this is is old beause now carleso is 2861 and beat the world record

Vietnamese talent will depose Carlsen soon

gooo carlsen... go liem .... d best of d best in chess

btw there was an elderly player at the club i played at who had a stroke. the club organizers wrote to the uscf about it and they dropped his floor a few hundred points; he got to have an accurate rating and continue playing for years.

"I am curious what specifically the problem is/was with sandbagging, and how the rating floors were intended to help avoid it? "

This link may be useful: http://www.timmybx.com/uscf-president-agrees-with-timmy-on-sandbagging/

But anyway, I guess everything is relative to what you consider to be a lot of money....

Consider the upcoming US Open. Class prizes are $2500 which I happen to think is a lot of money, and for such prizes you only compete against players in the same rating class as you. So if there is no rating floor on ratings, and say you are master strength and currently have a master rating, then what you can theoretically do is play in a few cheap weekend tournaments, lose every game, which will artificially lower your rating far below your real strength to say -- class B...Then you can enter a national tournament -- such as the US open -- and unfairly compete against B players for these large class prizes, when in fact you are really much stronger than class B. 

Unethical, yes, but such things are an unfortunate side effect of what happens when large amounts of cash find their way into a sport, in this case, chess. On the other hand, our top players certainly need/deserve large cash prizes, so what to do?

I imagine the sandbagging involved people purposely making their ratings much lower than they should be (by throwing games) so that they could enter a big-money tournament in a rating class that is far below their actual rating, and then stand a very good chance of winning. The person would end up competing for a large money prize against players that are all much weaker players.

However, if this was the main issue, there is an obvious solution that doesn't defeat the entire purpose of ratings. Simply keep track of the highest rating a player has ever achieved, and use that for determining elibility of tournament sections, prizes, etc, not the current rating. A player who was formerly 2250 wouldn't be eligible for a big money U2000 prize, so there is no point in him purposely losing games so that he can get his rating low enough to dominate that section.

Does anybody see any problems with this idea? The only issue I see is that the small number of people who are much weaker than they used to be would be forced to play against players much stronger than they are, but that seems much less of a problem than those that result from abusing the rating system. And maybe there are other alternatives like let them play in a section below their highest ever and let them get the rating points, but keep them ineligible for prizes regardless of their performance.

It does seem odd to co-opt the use of ratings (originally designed for relative comparison of player strengths) for encouraging participation by an audience of elderly players (through the addition of rating floors).  

It's easy to see, as David mentioned, how this has inflationary effect and in a way defeats the purpose of the original design, though for most players of the same generation it still keeps the comparison relatively accurate.  Players across generations become incomparable then, at least not without an adjustment for inflation (as we tend to do with currencies)...

I am curious what specifically the problem is/was with sandbagging, and how the rating floors were intended to help avoid it?  fish_food, would you briefly clarify for those us who are not in the know.  Thanks!

USCF rating floors were created to prevent sandbagging...for those old enough to actually witness the rise of obscene levels of cash prizes in the big open tournaments (world open,etc) we actually remember them being implemented for that reason.

im dpruess: Thanks for the explanation. That makes sense. Ratings floors still seem like an obviously bad solution though to the problem they were intended to solve, since the whole point of a ratings system is to predict the expected results of games between players and give an estimate of a player's strength relative to other players. Trying to solve an unrelated problem by compromising the integrity of ratings seems a poor choice. One can see the absurdity of the policy by considering a 2400+ player who has a stroke that impairs his chess ability so that he plays at the level of a 1500 player. If he continues to play chess, which he might if he still loves the game, he will wreak havoc on the ratings of everybody he plays and loses to that is much lower rated. In effect, that person would become a "ratings pump" that lower rated players could use to increase their rating again and again, and those people will then have over-inflated ratings that would cause other people who beat them to gain more ratings points from the win than they should.

As Glickman says in the article, this makes the ratings system serve a purpose it wasn't designed to do. His suggestion of a good system of titles for achievement seems to make more sense (so people won't use ratings as a mark of achievement and care so much about losing rating points). He also remarks that the problem of scholastic players whose rating is wildly inaccurate due to massive leaps in improvement is solved by the Glickman-2 system, which he says has been introduced in Australia with great success (young players there that are rapidly improving are tracked accurately).

sapientdust, i believe the (or one) rationale behind the uscf ratings floors was this: as players got older, their abilities declined a bit. therefore, they would not play in tournaments, because they did not want to lose lots of rating points to younger players. but a theoretical goal of the uscf is to increase tournament participation, so ratings floors allow: the young tigers to gain points off the old guys, and for the old guys to keep their ratings! win-win!

(btw, that policy has an obviously inflationary effect, since it creates rating points out of nowhere; however the pool as a whole was deflated bc of the influx of scholastic players who made up a greater and greater percentage of the uscf rated population. they would often earn ratings of e.g. 500, then take classes, improve a thousand points, and then take those points from the rest of the rating pool).

fish_food: interesting article. Thanks!

The "rating floor" that Glickman is talking about in that article though is different than the minimum performance rating required to get a FIDE rating (maybe that's not supposed to be called a 'ratings floor', but that's what Edwards called it too). Glickman is talking about the ratings floors in USCF below which your rating cannot drop after you have already achieved a rating, no matter how many games you lose. That seems obviously like a bad idea, as it defeats the purpose of having a rating if they no longer accurately represent how well one is expected to perform against other players.

@spatientdust-- the rating floor mechanism would have to be verified by computer simulation also.

Perhaps this link http://www.glicko.net/ratings/cl-article.pdf will be of interest...the rating floor, coupled with loss of established stronger players in the USCF, at one time resulted in deflation rather than inflation in American chess.

You might note the FIDE is using an older rating formula somewhat different than the USCF formula -- which has undergone considerable tweaking.

fish_food, you're right, it would definitely not be bell-shaped. Thanks for the correction.

im dpruess: Thanks for the explanation. That makes sense.

I just noticed an update to the original page I linked to earlier that I hadn't noticed previously:http://members.shaw.ca/redwards1/update.html

He discusses some explanations people have put forward, including those of fish_food and im dpruess, and he concludes that the rating floor explanation seems the best one at this point, but he wasn't able to get a response from FIDE when he tried to get more information.

@sapatientdust ---"I don't see why that would be expected, rather than a bell-shaped curve with most performing about average and those performing better being balanced by those performing worse."

USCF published the curve for its membership some years ago -- before the inflation I believe-- and the curve was highly skewed towards the low end, and so not really resembling a Bell curve, with most players having really low ratings.

That is kind of what I would expect...players who are not that strong try out a rated tournament to see how they fare, must get crushed, and lose interest, and never improve.

Regarding my weak pool and strong pool idea. One could write a computer simulation and simply vary the average size and strength of the rating pools, vary the influx and out flux of new players into the pools, and then let them interact and see if there was any observed inflation...Since the ELO rating gives the statistical probability of a win, draw, or loss, this is a simple problem for a computer (but impossible to do on paper from first principles).

But you are quite correct, my explanation could be wrong, it is just something I have postulated based on my empirical observations.

sapientdust, imagine you need a performance of 2200 over at least 10 games to get a fide rating. a player with a "true strength" of 2150 will not earn a fide rating one year; then the next year he won't earn a rating; then in some rating period he will have a string of two good tournaments, and earn a fide rating of e.g. 2216.

the same happens wherever they set the bar -- 1800, 1600, etc... a certain number of people having sub-par tournaments won't earn fide ratings.

and even if this is only a small effect (like an average +5 bias on incoming ratings), because there are far more players rated 2200 than 2600; and far more rated 1800 than 2200, i think it could push a significant inflation up towards the top.

(although players also improve, as per my reasoning below).

fish_food: another interesting idea. The suddenness of onset of the ratings rise though seems to argue against your explanation, because the number of soviets playing outside the USSR wouldn't have gone from the pre-1986 rate to a different rate overnight. What you would expect would be a slower upward trend, as some are able to play outside the USSR for the first time, with that upward trend accelerating as more and more players start playing outside the USSR and then leveling out to a flat line again when it reaches equilibrium.

p.s. the article I linked to isn't mine. It's by Rod Edwards.

im dpruess: very interesting explanation! That certainly sounds like it could explain the sudden onset of ratings increases if there was a change like that at the end of 1985/1986.

What is the reason though that we would expect "among players earning their first fide rating a disproportionate number would have been having an above-average performance for themselves"? I don't see why that would be expected, rather than a bell-shaped curve with most performing about average and those performing better being balanced by those performing worse.

@sapientdust- Nice work.

The cause of the inflation is open to question...

It used to be common wisdom that (before Perestroika ), a Soviet candidate master was equal to a USCF master. Anytime there are pools of players that are isolated from each other, such rating differences can develop over time. Players in the Northwest US (Oregon, Washington, BC) used to be underrated relative to California. I can remember some NW players talking about how, after they made a trip to a California tournament, the California masters were "dirt". However, now international borders are more or less open, weaker pools can mix with the strong, and the strong can travel abroad to feed on the weak. Other countries with isolated pools can boast similar disparities. West Australia used to be cut off from East Australia, and so if I am not mistaken, East Australian players remarked on how inflated were the West Australia ratings were...now however, there is a good highway and you can DRIVE from Sydney to Perth...it was not so easy in the old days.

You mention 1986 in your article. When was Perestroika and when did the opening of the borders take place. About the same time.

However, in the final analysis, no statistics is required. Play over the modern games, and compare with the old games. Nothing remarkable is seen in terms of improvement, except, as I have mentioned, we now have exact 6 piece endgame databases and computer analysis of openings....

So I have to smile when I read the claims on this thread how players are stronger now. I have seen the opposite. Players who currently rank as masters, yet who formerly would not rate quite so high.