webdiplomacy.net

If everyone 7-way draws we can make this an FvA tournament. Would save me having to run another Champions' League!

Peregrine Falcon wrote: ↑

Mon Apr 12, 2021 4:48 pm

I resent your assertion that I live pretty much on the great lakes

Considering Toronto traffic, we're probably the same driving time from getting a fresh sip of the effluent.

goldfinger0303 wrote: ↑

Sun Apr 11, 2021 3:42 pm

You know, I don't have a tiebreaker in place for the Group Stages because its just so unlikely it would ever happen....but this was close indeed. And if Sons of Vlad and Great Lakes tie in the Finals, that does make the tournament's only tiebreaker invalid.

I would honestly have to consult with the moderator team on constructing a second tiebreaker that seems fair.

Well the first thing that you can do is eliminate the scoring of the 1v1 games, ha ha. What heresy led to their inclusion in the sanctified space that is the webDiplomacy World Cup, anyway? (By the way, while we're on the topic of changing the tournament format, next year let's split standard press games into a standard press game and a rulebook press game with hidden draw votes!)

But in case ignoring the 1v1 games doesn't suit your fancy, I think that I have a solution for this. DjEcc24 designed the tournament to first respect WTA scoring, then simulate PPSC scoring to break ties:

The points are scored as follows.

420 for a win
420/n for a draw
0 for defeat.

Plus the SC count (any win has an SC count of 18)

With half-points for the Gunboat games

WTA scoring is fully reflected. However, the rules only include a simplified version of PPSC. I would change the PPSC portion to be something like this:

• Plus the SC count (any win has an SC count of 18)
• In the event of a tie, the above scoring will only apply to games that end in a win. In games that end in a draw, games will be scored as 34/n. For example, a game that ended in a draw with a 17-SC France, a 15-SC Turkey, a 1-SC Germany and a 1-SC Russia is usually scored under tournament rules as 122 points for France, 120 points for Turkey and 106 points each for Germany and Russia. Under tie-breaker scoring, all 4-way draws in the tournament are scored as 113.5 points (or 56.75 points in gunboat games).
• If results are still tied, games that ended in a win that had an SC count greater than 18 will receive special attention. In these games, one or more of the 34 'PPSC' points is not awarded since the winner can only be awarded exactly 18 points. To break a tie, a fraction of the extra point(s) will be awarded to the defeated survivors as a function of their SC count at the end of the game. For example, France wins a game while occupying 19 SC's, while Turkey survives with 10 SC's and Germany survives with 5 SC's. Under regular tournament scoring, France would win 438 point, Turkey would win 10 points and Germany would win 5 points. To break a tie, Turkey's team would win an extra ⅔ of a point while Germany's team would win an extra ⅓ of a point.

If we have half decent stats on how each country performs, we can use a 'strength of schedule' to break ties.

I like the idea of differentiating the full press games. I'll take that into consideration for the next iteration.

However, I feel that 1v1 as a format has risen to a position in the online diplomacy community to merit inclusion, and that the overall weighting given to it (120 points for a perfect run) is of appropriate size to not drastically influence scoring. Given reasonable levels of competition, the most it can do, on net, is to elevate your score by 40 or so points - or about the equivalent of changing a 4-way draw to a 3-way draw.

I think a good tiebreak would come down to the true WTA scoring JECE suggested, or a strength of schedule Yonni suggested. Its something that will ultimately only be discussed if we need to though :) (But will probably be fleshed out before the next run)

goldfinger0303 wrote: ↑

Tue Apr 13, 2021 2:21 am

true WTA scoring

true PPSC scoring ;-)

I'm not sure what JECE suggested is a true PPSC or a true WTA... I would rather call it a true DSS

I like the idea of making one of the FP games RB HDV! Would change up the dynamic and requires somewhat different skills to work with.

Like a penalty shootout in football, a tiebreak should distill the game down to its essence. How about if there's a tie, all the other teams vote for the winner?

Teccles, that is both devious and genius!

teccles wrote: ↑

Tue Apr 13, 2021 5:57 pm

Like a penalty shootout in football, a tiebreak should distill the game down to its essence. How about if there's a tie, all the other teams vote for the winner?

Well, the penalty shootout is about as far from the essence as you can get, but that's another story.

How about an anonymous tie break FP game with all seven teams, where the better result between the two wins?

That could end in another tie, no?

I endorse Teccles' idea, but I request election speeches by the tied teams members on this forum

RoganJosh wrote: ↑

Wed Apr 14, 2021 7:43 pm

That could end in another tie, no?

Theoretically. For the vote idea, if one team abstains, a 2-2 draw is possible.

It's pretty hard to have an exactly equal result if you have tie breaks:

Solo obviously wins.
If both are part of a draw, most center wins
If one makes draw, that one wins.
If both are eliminated, the latest elimination wins
If eliminated on the same turn, then turn weighted number of centers (-1 for Russia as long as it holds all four home centers) is the tie break.
Failing that? Maybe a 1v1 for all the marbles.

Or maybe just have 1 v 1 for all the marbles to start with.

Scratch both teams and give the previous winners a bye to the finals? Just spit-balling here.

A lot of tournaments do that the person that had the largest point draw gets the advancement.

Yeah, imo the best tie-breaking list would look something like...
1) # of solos
2) Points from draws not including SC bonus
3) Total SCs
4) # of 1v1s won
5) Sudden death 1v1 challenge

For 1), if one team solos and the other does not, the team that solos deserves to go on. Yes, this means solos are heavily weighted both in terms of points and the tie-breaker, but I'd say this a feature, not a bug.
2) Basically looks at which team was is the smallest draws in the classic map games.
3) If still tied (ie. two teams had the exact same distribution of draws), then look at SC differences. (Since 1v1s could hypothetically mask those differences.)
4) If still tied (somehow), then the team that won more 1v1s proceeds. (Since I think we can all agree that the classic map games are more important.)
5) If the two teams still somehow also won the same # of 1v1s, then it just goes to a sudden death resolution via 1v1. Since 1v1s are relatively short, this shouldn't take too long.

Pretty rare that any of these would ever see use, though.

(In quick response to other ideas proposed: teccles' idea is neat, but would be completely decided by meta considerations, which is not something I think we want. Ogion's would just take way too long. Yonni's would be neat, but getting everyone to agree on a power distribution would not be. JECE's is similar, but perhaps overly convoluted. VI's has merit.)

Peregrine Falcon: Have you checked the math on that? 1) through 3), if I understand you correctly, are basically the current scoring rules stated differently.

I don't agree with 4) since, as you say, the classic map games are more important.

I think that a somewhat convoluted but more precise tiebreaker system is better than a tiebreaker system that perhaps oversimplifies the regular scoring system.

I like the sudden death idea as a final tiebreaker. But it may take longer than you think if you need multiple rounds of sudden death!

What do you mean by this, by the way?

Peregrine Falcon wrote: ↑

Thu Apr 15, 2021 4:12 pm

(Since 1v1s could hypothetically mask those differences.)

VillageIdiot's idea has merit, except that teams rarely stay intact for the next edition of the World Cup. Also, it wouldn't help us break a tie in the finals.

By the way. Since I have your attention, Peregrine Falcon, what do you think of adding the press variant of rulebook public press?

JECE wrote: ↑

Thu Apr 15, 2021 6:24 pm

Peregrine Falcon: Have you checked the math on that? 1) through 3), if I understand you correctly, are basically the current scoring rules stated differently.

If you're asking if I have run models to estimate what distribution of points we'd actually be likely to see in a tie, I have not. If you're asking whether I glanced at how my tie-break would work for Sons of Vlad v Great Lakes, I did.

And yes, that is exactly what I was suggesting. For regular position determinations, all the points, no matter what they're from, are aggregated into a single number. This tie-break suggestion would in effect disaggregate those points and rank them based on what is determined to be most important.

JECE wrote: ↑

Thu Apr 15, 2021 6:24 pm

What do you mean by this, by the way?

Peregrine Falcon wrote: ↑

Thu Apr 15, 2021 4:12 pm

(Since 1v1s could hypothetically mask those differences.)

Lets look at the Sons of Vlad v Great Lakes example:

1) No solos. Moving on...
2) Draw sizes:

• Sons of Vlad drew FP2 (3WD = 140pts), PP (7WD = 60pts), GB1 (3WD = 70pts) for a total of 270 points from their draw sizes
• Great Lakes drew PP (3WD = 140), GB1 (5WD = 42), GB2 (3WD = 70), for a total of 252pts from DSS
• Thus, we have a winner. Sons of Vlad made it into the higher-point draws, and so comes out on top. But for sake of argument...

3) SCs

• Sons of Vlad got SC points from: FP2 (17 SCs/pts), PP (4 SCs/pts), GB1 (9 SCs = 4.5 pts), for a total of 25.5 pts from SCs
• Great Lakes got SC points from: PP (9 SCs/pts), GB1 (5 SCs = 2.5 pts), GB2 (4 SCs = 2 pts), for a total of 13.5 pts from SCs
• So Sons of Vlad *also* were larger powers at the draw, and again comes out on top. But still...

4) 1v1s

• Sons of Vlad won 3, for 60 pts
• Great Lakes won 4 and drew 1, for 90 pts
• Now we see where Great Lakes got their points from. They won and tied an additional 1v1. But this comes after the Classic map stuff, so it doesn't play into the tie breaker.
• Note: If it ever comes down to 1v1 acting as the drawbreaker, then they actually would have the exact same amount of points in 1v1, since we already examined the rest of the points, and they were still tied. That's why it's specifically 1v1s won, and not 1v1 points that act as the draw breaker (1 win = 2 draws according to the points). If one team won one 1v1 and the other drew two, the team with the win would progress. However, there is an argument to be made that making it into two draws is actually more impressive in FvA than winning once, so an alternative could be to look at # of games not defeated in (thus counting wins and draws the same).

Perhaps that clears it up? I think this shows that it's a decently robust tie-breaking system, even if it's basically never going to be needed.

JECE wrote: ↑

Thu Apr 15, 2021 6:24 pm

I like the sudden death idea as a final tiebreaker. But it may take longer than you think if you need multiple rounds of sudden death!

If the 1v1 draws (which is possible), it's an unlikely enough occurrence that I think it's fine to wait a bit longer to play another. But given how much Diplomacy players like luck-based game mechanics, perhaps the final tie-break should instead be to flip a coin.

JECE wrote: ↑

Thu Apr 15, 2021 6:24 pm

By the way. Since I have your attention, Peregrine Falcon, what do you think of adding the press variant of rulebook public press?

Peregrine Falcon wrote: ↑

Tue Apr 13, 2021 4:50 pm

I like the idea of making one of the FP games RB HDV! Would change up the dynamic and requires somewhat different skills to work with.

As a final point, a tie-break list is probably most needed to break a tie to choose the 7th team to get into the finals. I would actually go so far as to suggest that there *shouldn't* be tie-breaking in the finals round. If two teams manage to tie for first, it's probably the consequence of some sort of diplomatic outcome on the board—and that should be reflected in the final scores.

Peregrine Falcon wrote: ↑

Thu Apr 15, 2021 4:12 pm

Yeah, imo the best tie-breaking list would look something like...
1) # of solos
2) Points from draws not including SC bonus
3) Total SCs
4) # of 1v1s won
5) Sudden death 1v1 challenge

For 1), if one team solos and the other does not, the team that solos deserves to go on. Yes, this means solos are heavily weighted both in terms of points and the tie-breaker, but I'd say this a feature, not a bug.
2) Basically looks at which team was is the smallest draws in the classic map games.
3) If still tied (ie. two teams had the exact same distribution of draws), then look at SC differences. (Since 1v1s could hypothetically mask those differences.)
4) If still tied (somehow), then the team that won more 1v1s proceeds. (Since I think we can all agree that the classic map games are more important.)
5) If the two teams still somehow also won the same # of 1v1s, then it just goes to a sudden death resolution via 1v1. Since 1v1s are relatively short, this shouldn't take too long.

Pretty rare that any of these would ever see use, though.

(In quick response to other ideas proposed: teccles' idea is neat, but would be completely decided by meta considerations, which is not something I think we want. Ogion's would just take way too long. Yonni's would be neat, but getting everyone to agree on a power distribution would not be. JECE's is similar, but perhaps overly convoluted. VI's has merit.)

To be clear, I was proposing a playoff if the *final group* is tied for the final winner. We can create all kinds of tie breaks to make it easier, but if they're tied after those tie breaks are worked through, then what?

As for a system of tie breaks, I like this set as well as any, I suppose.