"of course your friend will backstab you if he get the chance."
If he does, he is usually, though not always, sacrificing his own standings for ... usually not much. Most of the times when 2 people across from each other attach each other, the real beneficiaries are the other 2 people. I would go so far as to say that attacking your friend is like losing your own pieces.
I have to agree with you @Omega60, at least to some extent. When I made that statement I had not actually realized to what extent your friend mattered. I was relatively new to the game.
A couple of things I've noticed.
My rating tends to fluctuate between 1300 and up to 1600. Here are my thoughts.
The higher the rating, the more defensive people play, and the more pawns get shoved to the center of the board.
Getting an initial extra queen does not seem to be that helpful. I think it is because it makes you too much of a target, and this more than outweighs the benefit of the queen. By contrast, if someone resigns - GO FOR THAT KING. You may be a target, but 20 points is a lot.
Getting 1st or 2nd gets you points. If you are in a bad position, work to help whoever is strongest with the idea of going for second.
Sometimes it's easier to play against stronger players. Less randomness.
At higher levels, chess skill, especially knowing how to recognize mating combinations, becomes more useful and important.
Getting an initial extra queen does not seem to be that helpful.
At the current value of promoted queens, it is helpful, a lot. Do not withdraw your new queen back in your base. Keep her in the center, protected by a pawn, preferably in the diagonal of the fianchettoed bishop of your backstabber. She will cast deep fear in his heart for the rest of the game!
Getting 1st or 2nd gets you points. If you are in a bad position, work to help whoever is strongest with the idea of going for second.
This is indeed a legitimate strategy right now. I have made a suggestion that would reduce the attractiveness of this strategy though. I proposed that the winner's rating, if he also happens to be the last man standing, is boosted by +25%. His extra rating points are going to be deducted from his opponent's ratings.
Sometimes it's easier to play against stronger players. Less randomness.
I can confirm this from my experience too. Players above 1600 are generally aware of what is happening on all regions of the board, and they don't miss the chance to engage in backstabbing operations presented by their best friend. But try climbing above 1900 and the randomness reappears with a different face. Your best friend will willingly betray you, and ignore the opportunity of exploiting your sacrifices, if he judges that your common enemies have been weakened enough.
I can confirm this from my experience too. Players above 1600 are generally aware of what is happening on all regions of the board, and they don't miss the chance to engage in backstabbing operations presented by their best friend. But try climbing above 1900 and the randomness reappears with a different face. Your best friend will willingly betray you, and ignore the opportunity of exploiting your sacrifices, if he judges that your common enemies have been weakened enough.
This actually sounds appealing. And much better than... a person screwed me over, especially in a tactical situation where I am working with someone else in a joint attack, but also screwed himself royally in the process. So what you are implying is that awareness of the power relations between the 4 players increases as rating increases. Cool !!
Getting 1st or 2nd gets you points.
You get some points as the 2nd most of the time, but not always. If you are 1900 and play with 3 1500 guys (which still happens quite often) you'll lose points being the 2nd. (It gives you 2 small "+" and then 1 large "-".)
But if the rating of all 4 players is similar (+-100) the 2nd place gets you points.
But try climbing above 1900 and the randomness reappears with a different face. Your best friend will willingly betray you, and ignore the opportunity of exploiting your sacrifices, if he judges that your common enemies have been weakened enough.
Even worse. I have a personal list of very strange 1900+ players. There are only 3 names in that list now, but it is going to grow fast.
Two of those players just never help the opposite player ("best friend" in @Martin0's classification or frienemy in the chat variant). Do not know how did they climb to 1900+, seriously.
And the 3rd guy from the list just plays to be the 2nd (not the 1st). Always. From the very beginning of the game he selects a target (a player at left or at right) and then exchanges pieces like crazy with that player, attacks him all possible ways even if he gives a rook for a knight etc. This way he helps his "best friend" to checkmate the targeted player, then they go together to mate the 4th player and finally the jackal gets +20 from his master after "Claim Win". And sometimes this works.
Ha ha! I can imagine that the playing style of the 3rd guy guaranties great fun for the most part of the game! 
If the other two people figure out what's going on, it should make for an interesting 2v2.
Interestingly, I agree 100% who your best friend is. But based on what I saw, not very often do people act this way.
Personally once a player in front of me was getting attacked and I could have checkmated him without any concessions. But i chose not to. It was early in the game. Because even 20 pts don't convince me to put myself between 2 players in the middle game, where i should have at least 25-35 points advantage to overcome this or something like 2 extra queens. People don't tend to see things the same way.
Many times tho opposite guy attacked me in addition to other two already, without scoring anything for him/herself (not check mate, not many points...) just to play difficult middle game against two with less points on scoreboard.
By contrast if one of the players on my flank gets eliminated, I instantly consider ''my best friend'' to be my worst enemy, so I normally just defend against the one in the middle and take what is for grabs without any concessions, but focus 75% or more on ''my best frined''.
That isthe basic yes. Ennemis to your flancks. ally in front. Then everything changes once one player dies.
Yeah, once a player is eliminated I think everything changes as well. Material, points and potential promotions are some of the things to take into consideration then when it comes to mutual interests. Also if your aiming for first or second place matters. Your king can become quite exposed if the player in front of you is allowed to promote several pawns on the same side as your king.
@rogerstar, you do not checkmate a player in front of you, because you are going to be sandwiched. (Quotation: "20 pts don't convince me to put myself between 2 players".) So, you think it is bad to be sandwiched, you expect 2 "bread" players to attack a "ham" (or "cheese") player between them and consider it as an optimal strategy.
But later when you discuss the 3 player middlegame you say: "I instantly consider ''my best friend'' to be my worst enemy, so I normally just defend against the one in the middle ..., but focus 75% or more on ''my best frined''". So, you do not focus on the sandwiched player. You, as a "bread" player focus on another "bread" player, not on the "sandwich filling".
Do you see a contradiction here?
To you, BabYagun!
Thank you for your reply.
Point well taken!
So you think that if other players would share my opinion about ''sandwich'' business, as you described it, I would have been in contradiction and therefore advised (by our common expressed knowledge) to
1) Checkmate the player across the board, and
2) Be the ham in the sandwich, counting on other two players to think the way as I do and be fine.
To be honest, my experience is that players most of the time don't think the way I do, but normally go against the ''ham''. Now look, even if I concentrate 75% on other ''bread'' player I could still completely destroy the ''ham'' in couple of moves, given that other ''bread'' player plays mostly against the ham.
Still as a ''bread' I don't take the ''ham'' as a great danger for victory (if doesn't have 30 pts advantage and 2-3 extra queens), so given that I want to win, I have to concentrate on the most serious candidate to win, that is other ''bread''. Sometimes 25% of force can be more deadly than 75%.
Look, the ''ham'' does not have bright future even in best case scenario (against 2 players with my mindset), bacause is just so easy to destroy (normally).
Contradiction - yes in words on paper, but not in practice, where:
1) Most people (my so far co-players) don't play as I do (after all 4player, above all, is fun and isn't it fun to just get at someone - which normally is ham)
2) Is usually not great to be in the sandwich even if the other two have the same approach as I do.
In short - I don't play the way I expect other players to.
(But I am not a very expert player - around 1500 most of the time)
@rogerstar, thanks for the detailed answer.
Now please compare 2 cases:
1. You are Red. Green was mated.
2. You are Red. Blue was mated.
Who is the most dangerous for you in 1.? Blue. Because he moves first, then Yellow can support his attack against you. Who is the most dangerous for you in 2.? Yellow. Because he moves first, then Green can support his attack against you.
In both these cases you are one of the bread slices. But the ham is more dangerous when it is at your left. So, there are different sorts of sandwiches.
@BabYagun not only there are different sorts of sandwiches, but they exist simultaneously on the same board. Because each bread player sees a different sandwich. One has the ham at the right, and the other at the left.
I'm not sure if the order the players move is the deciding factor there. Let's say you are red and all players castled kingside. Green gets checkmated by yellow. Now yellows pieces that attacked green are close to the red king. Yellow also wants to push his pawns on his queenside to promote pawns. Not only will this make yellow ahead in material, but it will create half open files towards the red king and the promoted queens are also close to the red king.
While the order of moves should make blue your worst enemy, he is stuck between 2 players and therefor have a disadvantage. Prolonging the game and play passively does not sound that appealing for him. Typically he should either focus on promoting pawns in the center while the others focus on the flanks, or he needs create weaknesses for his opponents, even if it means making himself weaker. If he decides to attack yellow, while weakening his own king, then he becomes closer to a friend to red than his worst enemy. Then both red and blue has mutual interest to attack yellow (especially if yellow was ahead in points after his checkmate).
If on the other hand, blue decides to attack red and does not care about yellow getting 1st place (blue only aiming for second), then things will be really hard for red.
I think trying to figure out mutual interests when there are 3 players left is more important than thinking of the order of the moves when it comes to which players to fear. Of course tactically positioning still matters.
From a balance perspective I see where you guys are coming from, but I am in favor of the current setup. It would be annoying if my king started on the same diagonal as an enemy queen and the way you suggest 2 queens would start on the same diagonal as my king. My pawn in front of my queen can be pinned early and if I move my pawn in front of my queen first I would be exposed to a potential check.
I like the current setup because the kings starting on the same diagonals does not make them exposed to each other.