webDiplomacy's interpretation of the victory rules

[Nescio]

The official Diplomacy rulebook (50th anniversary edition, 2008) is quite clear:

OBJECT OF THE GAME
As soon as one Great Power controls 18 supply centers, it’s considered to have gained control of Europe. The player representing that Great Power is the winner. However, players can end the game by agreement before a winner is determined. In this case, all players who still have pieces on the game board share equally in a draw.

Furthermore:

After all the orders have been revealed and read, the moves made,
and the conflicts resolved, any dislodged (defeated) units make
their retreat. These retreats are written down (just like orders) and
revealed immediately. No diplomacy or discussion takes place prior
to writing retreat orders—all countries are on their own.

and:

As with retreats, gaining and losing units (collectively known as
“adjustments”) are written and revealed simultaneously without
discussion or diplomacy of any kind.

webDiplomacy has its own, less strict interpretation. It distinguishes between alive (“survived”) and eliminated (“defeated”).
An example where a player (Germany, I, Nescio) survived despite having 0 units: http://webdiplomacy.net/board.php?gameID=356521
An example where a player (Italy, Kestas Bot) survived despite having 0 centres: http://webdiplomacy.net/board.php?gameID=356222
An example where a player (Austria, Jane) was included in a draw despite having 0 units or centres: http://webdiplomacy.net/board.php?gameID=356144
Strictly speaking these are violations of the official rules: in all three cases the player ought to have been defeated.
The first example could be addressed by considering players with centres but neither units, nor the ability to train any to be defeated; the second by performing the solo-victory check at the end of the build phase, instead of the start; and the third by not processing any draw votes during the retreat or build phases.

[teccles]

I agree that WebDiplomacy disagrees with the official rules on who is in a draw. Personally, I think the official rules are a bit weird here - if all my units are gone but I have a centre and am about to build, I've pretty clearly not been eliminated.

For survived vs defeated: both the rating systems on the site (points and GR) treat these the same, as do most players; both are a loss. So the distinction there is probably not super important (but it would probably make more sense to count people on 0 centres as eliminated).

[Nescio]

Please look at the examples: having no units and no home centres means you will never be able to build, regardless how many other centres you have. In the first example, my last unit was dislodged in Autumn 1907, yet I “survived” until the end of the game in Autumn 1909, because Denmark was still mine.
Yes, once a game is finished, both survivals and a defeats are losses. (Against bots I consider draws failures too: only solo-victories count.) However, as long as the game is not finished, the distinction between survived and defeated is super important: survivors are included in draws. Therefore it's wrong to count players with 0 centres or 0 units and no home centres as surviving.

[teccles]

Please look at the examples: having no units and no home centres means you will never be able to build, regardless how many other centres you have. In the first example, my last unit was dislodged in Autumn 1907, yet I “survived” until the end of the game in Autumn 1909, because Denmark was still mine.
Yes, once a game is finished, both survivals and a defeats are losses. (Against bots I consider draws failures too: only solo-victories count.) However, as long as the game is not finished, the distinction between survived and defeated is super important: survivors are included in draws. Therefore it's wrong to count players with 0 centres or 0 units and no home centres as surviving.

I agree that the distinction matters in the case of a draw.

I did look at the examples, and I realise that in the example you show, the centre isn't a home centre and you could never build again. But the official rule is 'no units', not 'no units and no home SCs' - so I think the official rule here is pretty odd.

Personally, I think the current rule of 'no SCs' is simple, and the outcomes are fine (obviously it's a matter of opinion, but I think drawing with Denmark and no units is perfectly sensible). 'No units and no home SCs' also seems fine, but it's not actually what the rulebook says and it's a bit more complicated.

[RoganJosh]

In the first two examples, the problem is that the site differentiates between "Survived" and "Eliminated," but the official rulebook does not. It would be better (now that PPSC is gone) if we'd relabel both to "Lost."

It is odd that the official rules has a winning condition which is based on supply center count, and a drawing condition which is based on a unit count. teccles example is still relevant here: by the official rules, a player with a home center open for a build, but no units, would not be included in a draw. That, of course, contradicts Calhamer's opinion that the players who share equally in the draw are those that theoretically could still win the game.

Of course, if you'd base participation in the draw on whether a player could theoretically win the game, then I'm not sure what to do with a game like http://webdiplomacy.net/board.php?gameID=242215

Anyway, I don't think the sites should aim at following the official rule, because it's a bad rule.

[Nescio]

“That, of course, contradicts Calhamer's opinion that the players who share equally in the draw are those that theoretically could still win the game.”
... which is exactly why I think players with 0 centres shouldn't count as survivors and shouldn't be included in a draw.

[Valis2501]

I am a very serious person when it comes to Diplomacy and I assure you this approaches "How many angels can dance on the head of a pin?" territory.

[Peregrine Falcon]

I personally find the tension between rationality and irrationality one of the more interesting parts of Diplomacy.
I like that you can include zero-unit, zero-home-SC powers in the draw. If a power without any units can convince everyone else to still let them live, they deserve that draw.

(And I do agree that 0SCs should count as eliminated... I thought that change had already been made, actually. I'd be for counting both 'Eliminated' and 'Survived' as sub-categories of 'Lost', although SWS/PPSC scores do still persist, so idk how that would be handled.)

[Foxcastle]

Doesn't this just come down to deciding whether to process one more set of orders? If the rules say that gaining and losing units (or centers, for that matter) only happens during adjustments phases, and you agree to draw before the adjustment phase, then you've made a choice to include them in a draw (and foregone the choice to exclude them by processing one more turn in which they would be eliminated).

[Nescio]

I am a very serious person when it comes to Diplomacy and I assure you this approaches "How many angels can dance on the head of a pin?" territory.

While the first example (surviving with 0 units) is unusual, the second (surviving with 0 centres) is not uncommon, it happened in in 7 out of 60 games I played vs bots. And both can lead to players being included in draws in webDiplomacy, as can be seen in the third example.

I like that you can include zero-unit, zero-home-SC powers in the draw. If a power without any units can convince everyone else to still let them live, they deserve that draw.

It's not like you can properly negotiate with bots.

Doesn't this just come down to deciding whether to process one more set of orders? If the rules say that gaining and losing units (or centers, for that matter) only happens during adjustments phases, and you agree to draw before the adjustment phase, then you've made a choice to include them in a draw (and foregone the choice to exclude them by processing one more turn in which they would be eliminated).

If everyone voted for a draw prior to the retreats and build phases, then the game should end as a draw prior to the retreats and build phases. There is a good reason why the official rulebook says no diplomacy of any kind during processing orders, retreats, or builds.
The trouble is webDiplomacy allows players to change their votes at any time, and more importantly, it processes draws during any phase, before checking defeats or victory. Here's an example of a game that was drawn, despite one player having 18 centres (yes, seriously): http://webdiplomacy.net/board.php?gameID=349971 That shouldn't have happened.
As I wrote in the opening post:

The first example could be addressed by considering players with centres but neither units, nor the ability to train any to be defeated; the second by performing the solo-victory check at the end of the build phase, instead of the start; and the third by not processing any draw votes during the retreat or build phases.

[teccles]

I think that the draw with 18 centres is a quirk of the rule that bot games draw when you get eliminated - it is impossible in human play. This game where Austria drew in your opening post actually also looks weird - the orders say the retreats happened, but on the board Vienna is still red and Austria has 1 centre. Maybe I'm missing something, but I can't see how that can happen - according to my understanding Austria would be eliminated immediately after retreats. I wonder if this is another bot-specific problem, where a bot somehow votes draw after retreats process but before centres change hands?

It's worth noting that for full press games WebDiplomacy deliberately diverges from the rulebook on press during retreats/builds (which I think is very sensible for asynchronous play). So I think it would be odd not to allow draws during these times.

[Nescio]

This game where Austria drew in your opening post actually also looks weird - the orders say the retreats happened, but on the board Vienna is still red and Austria has 1 centre. Maybe I'm missing something, but I can't see how that can happen - according to my understanding Austria would be eliminated immediately after retreats.

Precisely! Which is why I started this forum thread, to highlight situations that shouldn't have been allowed.

[teccles]

Yep - I hadn't realised that some of those situations are bot-specific. I think the root of these could be that the bots vote draw at a time when people can't - between two phases, rather than during one or the other.

That's not all the situations you discuss, but in my opinion the others are fine.

[jmo1121109]

This game where Austria drew in your opening post actually also looks weird - the orders say the retreats happened, but on the board Vienna is still red and Austria has 1 centre. Maybe I'm missing something, but I can't see how that can happen - according to my understanding Austria would be eliminated immediately after retreats.

Precisely! Which is why I started this forum thread, to highlight situations that shouldn't have been allowed.

Bot's have different draw rules. As soon as there is no human in the game, the game is instantly drawn. Without this check, the bots have entered orders instantly for centuries of play crashing the entire site. Bot's also don't know how to draw, so the site handles that logic for them.

"Bot Voting:
The bots in this game do not get a pause or unpause vote, pausing and unpausing only counts human votes.

If a bot is winning a game and has gained supply centers in the last 4 turns, it will stop the game from being drawn or cancelled. Otherwise bot games can be drawn or cancelled anytime."

If bots are ever developed that understand when to vote draw then this type of odd placement logic can go away, but for now it's needed.

[DougJoe]

Here's an example of a game that was drawn, despite one player having 18 centres (yes, seriously): http://webdiplomacy.net/board.php?gameID=349971 That shouldn't have happened.

Hey, that's one of mine, didn't expect to see that! At the time, I didn't even realize that it was reported as a draw instead of a win for ol' Zultar Bot.

(Stupid trivia - that round of 7 games versus the bots were all named after the characters from the Bard's Tale 3 cluebook, if that means anything to anybody.)

[Nescio]

While I understand bots behave differently, I assumed the games were processed the same way as against humans.
Nevertheless, could the draw check be performed after the defeat and solo-victory checks?

[Doom427]

Not only are survived and defeated functionally the same result now that PPSC has been removed, but even when we had the bad get-points-even-though-you-lost system, a 0 unit survival is the same as a defeat. This only matters for drawing when some one is at 0 units and well, that only happens as we said before if people like you enough to keep you around. Maybe a new player gets punked by it once or twice, but there's plenty of rules that can do that to a new player