Some new (?) ideas, about rating calc. (FFA)

a) the current rating calculation only takes into account the average rating of the players involved. It's inadequate, because it's much easier to win (if you are 2400 for instance) with a 2600 in front with two 2300 sides, rather than with a 2300 opp and sides rated 2600 and 2300. But with the current system the value of a victory would be the same. We must take into consideration the configuration of the board.

b) in the classic 2p chess, the rating is a very accurate predictor of the outcome. If you are 1900, it's unlikely you would lose against a 1600, the probability exists but it's under 0.001%. In 4p chess for many reasons it's VERY EASY to lose even if you are 2800 and finish 4th if three 1800 sides cooperate against you (3 vs. 1) or simply if you play normally and put your Q at risk and your opp is stupid/incompetent, etc. 4th places are MUCH MORE PROBABLE in 4p chess (regardless of your rating) than the 2nd place in 2p chess.

As a result, most if not all good players avoid playing with lower rated: if they lose, they lose too much, if they win, they earn too little. 

I think it is normal to earn very few in such cases (we can't encourage farming), but it's not good that the cost of a total defeat is that high. I'm currently about 50/55th in bullet, have accepted two low-rated games and had 4th place twice in a row (paf, -50 points at least). The rating shouldn't be that volatile, it doesn't make sense. I propose we simply LIMIT the maximum progression or decrease in points to 10 points. As a result, high rated players would accept much easier matches with lower rated, and like that everybody would profit from their experience and learn faster.

c) we've had A LOT of discussion as about whether it should be pure solo (3 -1 -1 -1) which pushes people to play for 1st, or like previously 3 +1 -1 -3 (that unfortunately encourages playing for 2nd sometimes and makes less interesting games), or something in between.

It's pretty clear that the system can't be the same for low rated and high rated players.

I believe the current system is rather a good compromise, BUT: I think it should become solo earlier. 

Otherwise, I've always said that I think that the 4th place shouldn't be punished that much, because the difference between the 3rd and the 4th usually is not high and often depends simply on lack (for instance, with 2 good sides if your opp does nothing, you're 4th necessarily as nobody can resist a coordinated attack from 2 excellent sides). I also believe that the difference between the 2nd and the 3rd shouldn't be high because the 2nd in fact is the MAIN loser: he could've won but didn't. Often 3rd is 3rd for trying to become 1st (which is noble) whereas the 2nd is 2nd for having accepted that the 1st wins and having settled for 2nd (which is cheap).

Therefore I think that if we ever had to modify the overall system for having just ONE formula, as some asked (for instance Radon), a good system could be something like 3 -0.5 -1 -1.5. In this case the 4th is close to 3rd and is not punished too much; the 2nd and the 3rd are close as well; the 2nd still loses rating and thus people are not encouraged to accept the 2nd place too easily. And in the same time, it is not the abrupt Solo 3 -1 -1 -1 which, I know, shocks many, especially the players who are under 2200 and in fact they are the majority...

d) I remind that 3 years ago we had a very simple formula like 3 1 -1 -3 (I'm not sure), and it was then added the rule according to which if all players were above 1600 I think (which is today 1800 as +200 were given to all) were automatically Solo, or Winner Takes All as it was called then. This level corresponds today at least to 2100-2200 because of the inflation.

Maybe we should revert to such a simple system. But I think that if we do, it should be applied if there are TWO AND MORE players above some level (and not "all" like it was the case).

Here's an extremely simple solution that I've liked for a while:
Split FFA and Solo again but have them both go to the same rating for their respective variant and time control group. 

(For example, 4 min FFA and 4 min Solo both go toward 4P Standard Blitz FFA)

If we want to recalculate the FFA +3 +1 -1 -3, that's fine by me, as long as it doesn't hurt the experience of others.

The shared rating system simply doesn't work. In my personal experience it's not because I hate solo, but because other people don't play it like it's solo. 

Yes, that could be a good solution: simply let's allow the players who launch the game the formula THEY prefer (among 3-5 to determine, from pure Solo to pure FFA...) and the players who join will accept it or not. Exactly like they join or not today games with different timing options, or anon./non anon. games, or en-passant implemented or not, etc.

I.e. the rating would still be the same, unique, for "Individual 4p chess" as opposed to Teams, but with different options chosable by those who launch and join games.

Another example, why I believe that a maximum rating variation (up or down) should be implemented.

I've just played another bullet with no limitations. It's clear that my level didn't really change a lot between the beginning and the end of this game of 2-3 min. (everybody does agree on that I presume?). But I still dropped 25 points. I was 2000, my opp was 1549, the sides were 1805 and 1953. The opp attacked me ruining my defense and weakening me, so OF COURSE the sides took some 3 moves max. to kill me. As a result: I'm 4th, minus 25 pts. Am I upset to lose? Nope, IDC, it's inevitable to be 4th under these conditions. But to lose 25 pts is clearly unfair, and what will 99% of the strong players do? They'll avoid playing with players who are more than 100 pts lower. Therefore there are much less games that are played; average players do not have the opportunity to learn, playing with stronger players; and everybody has to wait for a while before a game is available. It's a collective lose-lose situation. And it can be easily addressed: the maximum variation in rating. If strong players don't risk too much of their rating, they would accept joining lower rated games much easier. It's impossible to pretend the level of the 4 players increases/decreases by 20-50 pts just in 10 min., it's absurd. The ratings are anyway something approximate and shouldn't vary by more than 10 pts in just one game, I believe.

I suppose one counteraction is to decrease the Glicko RD (rating deviation) of everyone. I still play open 1|1, I just hate how -15 for 3rd and -30 makes it impossible whereas 1|0 2300+ I get -10 at most for 4th and compared to 1700s, 2300s are actually easier to play against both because a) I don’t care if I lose some games and b) the players there don’t target on the basis on rating.

But of course, the rating system will never be perfect. Farming always exists and there still needs to be something to prevent that as well.

Is it your fault if you get 4th because your sides team on you and your opposite is passive or attacks you? No, right?

Is it your fault if you get 3rd in the 3 player stage because you played badly? Yes, right?

So obviously, 4th should lose less than 3rd, because 4th doesn't really have a chance to play for a win in such situations, but 3rd can. However, this is 100% counter-intuitive and it's just stupid to make a rating system like 3 0 -2 -1. The best we can do is make 3rd = 4th.

Playing for 2nd is also a problem, but again, we can't punish it by making 2nd lose more than 3rd or 4th. So in the end, it seems that we'll have to stick to 2nd-4th lose equally, which just comes down to Solo. Simple logic seems to show that this is the only system that works, at least in a moral sense. Is there anything wrong with what I said?

I used to be a big fan of FFA and hated Solo, but now it makes more sense why WTA is a better option than FFA. I think it's the only option that would work. Maybe we could decrease the loss amount so that players don't lose that much for placing 2nd-4th because other players don't understand the game well. I'm no expert in FFA or Solo, but these are just my thoughts.

TheCheese, 3rd=4th, absolutely! But I think it concerns more high rated games or at least average games. In games under 1800 clearly the 4th is the one who was the worst as people mainly do not cooperate. But yes, I would agree with 3 = 4, and it has been discussed. (Also, I've made once a provocative thread saying that 2nd is the main loser... Hence 2 = 3 = 4 which is perfectly justifiable philosophically. But still, people mainly won't agree).

I think we should just revert to 4 -1.3 -1.3 -1.3 again, honestly

-7  -2  +3 +6  ?

hey

why not just put back the historic rating calc. ? 

look the game #1

it was the first game played in 4pc 

so we just can come back to this rating calc. , the historic one  

why not?

No, it was not the first, at all.

In the beginning, millions of game were not saved. It only was implemented about 3-4 years ago.