Rating inflation because of 1800 option as provisional rating

@BobbyTopalov,

Jeez I'm sorry to have pissed you off.

1. again I haven't said my numbers are accurate, I'm saying my numbers are more accurate than "I read it in a news article somewhere" I can't find the article myself. I don't know when it's from and I don't know who posted it. I trust the actual stats over what a random player on the internet says. I apologize for my arrogance.

Yeah you're right, the people of my rating range would also get a higher rating(which is the problem I posted in my original comment? lol). Also no we will get paired up against weaker/stronger opponents than their rating represent because anyone above 1300 can beat these 1800 provisional players. I go way up to these super high rated players, or low rated players go up to my rating, meaning pairing is less accurate.

Quoting you:

 assuming your other opponents were even matches, works out to be roughly 1 point extra per win

Also you:

So, a score of 5.5/10 has net you a total of 9 point. 9/5.5 = 1.6 points per win. 

So now 1.6= roughly 1 that's some voodoo magic right there.  Sorry I tried to rationalize your numbers... You really seem angry.

Prevent people from being happy because they think they're being cheated against even though they can see that their opponent's rating is provisional. -___-

Also, who the heck quits chess after getting beaten badly once?

I don't want to offend anybody... but I would like to point out that assuming everyone is the same rating (let's say arbitrarily 1800) shouldn't the factor of different ratings come into play? I didn't read closely enough to how the wins and losses come, but what if it's 5 wins before all the losses?

@BobbyTalparov 

Can you link me to these news articles that report this statistic? Thank you.

Have you considered that someone rated around 1800 will more frequently stumble upon provisional players than someone rated around 1200, and have you considered that players at the 1800 level would also not prefer to play against people that are way too weak?

The possibility of choosing 1800 rating should lead to some rating inflation, but...how much? That is difficult to tell. It's probably negligible short-term, but we'll see in a few years whether it affects ratings more clearly long-term.

Rated games against the computers (easy, medium, etc) are no longer available, as they were used by some to "game" the system (e.g. by playing only computer easy to get to high ratings like I did). Perhaps they will do the same with the 1800 option, even if it's only a few people doing it.

The idea that chess.com's ratings have suffered inflation due to changes in how it allows players to identify themselves (novices, experts, and so on)  is a thesis that can be tested. 

Just take a look at the rating graphs at different times. I have seen no evidence of rating inflation since the changes that chess.com implemented. 

Just a few points:

- 20 million is bogus number.  Chess.com is counting every account ever made in that number.  No biggie, every single membership website that puts up ads and follows a "freemium" model lies this way, to inflate their importance in press releases, etc.   If 1/10th of those numbers are still real/active accounts, it would be very surprising.  If there are 20 million chess players here, then the live play numbers are absurdly low, don't you think?  Using the "just a drop in the 20M bucket" argument is disingenuous.

- The ratings pool is already borked, because chess.com has fudged it before.  Blitz ratings are too low, standard and correspondence ratings are too high (standard was too low also in the past).  The argument that the pool is completely relative falls short.  If that were true, Chess.com would never have "fixed" them by adding 100+ ratings points to standard, would they?  Everyone knows that if a major chess site's ratings pool runs amok compared to others, that site will lose members.  If you have a 1800 OTB rating, 1900 at Lichess, and 2350 at chess.com, then chess.com's ratings become a running gag, and the site loses credibility.  That's why they make adjustments, but making wholesale adjustments to borked system is just two wrongs that don't make a right.  Making the "actual ratings in the pool don't matter" argument is also disingenuous.

A ratings pool is not designed to have members that can just jump out and abandon their ratings and start over and over again trying to maintain a higher level.  You can say that the overall pool will be okay and "balanced", but who cares if a gazillion abandoned ratings for inactive accounts are technically balancing your numbers?  They might as well not exist, nobody is using them and you will never play or even see those ratings.  They are effectively dead players.

If a player tried to go to an OTB tourney and sign up as a new provisional player every time their rating dropped below their starting point, they'd be banned.  And admit it to yourself, if this behavior were not banned OTB, it would be happening everywhere, all the time.  Every time some 1500 player had a bad streak and dropped to 1350, he/she would "reset" to 1500...why even bother to fight back from 1350? 

Here, not only can you do this as many times as you want to, but you can decide to self report at 1800.   1800 is too high up the food chain for self reporting, because in general people always think they should be rated higher than they are.  So why give easy access to reporting yourself at the 95%+ percentile of players?

Go ahead and keep arguing...nobody is going to convince anyone until time tells the answer and it's time for "I told you so".  This will be years away, maybe even a decade.

Let's just focus on the Blitz rating

The current average blitz rating on this site is 1040.x

Let's see if that changes dramatically. I have not seen any major change in the averages in the last two years here.

The frequency of an established 1800 meeting a new player with a self-chosen 1800 is obviously higher than it was when everyone started out at 1200, outside of tourneys (which really only impact Live since most Daily events require at least 5 games completed).

And I agree that player that are at 1800 would prefer not to play 1200's (or lower) but I would say that the number of times that happens, as a percentage of any established player's games, is likely minimal. You would have to be extremely unlucky to get randomly  paired with a bunch of new accounts (no previous games) where the player said they were an expert and were not a decent player already (within 200 points of 1800).

I wasn't comparing the frequency of an established 1800 player meeting a provisional player after the new system has been implemented to the the frequency with the original system, I'm comparing the frequency of the established 1800 player meeting a provisional player compared to the established 1200 player meeting a provisional player. There's a huge amount of 1200 players compared to 1800 players. One player joining as an 1800 will have much more impact than 10 people joining as a 1200, because 1200 is the most common rating, which is why  I thought, the provisional was always chosen as close to the most common rating as possible.

My point is that the best way to have accurate ratings is to let well established players play against provisional players as little as possible, and to let provisional players play only, and as many well established players as possible, which is why the provisional rating should be the most common rating. To spread out these provisional players over as many players as possible.

Otherwise the established rating is only a representative of how good you are at beating provisional players, whilst the entire point of the provisional rating is that we don't yet know the true rating of the player, thus making the established rating a sort of semi-provisional rating, which is by definition, inaccurate.

And no, I'm not saying it directly affects the provisional player himself, as he will soon be at the accurate rating range(except if he's much higher rated than the average or much lower rated than the average). It's about how many times a well established "high rated" player will encounter a provisional player.

And it won't happen very often. I did a few spot checks early on, when a similar topic came up. The vast majority of members were choosing 1400 and below. I'll have to see if I can find the topic but I think the majority were 1200 and below.