Play Diplomacy Online

I've made an attempt to sum up everything discussed so far. I've noted who's contribute/agreed with what and made an effort to link any posts with big suggestions so you can read the original.

We started with:
Results Table Available (mapmaker)
can view, save, print; can't edit
Changes:

• win/loss to win/played (InterMPC); done
• InterMPC given edit access (technical); done

Suggestions:

Sort by specific value (raphtown); can only be done with edit access
Sort by Total Points (QueenofHearts) AND
Create table of multiple scoring possibilities & stats (valent); this led to a discussion of SCORING

The following were all raised and generally agreed on:

• Should encourage trying to win (InterMPC)
• Should reward/penalize the unexpected result (InterMPC)
• Encourage to play more games (InterMPC); worried about rewarding players who play a lot too much (raphtown)
• Needs to be easy to update and understand
• Post to Explain Rankings System (Ceebs)

These issues need to be settled:

• Draw understood as win (fellowes) vs draw (bb82, echotwo, InterMPC)
• Never lower rating on draw vs straight scoring (fellowes)
• Draw if survive (GhostEcho) vs part-of-draw (InterMPC)
• Win streak bonus (QueenofHearts)
• 2-player draws more points than other draws (AndyCooke, QueenofHearts) vs just a draw (echotwo, bb82)
• Solos more than twice 2-way (Mapmaker) vs scoring all losses the same (fellowes)

A completely new scoring system was suggested early on, specifically designed for Classicist games: Cumulative points for win based on sum of skill of opponents (InterMPC); does not adapt to ELO

0 for loss; for winners
2 vs bronze
3 vs silver
4 vs gold
5 vs platinum
+10 for solo

After a brief discussion of a homebrew ladder system (InterMPC, raphtown, Iamnobody, fellowes)

Rating given 0-100
challenge possible
Penalty for inactivity
Gray boxes here

It was generally agreed that adopting the ELO system would be best (Purple boxes here).

However, this still left the question of how many points should be available and change hands, or the "weighting" of the system for each result. The following systems have been brought up. NOTE: For ELO to work, these would need to be adjusted (if necessary) so the max "weight" is 1.

• Site System (in passing, by most)
• 3/1/0 system a la FIFA, hockey, etc (Mapmaker, raphtown); this is essentially the same as
• 1, 1/7, 0 as base (InterMPC); intermediate draws weighted to suit; cf Chess 1, 1/2, 0)
• New System (bb82). Suggested independently; suggested to use Fading Echoes; I give conversion to ELO weight as well.
  0 for loss (ELO 0); then
  6/7 player draws 0 (ELO 0)
  5 player 1 (ELO 0.1)
  2-4 player 3 (ELO 1/3)
  solo 10 (ELO 1)
• fellowes, as noted above, suggested two different weighting systems. The first, designed to never lower rating from a draw, is compared here (as FRS) to yet another weighting system, I think InterMPC's original ELO approximation (given as ELO).

NOT ADDRESSED YET: This may just be me being dumb, but who is going to be responsible, and how, for maintaining an active ELO database.

Let me know if I've missed anything.

I agree with all that is being said by echo and bb82 here about draws and their value. However, why is the discussion being had? Too much talk with the classicists and not enough action.

As a follow-up, I'm just going to give some of my own views quickly.

I'm all for implementing ELO if possible, and I think we should go with 1, 1/2, 1/3, 1/4, 1/5, 1/6, 1/7, 0 weighting: split the point among all winners.
I'm in favor generally of implementing DIAS whenever possible.
I'd like to see The Classicists including (potentially) forum games, and as such I'm less bound to "site scoring".

Final question: how to score variants? 7-player variants can be scored just the same; but I'd suggest not scoring variants.

lannes wrote:I agree with all that is being said by echo and bb82 here about draws and their value. However, why is the discussion being had? Too much talk with the classicists and not enough action.

Too much talk! This is Diplomacy!

But really, there's lots of talk because not everybody agrees.

I don't agree with "too much talk", but I do agree with "not enough action".

Now that GhostEcho has succinctly summed up the argument I think some "leader figure" should decide how we move forward from here. Whether we have a vote for the best system, or it is decided on by said "leader figure" or whether a panel is selected to come to a consensus on the best solution.

What ever the outcome is, I can build it if needed and will make it as automated as possible. ie mapmaker enters the results of the games and the spreadsheet works out all changing values.

here here, let us just put it to a vote and we will see what the majority wants.

Ok, but what are we going to put up to a vote? The individual issues? Which voting system we will use?

I'm strongly against a system that rewards for playing a lot, otherwise 'rank' will just boil down to who can play the most.

Reward/penalize I don't like. All it takes is one 'bad-apple' to make a solo extremely easy. I feel solos get too much as it is, let alone adding extra for multiple solos in a row. Solos should be extremely rare when all players are decent.

I also don't like the idea of penalizing for losses or inactivity. It would be destructive to the community.

bb82 wrote:I feel solos get too much as it is, let alone adding extra for multiple solos in a row. Solos should be extremely rare when all players are decent.

I'm not sure your argument's entirely coherent. On the one hand, you say that solos should be extremely rare. On the other hand, you feel they get too much reward. Too much compared to what, exactly?

The rankings on this site are already skewed in favour of draws, in the sense that by accumulating n n-way draws, one can gain the 'equivalent' 'reward' to a solo, in terms of rankings points. In many ways, I think straightforward W/D/L tables (absolute and percentages) might make the most sense, with rank determined by the percentage of wins, draws then used solely as a tie-breaker.

echotwo wrote:

bb82 wrote:I feel solos get too much as it is, let alone adding extra for multiple solos in a row. Solos should be extremely rare when all players are decent.

I'm not sure your argument's entirely coherent. On the one hand, you say that solos should be extremely rare. On the other hand, you feel they get too much reward. Too much compared to what, exactly?

They should be extremely rare when all players actually know what they are doing. The problem is that in most games there is usually a 'bad apple'. With so many players that have such different skill sets of varying degrees, solos happen way too often to make any sort of ranking system too meaningful.

Or in the extremes, some will purposely exploit much lesser players solely to rack up some easy solos. Whereas others might actually be looking for challenging games, thereby if finding them will rarely net solos.

So who do you think the better player is? The one that plays easy games or the one that plays challenging games? That is why any sort of ranking system won't be too meaningful, not with 7 players. In chess the standard system works well because you know the skill of the other. But all it takes is one 'bad' player in diplomacy to skew the results for all.

A simple w/d/l would also be useless for the the above reasons. Even if I implement a system of my choosing, it would still be far from perfect. But it would be better than the sites current system. In diplomacy no ranking system will be close to accurate. The best you can do is try to do broad compromises that incorporates many different aspects of the game. The site's ranking system just needs some tweaking.


If there is a solo, then that usually means that at least one bad player is in the game. Keep in mind that I'm not saying this is a bad thing as it keeps games interesting. I'm only saying that ranking systems are far from accurate.

bb82 wrote:

echotwo wrote:

bb82 wrote:I feel solos get too much as it is, let alone adding extra for multiple solos in a row. Solos should be extremely rare when all players are decent.

I'm not sure your argument's entirely coherent. On the one hand, you say that solos should be extremely rare. On the other hand, you feel they get too much reward. Too much compared to what, exactly?

They should be extremely rare when all players actually know what they are doing. The problem is that in most games there is usually a 'bad apple'. With so many players that have such different skill sets of varying degrees, solos happen way too often to make any sort of ranking system too meaningful.

Or in the extremes, some will purposely exploit much lesser players solely to rack up some easy solos. Whereas others might actually be looking for challenging games, thereby if finding them will rarely net solos.

So who do you think the better player is? The one that plays easy games or the one that plays challenging games?

How effective do you think the pseudo-Socratic method is at conveying the impression that you're not actually interested in having a conversation with your interlocutor?

That is why any sort of ranking system won't be too meaningful, not with 7 players. In chess the standard system works well because you know the skill of the other. But all it takes is one 'bad' player in diplomacy to skew the results for all.

A simple w/d/l would also be useless for the the above reasons. Even if I implement a system of my choosing, it would still be far from perfect. But it would be better than the sites current system. In diplomacy no ranking system will be close to accurate. The best you can do is try to do broad compromises that incorporates many different aspects of the game. The site's ranking system just needs some tweaking.

If there is a solo, then that usually means that at least one bad player is in the game.

[citation needed]. And it won't do to cite the proportion of the recent games that ended in a solo that involved a 'bad player' (however you define a 'bad player') - you need to show that that proportion is significantly higher than the proportion of all games that involve a 'bad player'.

Of course solos 'shouldn't happen'. That's the beauty of the game: it involves setting out to do something that should be almost impossible. And sometimes getting to a position in which you can't be stopped from doing so. Of course, there are likely more solos on this site due to bad play (or NMRs/surrenders) than anything else, but don't ignore the extent to which bad play can be induced, and indeed that not everyone is skillful enough to take advantage of it. Even if there's a game involving one 'bad player', it takes good play to be able to make sure that you're the one who can take advantage, and indeed to ensure that the 'bad player' is in a position where you can take advantage in the first place.

If you're saying that wins against poor opposition or NMRing or surrendered opponents shouldn't be rewarded anywhere near as highly as those that come against committed opponents who are themselves trying to win, then I agree entirely. It's a point that I and others have made, repeatedly. But to claim that solos are in themselves over-rewarded is I think ludicrous. All other things being equal, it's draws that are over-rewarded.