I've been begging management for a provisional period to the ratings since I joined... they mutter something about 'glicko' and swing the incense and move on...  and there's always people that feel it's important to lecture about how ratings don't matter and we shouldn't care... which always makes me want to say: if ratings don't matter and you don't care, fine...  why are you bothering to post? Since they don't matter and you don't care about them... you don't care if there's a provisional period or not, right? So let the people that DO care discuss it.

Estimating one's own rating is subject to abuse by loonies, but isn't a bad idea -- Anyone with a real rating from USCF or FIDE should start at that rating... there's no downside to it I can see... it'd only help square Chess.com ratings with USCF and FIDE ratings. And truly, it's absurd to start someone who is a titled Master at 1200!...  my god it's not fair to have to play him/her in a tournament!  That must have been gut-churning for opponents... How you like for a tournament draw: NMReb rated 1200... errr... *gulp* oh boy.

@TNziggy... 1400 playing at 1600 level is within the realm of the acceptable, as you say you've been improving which is the point... it's 1400 hiding the skill of 2400 player that is just silly. And it's happened to me!  I managed to draw a player who is currently 2500+! An awesome achievement for me, unfortunately at the time he was climbing, and was rated 1700 and I _lost_ rating for getting the draw.

Why not just try to improve your game so you aren't affected so much by 'climbers' or instead just blame the fact that they are climbers for your loss and not the bad moves that you made. Not only will you get better by making excuses for your loss you will feel better about your losses too.

My gripe is the opposite of a "climber," it is a "faller." I was playing a game against a player with a long record of games who was rated above 2100 at the start of the game, which would have made him the highest rated player I had even drawn, and I did manage to draw him. But right before I claimed my draw, in the last day or two, his rating suddenly plummeted 400 points! It seems he was playing about 165 games at once, and lost a whole block of games on time. So I lost points for the draw instead of gaining! I think points won or lost should be based on the ratings at the start of the game, not those at the end. The ratings at the start, that's what you sign up for. In an over-the-board tournament, you gain or lose points based on what the ratings are at the beginnings of the games.

This is an interesting option. However it should be considered that correspondence chess is a lot slower than OTB, so people's rating tend to vary from a little to a lot depending on the length of the game and the experience of the player.

Despite that last thing, I'd gladly fix the endgame points to the ratings at the begining of the game.

Anyway.. like Olimar said, it's probably a problem of the system and not the players. And the thing that bothers me the most is really not the "rating-damage" but the unexpected high-level pawnage from a hidden master that can take me out from a tournament were I had decent chances.

 Chess.com uses the Glicko System.  It's identical to the Elo except that it adds a correcting factor (Glicko RD) to give a more-accurate picture of a player's strength.

 Only if you have a high RD.  I'm currently playing against "pruliere" and he has a 2000+ rating, while I'm at 1265.  I only get 25 points if I were to win this game.  He will get 0 if he wins.

I read alot of posts on this subject and it has a frightening number of posts where people are more concerned about how many points they are winning and losing then the actual enjoyment of playing the greatest game of all time.  A rating doenst determine the kind of person you are.  Learm from your losses, and enjoy the wins :-)

It's not that we don't care - it's that we just don't obsess about rating.  Your rating here should be a fairly accurate picture of your actual strength, although "best win (by checkmate/resignation)" would be a nice way to further define an upper bound.  In most cases, your Chess.com rating will be slightly more correct, as the sample sizes for OTB tournaments are much smaller, and they represent a tougher population filled with only tournament players.

All ratings are always approximations.  There's no such thing as an "actual rating" - that number is known only by God:

http://math.bu.edu/people/mg/ratings/approx/approx.html

I feel like clinging to and taking pride in your rating will help you get better though.  I know that I wouldn't get better at chess unless I felt some competitive desire to do so, and I feel like nothing brings this out better than being concerned with my rating.  It keeps me accountable for my games by making the losses and victories more significant.

I would claim, actually, that if EVERYone were to be concerned with their rantings it would make us all better the quickest.  If I lose to a 1050 person who doesn't care about his/her rating, it could just be that he/she played an excellent game.  If I lose to a 1050 who DOES care, I know that I lost to someone I should have beaten.  In the former case I get no good feedback because I can't know how well I was playing, whereas I would in the latter.

It seems like providing more flexibility in describing who can play in a tournament you are planning is the key.  More granularity in the number of games played was suggested.  Another would be to go directly to the Glicko rating so you could directly omit or include players whose ratings will rapidly change. 

But this can only limit the phenomenon, not stop it.  Climbers and Fallers can not be avoided because the result of the winning is an increase in rating.  I doubt anyone making it to the fourth round of a Quick Knockouts 1400-1600 tournament will still be rated in that range.  That person will have had to win too many games to get there to still fit that range.

I agree that the minimum number of games that a person plays before they can join most tournaments should be much higher, at least 15 or 20.  Although there should still be some "newbie" tournaments open to people who have played fewer games than that.