Wide discrepancy between online and blitz rating. Why?

Out of sixteen people who played me "online" four got banned for cheating. Fact. 

And it will continue to happen, and thats why i dont put a lot of credence into online ratings.  Unless its a titled player.

The bad guys get caught, at some point...

Solskytz, and that's one reason I won't play "online" chess. Aside from the fact that I think the use of analysis boards and databases DURING a game is an absolute joke, i've seen many many correspondence players banned. I don't need or want any kind of help while I'm playing. Chess is supposed to be a battle of ideas between two players, not a battle of opening databases and analysis boards. 

To each their style.

I find nothing inherently wrong with this form of play - which helps you learn a certain opening much better (the ideas, main games played etc), plus sharpen your analysis skills. 

Two points:

(1) There is no particular reason to expect that a rating of X in one system is the same as a rating of X in another.  They're based on a different pool of players with a different mean, after all (look at the bell curves for chess.com and you'll see this effect strongly).  So it may be that Online 2000 and blitz 1700 (my ratings) are a lot more similar than they sound.  (Similarly, hardly anyone who plays a lot in both systems has FIDE=USCF, even though they purportedly measure the same thing.)

(2)  Online Chess is a very different thing than any speed of live chess.  You can use opening databases, you can analyze on a separate board, you can take a few days to think about the position.  On the other hand you can easily get too many games at once and find yourself playing a simul.  I have beaten several titled players in Online because they had 100+ games at once while I try to keep below 20.  You can also lose a whole lot of games instantly if you have a bad day--either by time-out or by multiple blunders.  (That dip in my rating graph?  One bad day that slowly and painfully cost me nearly 100 rating points!)

In my case my 2000 Online rating reflects 'what Mary plays like with no time pressure, an opening database, and the right to move the pieces around before deciding' and my 1700 blitz 'what Mary plays like with time pressure, touch-move, and no outside help.'  Interestingly, my USCF and FIDE are in between the two (1900), suggesting that I just can't play fast.  Sounds about right, actually.  I do much worse at game/60 than at 40/2.

I just recent started playing correspondance chess and i am finding out i love it.  I like being able to use the analysis board, opening databases, etc.  I firmly believe it helps lazy people like myself improve.  

I think that 2000 online is pretty consistent with 1700 blitz. Online ratings are generally rather higher, and not everybody is as good in one as they are in the other. 

If it was 2000 online and 1350 blitz, on the other hand...

Totally agreed, my online is almost 300 points higher than my blitz.

Uh, some people say that these two rating rating measure different things and are not related to each other. I disagree. This is not between chess and swimming. There is a strong correlation in the bell curve of online and blitz rating. They don't match up exactly on the number, but neither are they separate as to be totally unrelated with each other. From what I can tell from rating data on this site, the difference for the majority of players appear to be between 150-200 points. Taking variations into consideration, I'd say a difference between 50-300 appears to be normal. One can argue ad infinitum about how more time will allow you to play better, analyze more, etc, but it doesn't mean just because you have more time your chess strength will simply shoot to the roof. There is usually a ceiling to how much you can improve, no matter how much time and how many database you have.

That said, since I posted this thread (6 months ago), I've played more than 100 online games (including 960), and by now I've come to realize that a big difference between online and blitz (by "big", I mean the difference is between 500-700 rating points) is a lot more common than one would think. It's a minority, but fairly common, especially for the rating range I usually play in (online between 1800-2000). I do not think that such a difference alone is enough to conclude anything about engine use, especially when the online rating is below 2000.

However, I'm skeptical of these 2 cases (assuming the person plays enough games in both blitz and online for the rating to be meaningful):

- A super big difference (more than 800 rating points), or

- A difference of >= 600 points and the online rating > 2000.

I don't want to say this definitely means engine use, but I think it almost always warrants a closer look. I would even suggest that chess.com perform automatic audit of all such players the moment such thresholds are crossed. Not that it's not possible for anyone to register such rating, but it should be extremely extremely rare. (One example I can think of is when you're like Korchnoi: still can make a good 2500 game, but too old/infirm/eyesight-challenged/computer-illiterate to be able to play well on the computer screen).

Some people also cheat while keeping their rating intentionally under 2000 - as they think that chess.com scans more +2000 level games. They "stay under the radar".