If the positions are limited to those that still have many pieces on the board, there is no way a cheater could input the position to their computer fast enough to solve them in the required time. Any player who is 2400 but does tactics at the strength of a 2000 player or lower should be considered suspect.

What chess.com does with those people is a separate issue. They could investigate them, or they could require them to purchase gold memberships if they believe that is the best solution (I don't like it, but I don't have any good suggestion yet), but whatever they do, there is no reason to penalize honest untitled players who have the tactics chops they would be expected to have based on their rating.

How can you really prevent a cheater from cheating on a tactics test??? If you make the timer too fast, even titled players who simply couldn't rap their head around the problems that fast could score low. AND if the timer is long, there would be plenty of time to still cheat. There are WAY too many variables in that idea. I like it, but it isn't practical... Plus some cheating programs can solve the problems and make moves directly, without even a relay of the moves -- making the timer irrelevant. It worries me ...

Yes, you can really prevent a cheater from cheating on CERTAIN KINDS OF TACTICS TESTS. Not any tactics test, only a test of just problems that can be solved in seconds by players of a given strength. You, Danny, might miss a problem or two here and there, but do you think it's possible that OVER THE LONG RUN (that is, over a sample of statistically signficant size) you would only be able to solve tactics problems of this kind at the level of a chess player 400 or 600 points below you? It's possible in the short term, but it is so unlikely in the long term that the chance of it happening is negligible.

I think maybe you misunderstand the kinds of problems that I'm talking about. They are the sorts of tactics that LEAP OUT AT YOU INSTANTLY, not the sorts of things that you have to spend signficant time on. But what leaps out immediately to a 2400 player does not leap out immediately to a 2000 or 1900 player. That is the key, along with doing a sufficiently large number of them and considering the statistical properties of all their responses.

5 seconds to look and a second to move is plenty of time to move, and there is no way that a person can input many pieces and solve the problem in that time.

I doubt very much that there is any program that can solve tactcs problems from an arbitrary website directly in the browser without even requiring any human intervention. That is a fantastic claim. Such a program would either have to parse a huge mess of HTML, Javascript, and CSS (and does it know how to parse Flash and other languages too?), extracting the position from that mess, or it would have to visually process the UI and extract the parts of the screen that correspond to pieces of a chessboard, correctly handling all the different piece designs and color schemes that can be found. It's possible that somebody was determined enough to write such a program that worked for chess.com only, which would be much easier than the general case, but I'd be very curious to see some evidence of that.

I wasn't aware of that. Is that only an "official" (FIDE/USCF/etc.) rating, or is it based on your Chess.com rating?

Official.

Danny:  I don't understand where you're getting 2400=OTB expert from.  Several class A and B players, myself included, have already noted that we're close to 2400 in CC here.  While hitting CM is a long-term goal of mine, I'll be surprised if it takes me any less than 3-4 years and it could well take much longer.

Perhaps simply raise the cap to 2600 or 2700 for CC?  Live ratings are significantly less inflated than CC ones in my experience.  If you want to minimize the range of OTB ratings you'll impact, this seems to make more sense.

As a slight variation of some ideas above (sapientdust and others), why not have a mandatory set of 10 Tactics Trainer problems when a new account is opened? That should give a ballpark start rating.

The Tactics Test is an interesting concept, but hard to put into practice. If you are going to have an instrument with such a high ceiling, you are going to have to find 2400+ players willing to help develop it. In fact, at that level, you might as well include items to assess positional understanding as well as sophisticated endgame knowledge. If you see where I'm going with this, the instrument, whatever its final form, will be time-consuming, and potentially expensive to construct. Even if a reliable and valid measure of this sort can be developed, I am not certain it will be worth the effort.

You could also whitelist longtime players such as shoop2.

Tactics Trainer already has high-rated problems, does it not?

The idea is worthy to try, but considering what I have seen, I tell you what is going to happen:

1) The main cheating thing will just move from 2500+ to 2100-2400 range and nothing significant will change. Cheaters will have 2400 and just pretend they are indeed much stronger, but they don't care. They will continue spoiling Vote games, tournaments, team matches etc. Again, nothing will stop them from creating new account when banned and they will slay 1500-2000 rated honest amateur chess enthusiasts on its way up.

2) It will become hard to improve in the 2100-2400 range, because it will be clogged up with cheaters. Finally, people will realize it and private blacklists will take place, which is good IMHO when people care, but very strong honest basic members reaching 2400 will get a place on that personal & group blacklists.

3) Gold membership isn't too expensive even for cheaters. Cheating is like obsession or addiction and some will spend a few bucks to maintain their ego. And maybe, cheaters will start to cooperate. I predict many 2300-2400 basic sock-puppet accounts feeding their main premium accounts from different IPs.

Btw, OTB chess elite also cheat: Dembo, Docx and Kacakovski cheated and at least two still active titled diamond members too (they are less active now after "Dembo scandal"). And some titled players active here just now have terrible reputation from other sites as cheaters.

Toiletgate.