Is this immoral

One more update, he just blundered and is about to lose a bishop without compensation in one move.

I think I may have a won game.

This point hasn't been given due attention.

@Tarikhk/Batgirl

Well yes, but why wouldn't they? They have nothing to lose in terms of rating...

There's some kind of fallacy that playing games -600 to you is a good way to gain rating. It is not.

A 2000 player playing a 1400 would have to win 30 games for every one he loses. If he beats a 1400 player 29/30, you can safely assume he is so much better than his opponent so as to be that strong.

Super GMs (2700+ FIDE) generally lose points when they play in events with lesser ranked players (2600-) because actually it's very hard to constantly win against people rated considerably lower than you.

Marvellosity:

Well yea, clearly...this 2067 rated player is about to lose to me, rated 1270.

But winning 30/30 is very possible if you only play players rated around 1000. Even I can probably win 25/30 of those games.

Yes, tornado, but if this 2067 player lost to a player of your strength more than 1 in 50 times, he would not be 2067. So you have to presume he would beat you 49/50 times. If he would not, he would not be 2067. See how this gets circular?

Btw, there's a big difference between 25/30 and 29/30. 25/30 would represent a massive rating loss for a player 600 points ahead, while 29/30 would just about keep him even. i.e. for 29 wins to be worth one loss, a win = 1 point and a loss = 29 points lost. So 29/30 is a 0 point net gain, while 25/30 is a 120 point loss.

marvellosity, I don't buy it. I agree in general that the ratings system is self-correcting. However, a rating based only on games against novices is not as reliable as a rating based on a wider sampling of the whole range of player strength.

I believe that is the rationale for GM norms. To become a GM, you can't just rack up victories against people who just learned how the pieces move. You have to hold your own against current GMs.

Come on, it's unlikely that the model would apply so perfectly. In an ideal world, what you said would be true, on average, in the long run. But not everything is so ideal as to fit the mathematics perfectly. Tornadofdoom's opponent is obviously nowhere near 2000 even by chess.com's inflated rating system (trust me, take a look at his games and it's obvious). He is a player who "always" beats <1200 players yet "always" loses to higher opposition--it makes sense though: you get good at what you do the most, and what he does the most is playing against <1200 opponents.

tornado, you're looking good in that game. keep it up.

I'm not a mathematician, but according to the FIDE rating calculator, if a player with a rating of 2067 wins a game from a player 1270, he gains 0 pts.  If this is accurate, there's absolutely nothing to be gained by a high rated player to consistently play low rated players. He would gain nothing if he won,  -7 pts. if he drew and -15 pts. if he lost.

"Doesn't the challenged person have to accept the challenge?"  challenges any inference that the high rated player is somehow taking advantage of lower rated players.  If a player challenges another player and that player accepts, then what's the problem? 

It seems to me that ratings gained in a non-tournament setting (the cheating question aside) are somewhat iffy anyway since no player is forced to play both colors and any player can refuse to play particular opponents.

'is this immoral'

no

the chess.com rather quirky method of grade calculation is the real issue, & comparisons with FIDE are dubious at best

please don't blame a player(s) for achieving wins

chessnetwork quick is the most extreme example I know. Win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point, win add a rating point ... to infinity and beyond, well maybe not infinity but 4200. 

>:)

That's true for live ratings, but not the case for online ratings. In online chess games, if you have a small enough RD and beat a low enough rated player, your rating won't change. But even the online rating system allows for inflations like the type you described. The RD is just too large.

It's a bug in the rating system that also used to affect the online games but has since been fixed.

Thanks chaps for correcting me.

I've not noticed the chess.com calculation method change though.

How recent was this fix eg 19th July, 2070 beats 981 online and becomes 2071. I would have thought this rd was sufficient to warrant no change for the higher grade.

... In online chess games, if you have a small enough RD and beat a low enough rated player, your rating won't change. Can you please elaborate did you mean a large enough rd or have I misunderstood.

This is a great idea!  I should try this from now on!

What a great way to increase my rating!  I'd have never thought of this on my own...

G

By the way, here's that guy's chart:

You can really see where his losses were!

Those losses must be devastating -- wiping out months of painstakingly tedious work every time one happens.

What I mean is that the RD values assigned to us are too large. They should scale down the RD for everybody so ratings can become more stable--our current RD values are too high and so even beating a much weaker play nets you points. Alternatively, they could make rating changes less sensitve to RD.

Just because I feel proud, it's down to a King+Pawn(for him) vs King+Rook+Pawn pair for me.

Hello, higher rating. It feels good to be up in the 1300 range again.

Here's the game: