Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[rrwoods]

Recently there was a post about Bridge (the game, not the card) which got me looking into the game briefly. Apparently it's common to run what is called Duplicate Bridge at tournaments. For those unfamiliar with the game, one hand of Bridge consists of each team member taking certain roles (based on their position relative to the dealer), and the four players each get thirteen cards (so the whole 52-card no-joker deck is dealt). In Duplicate Bridge, several rounds are played where several sets of opponents play the exact same thirteen cards, and the scoring systems are (in theory) designed to balance the effects of "poor" hands vs "strong" hands and measure each teams skill with those cards. Apologies to the Bridge players if I've butchered anything, but what I e described above is a sufficient intro to what I propose here.

I don't have a system fully fleshed out here; indeed, the reason for this post is to gather more ideas on feasibility and implementation.

The outcome of any given game of Dominion (wait for it) depends on the board (!) and the starting hands. Some "solutions" have been proposed to the starting hands "problem" -- identical starting hands, force 4/3, agree on 4/3 or 5/2 and force what is agreed on, and many others. What I would like to do for a physical-cardboard tournament is to keep the individual games as the rulebook describes, and find a match scoring system that balances the effect of the starting hands and boards by somehow comparing the results of other games played under identical conditions in the same tournament.

The match structure would involve arranging the boards and drawing all the starting hands for all the round 1 games before any players are seated, and recording the starting hands (leave the boards on the table between games). Then for round 2, reset the starting hands to the same as they were in round 1, and shuffle the players around in some predetermined (semi?)random fashion. Repeat until a sufficient number of rounds are played; for very small tournaments, this may be when all players have played all others.

I'm thinking the scoring system would either (1) reward more for wins from "hard" situations (ones with a low final win percentage), or (2) penalize losses from "easy" situations (ones with high win percentage), or both. This incentivizes players to optimize their performance in each individual game, which is what I want -- no sort of VP comparison would be acceptable. Exactly how to form those metrics (what the factors are for rewarding/penalizing), or alternative "balancing" metrics, or even whether acceptable metrics exist, is the question posed by this post.

[florrat]

There are some major differences between Bridge and Dominion, and I think these differences make this tournament setup particularly suited for Bridge, and less suited for Dominion.

(1) In Bridge there are unequal opportunities between your team and your opponents team (in a single hand). Your team can have bad hands, and your opponent good hands. The tournament setup you describe is designed to eliminate this randomness as much as possible. In Dominion on the other hand, the supply is the same for each player, so the players have mostly the same opportunities. Sure, one player can open 5/2 while the other opens 3/4, but usually that is not that big of a deal (I think first player advantage is bigger than opening split advantage). So I think that when adopting this tournament setup you're trying to solve a problem which does not really exist in Dominion. Unless you want to count all the shuffling during the game as the opportunity for a player, but it's nearly impossible to give two players (who might have different strategies) equal shuffle luck throughout the game.

(2) In Bridge, the only randomness of the game is when dealing the starting hands. In contrast, in Dominion you shuffle multiple times throughout the game, so you can still have a lot of (bad) luck after your first 2 hands. In Bridge, your team has equal opportunities as a team with identical starting hands, while in Dominion your Sea Hag can still be discarded by your opponents Sea Hag on turn 3, which may not have happened to another player with exactly the same strategy.

(3) In Bridge it is easier to compare the outcome of different games with the same hand: you can look at the scores. In Dominion, the victory points (or the difference in VP) isn't a good indication of how the game went. So you shouldn't do any VP comparison (which is as you suggest). But then any scoring metric can only take into account the number of games won and the number of games lost on a specific board, which (if you play something like 6 rounds) isn't that much information to determine how "easy" or "hard" a specific board/starting hand is. So you would have very little information to determine how lucky someone got because of his starting hands on specific boards.

[rrwoods]

Your (3) is my biggest concern -- that the sample size is too small to closely represent how easy or hard a seat is. (1) and (2) I am not as concerned with (though shuffle luck has enough to do with the outcome that (2) may be the same concern as (3)).

I do disagree with the problem not being present in dominion. Your opening can have a significant impact on the outcome; iso even had an identical starting hands option to mitigate this! First player advantage is also something I'm trying to account for, and matching the starting cards is a simple enough process that its (possibly small) benefit I believe outweighs its small cost. The only issue is that you need players not to talk about their openings to other players, which may raise the cost to a level that is unacceptable.

[Polk5440]

I also think a duplicate-like tournament would be great.

However, I don't think coming up with "balancing" metrics would be worth it. Simply playing the same kingdoms, rotating around the room, and tallying up # of wins (or points if multiplayer) should be balancing enough -- if not, then duplicate is not worth it.

Wait -- now that I think about it, did the tournament in Michigan recently have this structure? Kingdoms were pre-set and you just walked around to your next table? Did everyone end up playing the same kingdoms by the end?

A benefit with duplicate is that Kingdoms could be pre-randomized according to particular criteria or designed! (I am definitely in favor of more tournaments using designed kingdoms. Or a tournament that randomizes from only two expansions at a time. Or whatever!)

[liopoil]

Wait -- now that I think about it, did the tournament in Michigan recently have this structure? Kingdoms were pre-set and you just walked around to your next table? Did everyone end up playing the same kingdoms by the end?

kingdoms were preset, but players did not play quite every single set. In the preliminaries, there were 6 kingdoms and 4 rounds, so each player played 4/6.

[Elestan]

Simply playing the same kingdoms, rotating around the room, and tallying up # of wins (or points if multiplayer) should be balancing enough.

The problem with duplicate Kingdoms is the number of sets required; to do six tables of the same Kingdom, you need six copies of each of the sets required.  Most people don't have that.  Rio Grande sponsored the Michigan tournament, providing 1-2 copies of each set, plus a bunch of base cards, and it was still a challenge to figure out enough fun Kingdoms that didn't overlap cards, so that I could play them at the same time.  In the end I came up with nine, so I could handle up to 27 players.

Wait -- now that I think about it, did the tournament in Michigan recently have this structure? Kingdoms were pre-set and you just walked around to your next table? Did everyone end up playing the same kingdoms by the end?

The number of sets depended on the number of players; we had 19, which made for five 3P and one 4P (2P games are not allowed in official tournaments).  We played four prelim rounds, because that's all there was time for; I'd have loved to have done six, but that would mean something like 9 hours straight of Dominion, which would probably strain even the most dedicated player's stamina.

A benefit with duplicate is that Kingdoms could be pre-randomized according to particular criteria or designed! (I am definitely in favor of more tournaments using designed kingdoms. Or a tournament that randomizes from only two expansions at a time. Or whatever!)

I also recommend designed Kingdoms.  Sadly, while random Kingdoms can be entertaining in that they're each technically unique, too often they fall into one of a few standard patterns (Slog, BM, etc), have some cards that you'd never buy, or that otherwise generate a humdrum game.  A well-designed Kingdom, by contrast, has some strategic depth to it, with real choices to be made, and every card should have a part in at least one plausible strategy.  Also, a good tournament Kingdom should make it difficult to take excessively long turns.  Not impossible (I think that trying to get a massive engine is part of the fun), but the players should have to work for it.  Most of the Michigan tourney sets either were a little light on +Card, or were designed so that an engine player would need to wrap the game up reasonably quickly to avoid being 3-piled.  Possession was the only card I refused to even consider using.

[Davio]

I think the only way to do a Dominion tournament is to do some kind of Swiss thing.
You should never, ever, ever use VP as a tie-breaker of any sort.

The thing is, Dominion just has so much randomness that you need quite a lot of Swiss matches to say anything meaningful about the outcome, I'd think at least 7.

[Elestan]

You should never, ever, ever use VP as a tie-breaker of any sort.

Agreed, and that was one of Rio Grande's rules.

The thing is, Dominion just has so much randomness that you need quite a lot of Swiss matches to say anything meaningful about the outcome, I'd think at least 7.

The Michigan tournament had four preliminary Swiss rounds, then the top nine players advanced to the semifinals and played three more, then the top three players (all of whom would receive prizes) played a final game to determine their placement.  The final was a single game for two reasons:  First, that's about as many rounds as anyone can play back-to-back and still have fun.  Second, a single game provides a climax to the event; if the winner were to be decided based on cumulative records, we could get a situation where we already know the winner before the final round, so there's wouldn't be much thrill in it.

[AdamH]

First, that's about as many rounds as anyone can play back-to-back and still have fun.

Challenge accepted.

[Elestan]

First, that's about as many rounds as anyone can play back-to-back and still have fun.

Challenge accepted.

Heh...okay, there may be exceptions.  But then you'll have to either get up an hour earlier, or drive back an hour later.

In order to have more rounds of play, I would probably have to make it a two-day (probably Sat/Sun) tournament, with the semis/final on the second day.  Which I'd be fine with doing, if it seemed like enough people wanted it; the hotel offers steep discounts to convention attendees who stay overnight.

[Kirian]

I do think a one-game final is... problematic.  While everyone likes a big showdown climax, we all know that--unlike chess--someone can get unutterably bad luck in one game.  (I don't like the one-game showdown in the Catan tournaments for the same reason.)  That said, I don't think it would be necessary to use cumulative score, especially since, for 27 people, three rounds of three-player Swiss is, ignoring ties, enough to determine a single top player--only one player can win three games in a row.

If you were to run a 2-day tournament, Elestan, I'd suggest:

Day 1:  Three-round Swiss qualifier.  Top 9 advance to the semi-finals.
Day 1:  Three-round semifinal.  Top 3 advance to the finals.
Day 2:  Three-round finals.  All three players play all three games, once from each position.

Obviously that doesn't work for just one day!  Maybe in January we can talk tournament theorycrafting while playing a game or two.  Maybe even Dominion!

[rrwoods]

First, that's about as many rounds as anyone can play back-to-back and still have fun.

Challenge accepted.

At MAGfest last year I spent 40 straight hours awake, and I think probably 50% of that time was spent playing Dominion :-P

[Kirian]

At MAGfest last year I spent 40 straight hours awake...

...

You are insane.

[Elestan]

for 27 people, three rounds of three-player Swiss is, ignoring ties, enough to determine a single top player--only one player can win three games in a row.

I'm not actually using a 'proper' Swiss ranking algorithm (where you match people after each round based on their scores), because I haven't found a good way to do so that also satisfies the following criteria:

• Supports up to 27 players, using at most 9 Kingdoms.
• 3-Player games whenever possible (No 2-player Kingdoms, and no more than two 4-player Kingdoms).
• Nobody plays a second 4-player game unless everyone has played a 4-player game.
• Nobody ever replays a Kingdom.
• Players play as many different opponents as possible (given the other criteria).
• Doesn't require a lot of time between rounds.

If you can figure out how to do a true Swiss matching while satisfying the above, I'll seriously consider using it.  Lacking that, I'm using shuffle charts, which move the players between tables in keeping with the above criteria. 

Maybe in January we can talk tournament theorycrafting while playing a game or two.  Maybe even Dominion!

Sounds good to me!

[Watno]

So where's the similarity of your system to Swiss?

[Elestan]

So where's the similarity of your system to Swiss?

Players still play all the rounds in the prelims, are given points based on their position in each game, and proceed to the semifinals (or not) based on their cumulative score.  This makes it much closer to Swiss than to a single or double elimination format.