I like it. A little draconian, perhaps, but if it encourages treachery, I'm in favor.

I’m curious what style of play you’re trying to encourage.

Also, if the game is still zero-sum, then the final point values are the same.

It stinks - it's not zero sum as it halves the original stake; I will aalways NEVER play a game that is rated that way. Also "In other words, a 2-way draw can at most have 25% of the total pot, with 50% of it being reserved for the solo victor." is nonsense - if it's a 2-ay draw their is a NO sole victor. Been at the giggle juice, Scooby?

Damn, sorry for typos - 5 a.m. and no autocrrect!

Yeah, I don't like the zero-sum aspect either, but it would be cool for a tournament where the "extra" points go to solos. Hard to implement in vDip though...

@nopunin10did, Draw-Disvaluated intends to encourage solos without the harshness of Draw-Zero.

@chumbles

The game is indeed not zero-sum, but who claimed it was?

The point of DDS is that a 2-way draw is *not* worth half a solo. A solo is the ultimate objective and should be way more rewarded - which is solved by reserving 50% of the pot apart.

So, it's also not a constant-sum scoring.

I must add that this system was invented by Thaddeus Black, intercontinental Diplomacy champion who posted it first in Diplomacy World #133.

To be clear, I wasn’t comparing it to Draw-Zero. I was trying to state that, in order for the system you describe to act in a zero-sum manner, the mathematics of it yields a result no different than standard WTA or DSS.

And if you left the system described intact without converting it to zero-sum, you would be left with a negative-sum system. Half the pot bet at the start of the match would simply disappear. Such a system rewards players for playing fewer games in the long term.

As for its use in tournaments, those games don’t have to be zero sum because there’s no “bet” at the beginning, as there has to be in an unstructured environment such as this.

Alternatively, what one _could_ do to achieve something similar to this scoring system would be this:

If the game ends in a draw, take half the pot and refund it, splitting it evenly between all players (including those who were eliminated or opted out of the draw).

Split the other half of the pot evenly between all draw participants.

I think this is a bad scoring system.

First of all, all games on this site are zero-sum, and I think that they should be. I have a different look at solos. I see a solo more as a failure of the survivors than as an achievement of the winner, and it is a fact that the worse the players are on average, the more likely it is that someone will solo. Giving a higher total payoff for games in which someone solos does not make much sense to me.

The suggestion of nopunin10did in the post above me makes more sense to me since its zero-sum, but it still would be very weird to have. It would incentive players to stop a solo EVEN if they KNEW that would cause their elimination. This is just as strange as PPSC scoring, where players are incentiviced to take more supply centers, even if they know that will cause someone else to solo. Also, a side-effect of giving all players, even losing ones, an extra reason to help prevent a solo from occurring, might decrease, instead of increase, the number of solos that you would see.

If you really want to encourage solo attempts, then one should have a scoring system that, in case of a draw, gives a higher payoff to players that are closer to achieving a solo than to other players.
Consider two scenario's where some player (let's call him Bob) plays a classic game.
- Scenario A: Bob plays for the draw from the beginning onward. He allies with 3 other nations and eliminated the other 3 nations with his allies. After that, all players are roughly equal in size, and everyone draws.
- Scenario B: Bob attempts a solo. He comes very close; in fact, he reaches 17 supply centers. But he gets stalemated by the 4 surviving players on the other side of the stalemate line (none of which is very big in size) and has to accept a draw.
Under the scoring system we currently use at vDiplomacy, Bob would gain more points in scenario A than in scenario B. A better scoring system would give Bob a higher payoff in scenario B, as that scoring system would encourage solo attempts. There does exist precisely such a scoring system, and it's called Sum-Of-Squares Scoring. I am not in particular favoring that scoring system, as it has its own problems, but I think this is the right way to think about scoring systems and how to improve them.

@Mercy: Also C-diplo rewards getting closer to a solo, but it also adds an interesting conflict for the defenders as there's a bonus also for second or third most SCs. I too don't want to go as far as to say that there is one-fits-all scoring system, but this one does add an extra twist to the stalemating game that a skillful player might exploit to steal a solo. Therefore I would argue that C-diplo promotes solo attempts just as much as SoS, if not more!

I've recently been playing on the Conspiracy! platform, a Diplomacy app that's available on Android devices. One of the things I like about it is that it factors in an ELO rating system, but that there is only ever three outcomes to a game Win-Draw-Defeat. No survive, no points for second place, no points for assisting another player to win, or failing to change course and join forces with previous enemies against a solo attempt.

If anything the whole WTA,PPSC,SoS etc argument is just adding complexity when there really doesn't need to be one. We should be able to simplify the whole ELO rating element here. The exceptions being variants that are deliberately unbalanced like Fubar or Renascimento which should be exempt from the ELO system of scoring

@d-ice: Sure, C-diplo would be a reasonable scoring system as well.

@kaner406: I have also played on the Conspiracy app (name: Genade) and I have to inform you that you are misunderstanding the Conspiracy scoring system. Conspiracy uses PPSC as a measure of your performance in a game. In fact, if you win with more than 18 supply centers, your performance is rated better than if you win with precisely 18 supply centers. Your PPSC scoring, together with your ELO ranking relative to the other players, determines how your ELO ranking will change. This can go very far: If your ELO ranking is far higher than your opponents, and you win with not much more than 18 supply centers, you can loose ELO ranking.
Also, I do not understand what you mean with your talk about the 'WTA, PPSC, SoS etc argument just adding complexity' and that we 'should be able to simplify the whole ELO rating element'. By definition, any ELO ranking operates in accordance to SOME scoring system.

Well I'm wrong regarding how the ELO system works then, but all the same, why not get rid of the points needed to join a game entirely, and restrict the amount of games a player can join based on their reliability? By removing the need to use points to buy into a game you are making a more pure version of the ELO system, that operates in the background and not as a dual use ppsc/WTA creation that grew out of a semi flawed inheritance from webDip. Getting rid of these dual (actually 3 types of game creation of you count unranked games) points generating system and basing something on a win-draw-loss scenario seems like a step in the right direction to me.

@kaner406 raises a good point. Points are supposed to determine both 1) the best players and 2) the "level" of a game. However, vDiplomacy points already are more well-regarded to decide the best players, and reliability can work just as well to decide who can enter a game.

Still, there's always Unranked. Some people do like the thrill of points!

@d-ice & @Mercy

There's parallel conversations about rank-based systems going on now in both this thread and another:

http://www.vdiplomacy.com/forum.php?threadID=76569

I have a description in that thread of how one might adapt a Carnage-based system to vDip. Carnage is a rank-based system that works on the principles of C-Diplo and has become very popular in US-based face-to-face tournaments.

I'm having trouble finding the full official Carnage writeup, but it's an iteration on Dave Maletsky's "Win Tier" system that replaces the "tier" concept with a simple solo-winner-takes-all-points mechanism.

http://www.world-diplomacy-database.com/php/scoring/scoring_class.php?id_scoring=14

Ok @nopun, but really the “solo-winner-takes-all-points mechanism” is default in almost all scoring systems I know of... the interesting part is how the system handles forced and agreed draws.

@d-ice

Agreed. The only quirk about rank-based systems is that if a solo winner still takes _all_ of the points, it means that an eliminated player's score isn't known at the time of elimination. Whether they get any points at all depends on whether the remaining players end in a draw or solo; this is the same problem with the hypothetical zero-sum variation of draw-disvaluated-scoring.

I've been working on the mathematics for a zero-sum rank-based / draw-size hybrid that would combine some of the merits of each. For a standard 7-player game, it would award 85% of the pot to a solo, rather than 100%, to make room for some menial consolation prizes that would be awarded based on the degree of loss (and would still represent a net-negative point change).

Correction: ~79% of the pot in a seven-player solo, not 85%

Ok, I see your point but personally I don't think it's that much of a problem that eliminated players can get different scores depending on what happens. Consider the 2-way draw in C-diplo, 5 eliminated players would then share third place bonus. Is it reasonable that eliminated players get more points in a two-way than a three-way? No. But the point is that there are probably so many different special cases that apply to different scoring systems that I don't think it's worth bothering. I'd turn it around and say that solo-winner deserves the 21% extra points, and players that have been eliminated are just as guilty as the survivors for creating the oppotunity to steal the solo. Therefore 0 points is reasonable.

A fair assessment. Either a C-diplo or Carnage system would work well here as an alternative to current systems, though I think Carnage probably scales better with varied player counts.