SFA (Solo-for-All)

continuing the topic of reuniting Solo and FFA (https://www.chess.com/clubs/forum/view/ffa-solo-simplify-the-game), I propose option for how to do this: SFA (Solo-for-All). the main principles that it contains are:
1. the FFA rating system is well suited for players with a rating up to 2300. it allows to quickly climb to the top if have the necessary skills, is friendly for beginners, since there are 2-3 winners in the game, and is also popular among most players
2. the Solo rating system is appropriate for high-level players over 2300. it allows to concentrate on playing for 1st (FFA, due to the presence of a disadvantageous 4th, distorts the game for 1st)
3. in contrast to the WTA system (FFA turning into Solo according to the rating border), it has pretty smooth transitions, allowing to get used to the new point distribution per game
here is SFA equation: x1 = -0.5 * (tanh(0.01* (x - 1800)) - 1.0) ; x2 = 0.5 * (tanh(0.01* (x - 2300)) + 1.0) for rating change 3+x2  x1-1.33*x2  -x1-1.33*x2  -3+1.66*x2. here's FFA equation: x=-0.5 * (tanh(0.0036* (x - 1800)) - 1.2) for rating change 3 x -x -3 (for those who want to experiment with functions: http://fooplot.com)
as can see, for players under 2200, almost nothing has changed, players with an average rating 1200-1700 in games play in FFA (31-1-3), 1900-2200 in modern FFA (300-3), 1700-1900 get used to 2nd = 3rd. top players 2500+ play in Solo (4 -1.33 -1.33 -1.33), 2200-2500 get used to playing for the 1st, not looking back with fear at the 4th

what do you think about SFA? do you think should correct anything in this system? if you have any alternative specific (function) suggestions?

Generally agree with the structure; people can debate about the breakpoints (e.g., whether Solo should kick in at 2400 or somewhere else), but directionally seems logical to me.

I would say though that perhaps it's better to not be "pure" Solo at the high end, but "almost" solo. If the goal is to encourage playing for first and discourage playing for second, I think all we need to do is to make 2nd place lose a token amount of points, but it does not need to lose equally as 4th place. By making 4th the same as 2nd and 3rd, I think we discourage aggression in the early game in the race to take out a side player since everyone wants to avoid being last, and that early aggression is an exciting part of high level FFA. Therefore, I think 4th should still lose more points than 2nd and 3rd, and 2nd just need to lose points as well instead of gaining a small amount, and psychologically I think the expectation of losing a token amount of points by not getting 1st will help motivate for playing for 1st, while the larger loss from 4th will continue to motivate early aggression to take out a side player.

Not sure how that plugs into the formula, but for 2400+, I propose something like +4, -1, -1, and -2.

I would suggest rather something very simple, for example what we had before 2019 was not bad (1500+ was automatically WTA, i.e. 1800 now). I.e. we could have "Individual" games (vs. "Teams") which would be FFA, but if the lowest rating of 4 is above 2000, that would be Solo (WTA).

But in this case we should also implement the following principle: if the highest and lowest ratings differ by more than 200, the game gains/losses are automatically reduced (by 3, 4 or 5 for example). That would a) reduce dramatically the farming, b) make it much less dangerous for high rated to play with lower rated, c) make the ratings more relevant/accurate. (And when people play "their natural rating" ranges, the ratings would really be impacted; and it's pretty easy for anybody to launch a game with for example 2400-2600 range in order to be sure to make a fully relevant game with relevant rating calculations afterwards).

Flying Plane, it's clear that the fact that all three losers lose equally in Solo favorises some perverse effects (once they understand they won't lose, the players may disconnect - which modifies sometimes largely the situation afterwards and it's unfair - or specifically attack someone just to punish him for their own loss, etc.); whereas the FFA structure also has bad effects (ppl sometimes play for 2nd; or simply play to avoid the 4th place at any cost. and of course it encourages the teaming beyond normality).
A good compromise could be: +4 -0.5 -1.5 -2 : in this case  to team up wouldn't be that effective; the difference between 2nd and 3rd, and between 3rd and 4th is not large; but it's still much better to be 2nd than 4th, etc.

here are two good examples showing how the presence of an unfavorable 4th makes the players deprive themselves and others of the chance for 1st:

red puts a checkmate on green so as not to risk taking 4th and receives a checkmate on the next move:
https://www.chess.com/4-player-chess?g=5984245-44;

the protracted stage of 4 players, in which yellow clearly dominates, where instead of playing blue/red vs yellow, they finish off green so that he is exactly 4th and destroy each other by gaining points so as not to be the last, as a result easy win for yellow: https://www.chess.com/4-player-chess?g=6189477-137
as long as there is even the slightest difference between 2nd, 3rd and 4th, players will continue to spoil the game in order to be at a minimum loss. therefore, in order to completely eliminate this in at least 2500+ games, I suggest Solo as the only option where possible

I also don’t understand why the player who took 2nd is better than 3rd or 4th. often the player who took 4th is the player who lost the team stage (which is influenced by his opp), and the 2nd is taken by the one who lost the most important Solo stage and ruined the game for himself and the 3rd, so I still hold the opinion, that a good option would be a system 3 -1.5 -1 -0.5 (#104-109https://www.chess.com/clubs/forum/view/possible-ratings-correction-to-stop-ffa-teaming?page=6)

I don't really agree. It's slightly more subtle, and depends on the chances' perceptions at any given moment, and on the risk aversion of the involved players which is a personal parameter. 

With pure solo (and same losses of the 3 losing places) yes there is no more concern of being 4th and yes ppl are pushed to play to win, and that's good. But it generates negative side-effects in the same time: a) some ppl resign when they realise that they won't win whatever happens (and this quite often makes the result unfair/irrelevant as some player who normally should've won doesn't win as someone else easily mates the zombie king for instance, or more often the pressure is released at some side and it influences the outcome unfairly), b) as ppl instead of winning focus rather at preventing another player from winning, it makes longer games as alliances become less relevant, and more generally it is not normal that in a multiplayer game ppl focus on making someone lose instead of winning first, c) as to get 2nd or 4th is the same, that deprives us of more interesting and subtle games where someone could be aiming at 2nd (if he understands that he cannot win anymore taking into account the material/points/time/position/psychology), whereas now for some players it's just a time waste. At least with 2nd place they would still be motivated to remain in game and to try to get the 2nd place. 

I reckon the most important is to have a huge difference between 1st and 2nd; and to have a significant difference between 2nd and 4th. Whereas the difference 2/3 and 3/4 should be small. Hence the proposal above. With +4 -0.5 -1.5 -2, it's still much better to be 2nd than 4th; and it's no use to help someone too much as you lose in rating being 1nd; and it's not very bad to be 4th comparatively to 3rd, etc.

@ Indipendenza

a) some ppl resign when they realise that they won't win whatever happens ...
usually this is due to the fact that the player with less material has weakened the one who resigned too much, so you can only blame yourself
b) as ppl instead of winning focus rather at preventing another player from winning ...
this problem can occur in any rating system
c) as to get 2nd or 4th is the same, that deprives us of more interesting and subtle games where someone could be aiming at 2nd ...
these are the games that spoil the game for the 1st, because the players give up the game too early changing the aim on lower

in any case, any rating system has its drawbacks, it is better to confine ourselves to considering the systems that we know. this topic is about the Solo and FFA systems, the existence of which separately does not make any sense, since none of these systems does not solve the problem of cooperation at the stage of 4 players, so need to look for something more radical. but before moving on, need to optimize what is already there

Yes I agree that we MUST re-merge FFA and Solo to Individual. The only real question is HOW.

I still believe the former system was the best (and yes there are drawbacks).

Besides your graphic above inspires me something. Couldn't we imagine a much more subtle system, where there is NO fixed parameters, but the system very PROGRESSIVELY passes from pure FFA to pure Solo depending on the average rating of the ppl involved (but, I insist, with only the games where the max. difference in rating between any 2 players involved is less than 200, otherwise it's discounted significantly). I.e. the coeffs evolve as per some formula like:

1st   2.5+0.5k

2nd   1-2k

3rd   -1

4th  -2.5+1.5k

k being 0 under 1800, then evolving from 0 (when it'll be 2.5 1 -1 -2.5) to 1 progressively up to 2300 (when it'll become +3 -1 -1 -1).

if I understood you correctly, then here is a similar option: #11 and implementation #29: https://www.chess.com/clubs/forum/view/rapid-ffa-solo-and-blitz-ffa?page=1

Yes, not exactly the same figures, but exactly the same spirit!

In my point of view, the rating system and the points are fine the way they are. You can choose between solo or FFA, I don't understand why coming out with a restriction rather than an addition. Why forcing players to play only solo, or only FFA, when they can choose now ? 

I don't think any merge of any types should be applied.

However, I would be interested to have some 1 table tournament using the  8 or 12 games systems I proposed on another post. The concept would be to have a 8 or 12 games tournament that involve the same 4 players, playing a round of 8 or 12 games, where everyone is the same amount of time playing the same colors, with different opponent every times. 

FFA + Solo = SFA!!!: This change is exactly what I had envisioned. I understand the graph. Low-rated FFA games are utter BS... and low-rated Solo is worse... and what's worse than these two is high-rated FFA games with someone playing for 2nd place. So this new change, in my opinion, solves all three problems.

Go for it. The only danger is that if I don't like it I decide not to play, but that wouldn't be a big change .

I think the choice of lower rated FFA to higher player Solo is perverse and will degrade one of the principle draws of Free For All 4PC for strong players.

As a beginner player I was confused by the shifting alliances and loyalties.  Now as a more mature player I not only understand what this is about, I understand the point of fluid alliances (of the FFA variety, which is qualitatively different from the permanent alliance in Teams).  
There are several points here.
1.  Weaker players DO NOT UNDERSTAND, and/or ARE NOT EQUIPPED, to make proper use of what I call the "social dimension" of FFA.  This only comes with maturity in the game and by my observation tends to kick in around 2000 ratings, which just happens to be where my own rating floats -- most of the time in the 2000-2150 range.  The sudden collapse of understanding of this dimension <2000 has led me to adjust my game filters to play only 2000+ games whenever I can, because I dislike being in a game where players don't know what's going on on that level.  You get accused of bad stuff by people who evidently misunderstand this element of the game (and the no collusion rule) and don't know how to respond when opponents do understand this, join in 3-on-one pile-ons when it's clearly in their interest to do the opposite, and so on.
2.  Above 2000 the social dimension looms larger and makes the game an interesting world of shifting loyalties and negotiation of tactics, guesswork and reading signals.  It becomes a game of complexity and rich dynamic that is far beyond both regular chess and solo, which tends to be entirely exploitive in any cooperation, or involves occasional perverse cooperation in which a player understands cooperation is not in their own interest but does it anyway, for reasons less germaine to the on-board game.  Or -- and this is what I predict will happen in higher level games -- it becomes a much simpler social dimension in which every 4-player endgame is a 3-on-one drag-down-the-leader game making anyone in a commanding position the target of withering opposition that they cannot withstand 95% of the time.  It will lead to perverse strategies such as holding back on point acquisition to a flurry at the last moment, and you'll see "slowest wins" races like the bike races in the olympics where all riders hold back for a mad dash at the end and creep along at a snails pace to avoid taking a lead and putting others into their slipstream.  I have no taste for such games.  I think it will degrade the quality and spectator interest of the high level game. 

3.  It essentially erodes the notion that FFA is a distinct mode from both teams and Solo, suggesting that FFA as we currently know it is only for beginners and weak players.  As I pointed out in point 1, lived experience with the game suggests the opposite, it is a dimension best explored by players with the highest-level skills.

4.  The gradient -- and any variant of this gradient accomplishing the same ends -- will make it inadviseable to play against opponents of very different rating levels, either upwards or downwards.  In current play some of these are my favourite games; I learn the most from games in which there is a large spread of ratings and the social dimension makes such games quite interesting, particularly where top & bottom rated players end up opposites.  This will throw an unpleasant wrench in such games:  the average rating of the other players may put one into a game with unfamiliar scoring conventions -- which has the corollary effect of unfamiliar on-board tactics.  The game itself will have a different feel, on top of the problem of being the balrog (or runt) in the game, both of which makes one a target.  It's a double-whammy.   It will create a pressure to want to play only other players close to one's own rating.

So my vote would be against this form.

On the other side, the *only* argument I'm seeing here for the change is 

a.  It's a simplification.  (I don't get why one would wish to simplify for the more advanced players but not the starters.)

b.  Beginning players do best in the "kinder" FFA mode while advanced players do best with "higher stakes" solo. (As I've argued this runs contrary to "lived experience", and I think it's killing the golden goose of advanced social dimension.)

You ask if we have alternative suggestions.  My suggestion would be to leave the present scoring defaults perhaps with minor adjustments to address small absurdities or ratings drift and to introduce new options allowing players to play along such a scoring gradient.  Perhaps solo/FFA could be a single game option instead of separately ranked types, with a gradient slider in the game creation window allowing one to select how far along the spectrum from FFA to solo one wishes to play.  I realise this may create new problems of reconciling rating systems, but if one throws FFA and Solo into a single cauldron long enough I think it will reach a natural stasis and after a settling out period it will begin to make natural sense.  I don't see that this makes that much of a difference ratings-wise, compared to the dramatic options one can already choose, within FFA and Solo chess.

Also, I think some better system of explaining the collusion rule should be used so that there is less confusion about it.    Also the game would benefit from some low-level tutorial material designed to help beginners understand the distinction between collusion and cooperation and the reasons behind why one might choose to form a temporary alliance or break it.  I don't see the sense in waiting JUUUUUST until players beging to understand the social dimension, and then take away the reward structure that encourages its utility.

Teams, of course, should never be conflated with the others.

Finally, here's a suggestion that hasn't to do with how ratings are calculated but as to how they are used:

When matches are made, I don't understand why the algorithm doesn't automatically do the obvious, namely place the highest and lowest rated players opposite. (The other player placements around the board can be randomized.)  This would go a long way to avoiding the most common type of unbalanced game.  My recommendation is to make that the first step of the game formation algorithm.

Thank you for your contribution, Robert , the problem being though that 95% of the millennials who are numerous here will not read because they are unable to.

To begin with, the system is not new; 3 years ago all games above 1550 (today it's 1850 as 300 points have been added to all in the meantime... Plus add easily 200-300 for the inflation that added since, 1550 then was therefore about 2100 today!) were AUTOMATICALLY WTA (if ALL players were above, the game was Winner Takes All, i.e. Solo in today's terms). I.e. we just revert to the old system (which was good), but with progressive shift from FFA to Solo rather than having an abrupt threshold (which is also good probably).

1.  Weaker players DO NOT UNDERSTAND, and/or ARE NOT EQUIPPED, to make proper use of what I call the "social dimension" of FFA.  This only comes with maturity in the game

YES you are right of course.

has led me to adjust my game filters to play only 2000+ games whenever I can, because I dislike being in a game where players don't know what's going on on that level. 

You will realise that it's the same later with 2200, 2500, etc. Different layers correspond to different mentalities. Just an example: after 2200 it becomes absolutely obvious that to weaken/attack/kill/fail to save the opp is simply stupid, counterproductive, as it gives you eventually the 3rd place usually. But as I find out now, after 2600 it is less obvious and some Top 20 players do not hesitate to kill you being in front if they dislike your moves/openings/ideas.

It becomes a game of complexity and rich dynamic that is far beyond both regular chess

Yes, 100%, ; I believe 4p chess to be much closer to poker than to 2p chess; it's extremely psychological and there are so many complex interactions that it makes it subtle and tricky.

in which a player understands cooperation is not in their own interest but does it anyway, for reasons less germaine to the on-board game.

Could you explain please?

Do you mean that some players will cooperate in some cases by inertia? Or that they will do it for the sake of cooperation in se? Because that's not what I see (I'm 2577 currently). Strong players (2600+) tend to stop any cooperation when appropriate, and change alliances every 2 or 3 moves sometimes, according to the current and mobile situation (I mean of course the 2nd stage of the 3 FFA stages).

It will lead to perverse strategies such as holding back on point acquisition to a flurry at the last moment, and you'll see "slowest wins" races like the bike races in the olympics where all riders hold back for a mad dash at the end and creep along at a snails pace to avoid taking a lead and putting others into their slipstream.  I have no taste for such games.  I think it will degrade the quality and spectator interest of the high level game. 

You're right to some extent, and many high-level Solo games are like that sometimes. But in fact it's also true for FFA, that high level players will play like Solo quite often because beyond 2500/2700 pts players are not interested in playing for 2nd.

3.  It essentially erodes the notion that FFA is a distinct mode from both teams and Solo, suggesting that FFA as we currently know it is only for beginners and weak players.

But that is not totally wrong! In fact if you look into the leaderboards and stats of the games played it becomes clear that Solo tends to be played in average by much stronger players, and that's why the average ratings in Solo are much lower: there are very few "fresh meat" points to take, there are much less farming and there is no free lunch. Whereas in FFA there is a huge inflation.

This will throw an unpleasant wrench in such games:  the average rating of the other players may put one into a game with unfamiliar scoring conventions -- which has the corollary effect of unfamiliar on-board tactics. 

Yes, true, but anyway the games with large spreads in levels (I would say, beyond 200 points) are much less relevant and quite often tend to be b...sht.

And people whom you describe just have to be cautious when they join a game; if you look into the level it suggests (2100+, 2200-2450, -1900...) it will already give a very good indication of what to expect.

if one throws FFA and Solo into a single cauldron long enough I think it will reach a natural stasis and after a settling out period it will begin to make natural sense.

That will precisely be the case with the changes they will implement. (And besides, you will see that it will still be feasible to launch a game "purely FFA" or "purely Solo" for the hard adepts).

As of today, to see TWO separate ratings for Individual games (as opposed to Teams); to see that quite often strong players don't play enough what they should (=Solo); to see that there is a clear lack of games because people are spread between too many modes and variants, is not satisfying. I've repeated for more than 2 years that FFA and Solo should be re-merged as it used to be the case in 2018, and I'm glad they'll finally do it. (I'm much less convinced by the fact of inverting K/Q for Blue and Green, but it's another topic).

Also, I think some better system of explaining the collusion rule should be used so that there is less confusion about it. Also the game would benefit from some low-level tutorial material designed to help beginners understand the distinction between collusion and cooperation and the reasons behind why one might choose to form a temporary alliance or break it..

Yes, you're right.

But in fact there are 1515 threads about it. The issue being rather that the beginners won't read them.

When I discovered the game, 3,5 years ago, I would've adored to see something like https://www.chess.com/clubs/forum/view/basic-ffa-aspects. Instead, I had to spend 3 years and to learn by experience.

I know that the majority of reports made to the admins are merely b...shit as it concerns precisely the teaming, that players under 2000 do not understand (yes, it was also my case).

When matches are made, I don't understand why the algorithm doesn't automatically do the obvious, namely place the highest and lowest rated players opposite. (The other player placements around the board can be randomized.)  This would go a long way to avoiding the most common type of unbalanced game.  My recommendation is to make that the first step of the game formation algorithm.

Interesting proposal.

About 2 years ago I proposed instead to TAKE INTO ACCOUNT the position when giving the final points of rating. To win when you have a very strong opp is much easier than to win having him as one of the sides, especially when he is left hand.

So, I don't believe this change will effect under-2300s a whole lot.

This is my bet on what will happen. 

Generally, I think this will accelerate "the best strategy", and an important question is: "is that meta good or bad for 4pc?"

More specifically, (POTENTIALLY) I believe players (2500+) will

A. Become better, and experience more stalemates

B. Realize that whoever has the most points (or can get the most points) at the beginning of the 3 player stage has a huge advantage

C. Adjust accordingly

But would C. Be good for the game?