Proposal: Reduce All FFA Standard Ratings by 700

I am actually strongly against any rating reductions at the moment... and it's not for egotistical prerogatives . I am wondering what we mean by "rating inflation". Yes, the ffa ratings are at record heights, but I don't see how they can get too much larger, at least for a while. At our current rating range, it is extremely challenging to maintain something like 3200+ because everyone is so much lower than 3200+ therefore the gains are minimal and the losses are punishing. As already pointed out, I only see the ffa leaderboard as a very appropriate reflection of the current rating distributions at the top in 2 player blitz.

From my point of view I see no real reason to reduce any ratings just because "that's how we've done it in the past" for a more "accurate" rating which is only subjective. Why change something that reflects our 2pc counterpart so well? If or when we get the top 20 to 3100+ maybe then we can consider rating reductions to be more "realistic".

From my perspective there seems to be a common misguided theme among the highest rated players (say 3000+) that everyone is "overrated" because we see 2800+ players play so poorly from our experienced standpoint. Yet if you would face our current 2800s as say a 2500 we would find them quite a challenge. It's a matter of perspective: If you look at Hikaru or someone beating 2800s in 2pc with ease it only seems natural; Hikaru could say they play "extremely poorly" but that is from his standpoint as one of the best. I think the same can be said for the 4pc ffa elite in their unfair assessment of lower rated players.

np lol. I will say I think it's a good idea to just shrink everything eta: refering to changing glicko. With that said, if I get no points for anything (there aren't a lot of high rated games all the time), why would I play? So I'd think if you'd reduce the glicko you also need to shrink ratings. Even then, you need to make sure a reduced everything system works...

Hi!

Two things. First, I disagree with your assessment that ratings will stop. The reason I give to suspect this is because, overall, I think everyone's ratings are going up. Secondly, I just don't see 3200 or 3300 ratings as unattainable in the current rating system, but we'll see.

Secondly, I think this comes down to either a fundamental disagreement on the role of the rating system of 4pc with relation to 2pc. I think if you agree with me that 4pc ratings should line up somewhat with 2pc ratings, then all my arguments logically follow.

But you don't agree (which is fine, I worry my words come off as snarky, which I don't mean at all) But anyway, the reason I give for 4pc ratings to line up to 2pc ratings is this: we can give 4pc ratings more weight if they are vaguely lined up to 2pc ratings. As is, all my 4pc rating tells me is my relation in 4pc to other players. 3200 has less meaning because it tells me very little about my objective skill. How good is 3200? I don't know! If we attempt to line it up with 2pc ratings, however, then that gives the rating some feedback on my objective skill. If I could say, I'm somewhat close to the equivalent of a 2500 rated (edited to change: sorry, 2pc) player, that carries a lot more weight, in my opinion. As is, 3200 might as well be 500000, and much of achievements people get, like "I'm 2000) are misguided. This is not to say it is meaningless to be 2000, or 3000, or 3200, but I think we lose out on making 4pc ratings more meaningful by having ratings disconnected from 2pc, and all the meaning and achievement that comes from those ratings

I hope that makes sense?

I pretty much agree here (partially disagree on the first sentence, for different reasons.) Isn't it sad that we often take for granted and lose the beauty of everything that becomes routine or "ordinary"?

And more specifically, agree as well. Most people have played for years and are either really naturally talented or are extremely skilled in 2pc (which overlaps), which leads us to somewhat discount or invalidate things that would be difficult or impossible to attain for almost everyone.

i don't understand the problem, if ratings are relative, we could give everyone +1million, and the relative ratings would be the same, (a 1,002,700 is good and a 1,001,000 is bad) so inflation of the ratings does not really hurt anything unless you are trying to be same ratings as regular chess, which is comparing apples to oranges.

2 player chess ratings are a measure of skill. I can tangibly say someone is a GM, or 2000 and have some measure of what that means.

4pc ratings are less of a measure of skill. If someone is 2000, or 3000, there is less of a measure what that means.

Connect 4pc ratings to 2pc ratings, and we can partially translate 2pc skill into 4pc skill, thereby adding "meaning" to 4pc ratings.

I hope that makes some sense

I mean I see “2000s” throwing easily won endgames and being 900 2PC. It’s just a quality of life thing in a way for supporters of such a change happening.

So, if those 2000 players become 1300. It will be more competitive at the top since it's easier to get to 2600 then 3300. Those 2000 rated players can never get to 2600 in order to participate in the World Championship qualifiers. Back in the day, only skilled players were at the top whereas now, it's misleading, a pool of unskilled players basically. Farming is a huge issue. Like, if you are at top 20, doesn't it bother you the fact that, there might be a player or two with the same rating as you, but they just let their opposite get checkmated and they never cooperate? That might be a rating issue as well since 4th loses a ton. Skill-wise 4 player chess is not competitive anymore and it becoming dull...

A flat change like this is fine if we are resetting what a certain elo "means", which seems to be what you are suggesting (bringing the ffa elo in line with 2pc). Chess.com did this a couple years back when they brought bullet ratings up a couple hundred points to be more in line with blitz ratings. A flat reduction, however, does have serious potential effects on the playerbase which I feel need to be strongly considered before anything like this is enacted. I would also like to see a more statistical method applied to determine what number of such a flat change should be. I'm also not sure I agree with your effort to bring 2pc ratings and 4pc ratings into balance. There are plenty of FFA players better than me who I could beat or at least compete with at an equal level in 2pc. What I'm getting at is I'm not sure if these 2 things can be compared with any degree of accuracy. Once again, some statistical analysis should probably be done towards this before any changes are made. This potential change is quite similar to the bullet/blitz equalization Chess.com made that I mentioned earlier. Yet every difference seems to pose problems. Blitz/bullet deal with the same game, and have a high degree of similarity. 2pc and 4pc do not. The bullet change involved bringing bullet ratings up, which made the change come with minimal complaint from the playerbase. Reducing everyone's ratings by a flat amount could cause huge complaints from our playerbase, and especially with the playerbase still fragile from the merge, I would be cautious towards the effects.

I also see a lot of discussion regarding rating inflation of certain players being prevalent. I don't want anyone thinking a flat change solves rating inflation. A inflated 3100 will just be an inflated 2400. This is because not everyone has inflated ratings, some have remained stagnant, while others (mostly those playing against lower players) have risen. Dealing with "farming" such as this does require a more nuanced and thoughtful approach, which may include changes such as more sharply decreasing the amount your rating goes up the lower the opposition is, amongst other things.

Also: how do you plan to deal with other time controls? The problem is those time controls have much more inactive players and few active players. Already on the ffa blitz and bullet leaderboards we see ranges of as much as 600 points separating the top players (hi ) and the bottom of the list. That poses problems, as does the issue of inactive players. Should be match certain time controls to their 2pc time control counterparts? A flat change seems attractively simple at first, I just feel it is way more complicated than appears at first glance.

LazyImp tiene razón sobre el cambio brusco, yo lo introduciría poco a poco en la próximas asignaciones de puntos hasta estabilizarse en la nueva baremación, trascurrido el tiempo prudencial y pertinente.

10.05.2019

there was hardcore time, I couldn't even reach 1900

although we already used cooperation in its infancy https://www.chess.com/variants/custom/game/1152492/56/1 (-300 to all ratings)

"A flat change like this is fine if we are resetting what a certain elo "means", which seems to be what you are suggesting (bringing the ffa elo in line with 2pc)."

You got it.

"how well can 2pc & 4pc ratings be compared"

that's the limitation of this all. I personally think they are close enough there is value to this idea. It won't be perfect; I think just getting it generally correct is acceptable.

I agree with your comments of statistical analysis and inflation

"adjusting for TC + playerbase"

The playerbase reacting negatively could be a problem, yes. I mean it shouldn't be, but it could be. As for adjusting for different time controls, I am not sure; that is a valid issue. I was thinking just leave blitz, bullet, and hyper alone for right now, until they have a larger and less volatile player base. I don't see much harm in having the rapid ratings lower and keeping theirs the same. Alternatively, they could be raised or lowered by smaller amounts for right now.

Thanks for your thoughts!

Mirad, yo creo que donde está el problema es en el reconocimiento social, le digo a mi sobrino que tengo 2300 de ELO en ajedrez a 4 y me dice "Ostras tío, que pasada" y en realidad soy un matao JAJAJAJAJA. Creo que de alguna manera hay que ajustarlo a la realidad social.

https://es.wikipedia.org/wiki/Ranking_FIDE Magnus Carlsen con 2853 es el ajedrecista que más ELO tiene, CLARAMENTE vivimos desajustados.

YES we could do that. But whatever you do, the inflation issue will still be there, and maybe 1-2 years later the ratings again will be too high.

The real problem behind is that new players arrive continuously, offer easy points to "old" players and disappear, and the overall system points tends to increase like that. The only way to address that is to reduce everybody's ratings proportionally every X days (for instance, every week minus 0.5%).

Also, I believe that many high-level ratings are easily inflated just because of farming (deliberate or involuntary). And in fact it's very easy to solve this huge problem, by taking into account the respective levels. My proposal has always been to severely discriminate rating changes for games where the level difference is too huge. Read about that here:

https://www.chess.com/clubs/forum/view/about-ratings

Some other proposals were also formulated. Among other, I strongly believe that the configuration of the board should definitely be taken into account.

As of today, only the average rating of the board is taken into account, and it's total nonsense. Because if you have a 2700 in front and 2200 as sides, it's twice easier to win than with a 2200 in front and 2200+2700 as sides.

https://www.chess.com/clubs/forum/view/some-weaknesses-of-the-current-ffa-rating-system

https://www.chess.com/clubs/forum/view/a-new-proposal-ffa-solo-ratings-points-system