Print

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

[Jive Junkie]

Say some crazy billionaire Dominion enthusiast was offering $10 million in prizes (split up among the top 16 or so) for a Dominion tournament, but only if it could meet the following conditions:

a) Luck should be minimized, and it should be a very good (but obviously not perfect) indicator of players' relative skills - at the very least, the very best players should reliably place near the top. [i.e. if you were to run the tournament again, you'd see a lot of the same people in the final rounds]

b) You have 2 weekend days, of 8 hours at most each day.

c) It should be able to accommodate a player pool of at least 200 people.



How would you structure the tournament? Some random things to consider:
- How many players per game?
- Should players play a single game, two games (each going first), or a longer match (3, 5, 7)?
- If multiple games, should the Kingdom be changed or remain the same?
- Should it be single-elim, double-elim, swiss with points, etc?
- How will players be matched up?
- How do you deal with ties?
- How should time be managed?
- How will Kingdom cards be selected?
- Should players know the Kingdom cards beforehand for any particular round?
- Should players have the same starting hand?
- Should there be any banned cards, or perhaps a veto system?


Have at it! Note that the inspiration for this thread was my friend telling me that Dominion would never work as a serious tourney game, since it was too luck-based. I know there are Dominion tourneys out there, but they'd probably be structured very differently if a ton of money were on the line.

[rrenaud]

Make it two player.

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over. 

[sitnaltax]

I have a little bit of experience as a chess TD. Here is how I would do it.

* Games are two-player. With big prize money on the line, 3+p will be fraught with opportunities for kingmaking and collusion, and the integrity of the tournament would be difficult to maintain.
* Until the semifinals, matches are best 2 of 3. Loser plays first. The winner of a coin flip of each match chooses: play first in the first game (before seeing the Kingdom cards) or win the tie if the match goes 1-1-1 (or, by some crazy chance, 0-0-3). (I would prefer another game, but this is my concession to limited time.) In the semifinals, matches are best 3 of 5.
* The tournament structure is double elimination. With 256 players, this will produce a winner in 9 rounds. It allows all players at least two matches, and there is no incentive at any point to throw a match.
* If an independent Dominion system is available (Iso, Goko, whatever) that system is used to determine seeding. Otherwise it is random.
* Time is 75 minutes per 3-game match, 125 minutes per 5-game match. This allows a little bit of slack time.
* Players are required to play briskly. At the 45-minute mark, if game 2 is not finished, a TD may observe the game. If one player is playing slowly, and the match times out, the game will be adjudicated in favor of the opponent. Blatant stalling in the judgment of the TD will result in a disqualification. If both players are playing slowly, and the match times out, the third game is marked as a draw, but if the match is 1-1, the winner is decided by coin flip, not the tiebreak mentioned earlier.
* Depending on the tournament, Kingdom cards may be selected either randomly or at the discretion of a celebrity set designer--i.e. DXV or his chosen heir. A randomly chosen Kingdom is mulliganed if there are not at least two cards costing $5, per DXV's statements about what makes a good Kingdom. They will not be revealed before the start of the tournament in any event. The Kingdom cards change each game, but during each round of the tournament, each set of players plays with the same Kingdom cards. This allows commentators to compare different strategies post-mortem.
* If random Kingdoms are used, players may opt to use veto mode. Veto mode is a tournament-long decision, and a match is played in veto mode only if both players have opted for it. In veto mode, 12 Kingdom cards are chosen and each player, in player order, chooses a card to eliminate. (I considered having the choice be made before the players know who is going first; but this prevents us from having the who-goes-first choice made before the Kingdom is revealed.)
* There is no identical-starting-hands rule.

[Powerman]

I think the key is having a fair balance of 2, 3, and 4 player dominion.

How I would structure the tournament:
-First, define many kingdoms, probably in the range of 50.  Announce these kingdoms (on like a sheet of paper) at the beginning of the day.  This way all players can view each of them and start to get an idea of what they could be dealing with.  Each game will be played on a randomly selected one of these, unless otherwise noted.
-In 2 and 3 player games, all players will start with the same starting hands (randomly chosen).
-In 4 player games, each player will start with a random hand that is not necessarily the same.
-All ties will be broken by number of wins within the round.  Then number of total wins.
-Divide the players randomly into pods of 16 (we'll say 15 pods, so 240)
   -Within these pods of 16, divide randomly into groups of 4.
   -Each group of 4 plays 2 4-player games. (Reverse seating order).  Points go 3-2-1-0.
   -Now, re-divide the players so a person who got the most points is in a group with a 2nd, 3rd, and 4th (1 from each of the 4 groups)
   -Now, each group of 4 plays 2 4-player games. (Reverse seating order).  Points go 6-4-2-0.
   -The winner of each of these groups of 4 moves on, so 4 of the original 16 advance, and 60 players should be left.
   -That stage (6 4-player games + movement) should take roughly 3 hours.
-Lunch Break!
-Divide the players randomly into pods of 12 (so 5 pods)
   -Within these pods of 12, divide randomly into groups of 3.
   -Each group of 3 plays 3 3-player games. (Each player starts 1 time in each seating position).  Points go 5-2-0.
   -The player that gets 3rd out of 3 in each group is eliminated.
   -Now divide the remaining 8 players of each pod into 4 matchups of a 1st vs. 2nd (from 2 different groups).
   -The players play 3 2-player games, with the person who won the 3-player group going first in the 1st and 3rd games, and the person who got second going first in the 2nd game.
   -The winner of each of the pairings moves on to the quarterfinal, so 4 of the original 12 advance, and 20 players are remaining.
   -That stage (3 3-player games and 3 2-player games + movement) should take roughly 3 hours as well.
-The 20 remaining players are split into 5 groups of 4 players.
   -Each groups plays 4 4-player games. (Each player plays 1 game from each seating spot) Points go 6-4-2-0.
   -The winner of each group (so 5 players) advances to the semifinals.
   -The 3rd and 4th place of each group is eliminated.
   -The 2nd place of each group advances to a qualifier for the last spot in the semifinals.
   -That stage (4 4-player games) should take roughly 2 hours.
-End of Day 1!

-Start of Day 2!
-The 5 second place players will each do a round robin 2-player contest.
   -This means each player will play 1 game against each player (4 total) and have 1 bye.
   -Each player will go 1st in 2 matches and 2nd in 2 matches.
   -A player gets 3 points for a win, 1 point for a tie (on same turns), and 0 points for a loss.
   -The player with the most points advances to the semifinals.
   -That stage (5 rounds of 2-player games) should take roughly 2 hours as well.
-Lunch Break!
-Now, there are 6 semi-finalists.
   -Divide the players randomly into 2 groups of 3 players.
   -Each group of 3 plays 6 3-player games. (Each player starts 1 time in each seating position).  Points go 5-2-0.
   -The players that gets 2nd and 3rd out of 3 in each group is eliminated.
   -The winner from each group (2 total) advance to the finals.
   -That stage (3 rounds of 3-player games) should take roughly 3 hours.
-Dinner Break!
-We're at the finals!
   -The 2 players will play a best of 7 set.
   -The seating order will alternate, and each kingdom will be used in 2 straight games.
   -The match will end when 1 player has won 4 games.
   -However, if after 6 games the 1st seat player has won all 6 games (series tied 3-3), the series will change to a best of 9.
   -If after 8 games the 1st seat player has won all 8 games (series tied 4-4), a coinflip will be used to determine who gets the first seat in the final game.

The end!

[werothegreat]

200 people - break them into 50 sets of 4, randomly.  Each set of 4 plays 5 4-player games.  Whoever gets 4th place twice is removed.  If there is a relevant tie for 3rd place, do another kingdom.  If someone gets 4th place twice before the five games are up, eliminate them then.  Each game will be a totally random selection of cards, shuffled out and drawn by a judge.  Each table will play with a different kingdom.  Black Market will not be allowed.

You now have 150 people - break them into 50 different sets of 3, randomly, making sure no one is in a group with someone they've already played.  Each set of 3 plays 4 3-player games.  Whoever gets 3rd place twice is removed.  If there is a relevant tie for 2nd place, do another kingdom.  If someone gets 3rd place twice before the four games are up, eliminate them then. 

You are now down to 100 people.  50 sets of 2.  Start doing 2-player games, in a bracket system.  In case of a tie, do another kingdom with the same pair.  After the second round of this, there will be 25 players - randomly select a set of 3, and take the winner of that.  This will lead to 12, then 6, then 3.

50 sets - play with Base
25 sets - play with Intrigue
12 sets - play with Seaside
6 sets - play with Prosperity
3 sets - play with Hinterlands

Have four rounds of a 3-player game.  Whoever gets 3rd place twice is removed.  If there is a relevant tie for 2nd place, do another kingdom.  If someone gets 3rd place twice before the four games are up, eliminate them then.  Two of the rounds will be a mix of Alchemy and Cornucopia, and two will be Dark Ages.

Finally, do best out of three for the final two players, randomly selected (no Black Market).

Colony/Platinum and/or Shelters may be present in any game, at the discretion of the judge selecting the kingdom.

[antony]

I think 16hr is way too short for 200 players, but well.

I have very little experience with non-isotropic play :-) but let's assume 20min is a reasonable amount of time for one game to be completed (I am not considering >=3p games here).  Note that with that many players you also have to include time for players to report results, to find their opponents, etc (anyone who played in a big Swiss in of any game -- in my case, bridge and go -- knows that quite some time is always lost there).  I'll group the games by two so that hopefully, if the first game is too long then the second one will be shorter (and you gain time in reporting, etc.).  This means you can fit 8/(2/3)~12 double games (optimistically) per day.

I would just run a full 2-day, 24-round Swiss, cutting half of the field away after day one (cutting two thirds away may even be better for luck minimization but that would be quite unpopular amongst the players).  The reason for prefering a Swiss is that I think 2 of 3 is just too random for a KO, and if you start having 3 of 5 then 1/ you don't have enough rounds (you're basically limited to 5 rounds) and 2/ you have to allow time for games that go all the way to 5 matches while that could be a rare occurence.  For each Swiss match, a player can score 0, 0.5, 1, 1.5 or 2 points (victories) (so total 24 pts per day); each player is P1 for one of the games.  A common trick in bridge Swisses to keep the second day reasonably interesting if there are players far ahead of the field is to limit the VP spread between the day-1 leader and the last qualifier to some preset limit (6-8 pts sounds reasonable to me); if the spread is larger then divide all scores accordingly to bring it back to that limit.  Another advantage using this technique is that because of the somewhat aribitrary divisor, players have some fractional score on day 2 which serves as a tiebreaker.

Card selection (in particular, which sets to use) is really up to the organizer (I mean, if I want to run a Base Dominion tournament, I can do it.  If I want to run an all-expansions Dominion tournament, I can do it.).  Probably an interesting way to do so is to publish 12 card sets approximately 10 min before each round starts (so while the previous one is running!) and use isotropic-style veto mode.

By the way, another idea I would like to see experimented (but perhaps that is the subject for another thread) is komi-bidding: after the kingdom is revealed (but before veto, if used), both players secretely write down how many points (possibly a negative number, but that should be rare) they are willing to pay for the privilege of playing first.  Moreover the price must be a half-integer, which avoids ties.  The highest bidder plays first and his opponents starts with that many VP tokens (and a half :-)).  If both bids are identical then P1 is determined by a coin flip (and the bid is still paid).

[() | (_) ^/]

Say some crazy billionaire Dominion enthusiast was offering $10 million in prizes (split up among the top 16 or so) for a Dominion tournament, but only if it could meet the following conditions:

...

How would you structure the tournament? ...

Wait, is this really a hypothetical?  Just in case it is not, and you happen to be a billionaire (or your friend is), and you're posting this as part of your planning of such a tourney... then please consider this my sign-up.

Even though I'm way early, I'm proud to be the first to register for your tournament, sir.  Anything I can do to help organize it?

[theory]

Say some crazy billionaire Dominion enthusiast was offering $10 million in prizes (split up among the top 16 or so) for a Dominion tournament, but only if it could meet the following conditions:

...

How would you structure the tournament? ...

Wait, is this really a hypothetical?  Just in case it is not, and you happen to be a billionaire (or your friend is), and you're posting this as part of your planning of such a tourney... then please consider this my sign-up.

Even though I'm way early, I'm proud to be the first to register for your tournament, sir.  Anything I can do to help organize it?

I have some Nevada oceanfront property to sell you.

[() | (_) ^/]

Say some crazy billionaire Dominion enthusiast was offering $10 million in prizes (split up among the top 16 or so) for a Dominion tournament, but only if it could meet the following conditions:

...

How would you structure the tournament? ...

Wait, is this really a hypothetical?  Just in case it is not, and you happen to be a billionaire (or your friend is), and you're posting this as part of your planning of such a tourney... then please consider this my sign-up.

Even though I'm way early, I'm proud to be the first to register for your tournament, sir.  Anything I can do to help organize it?

I have some Nevada oceanfront property to sell you.

Sure thing, I'll just need your bank account information so that I know where to deposit the money.

[DStu]

Sure thing, I'll just need your bank account information so that I know where to deposit the money.

If you need help, I have already experience to forward money for former nigerian kings...

[DG]

- How many players per game?
  Early rounds of 4 player games moving into 3 player then 2 player games. Include a repechage for early losers.
- Should players play a single game, two games (each going first), or a longer match (3, 5, 7)?
  As many as you can accommodate in the time, at least first to 3 games for two players matches. First to 5 for a final match.
- If multiple games, should the Kingdom be changed or remain the same?
  Changed. Random kingdoms.
- How will players be matched up?
  Some seedings in early rounds with results from early rounds determining later rounds. Need a consistent rule for start player.
- How do you deal with ties?
  Either share the points or replay the game, either fair. Fair ranking/scoring for four player games will always be more difficult.
- How should time be managed?
  Fixed time period for kingdom assessment before play, then let referees monitor speed at slow tables
- How will Kingdom cards be selected?
  Random, or least random from a selected expansions (Intrigue + Cornucopia, whatever)
- Should players know the Kingdom cards beforehand for any particular round?
  No
- Should players have the same starting hand?
  No.
- Should there be any banned cards, or perhaps a veto system?
  Definitely no

[blueblimp]

Make it two player.

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

I like the goal of avoiding boring mirror matches, but there are plenty of kingdoms with a key card where player skill matters. For example, if Goons is there, you obviously would ban Goons almost every time, but Goons engines are interesting and require skill. About the same is true of Ambassador.

There are also boards where there are multiple critical cards to a boring strategy, where the players might choose different ones. Bishop/Chapel--ban the Bishop or the Chapel? (Okay, maybe the Chapel on most boards.)

[Axe Knight]

My major tournament idea from last year:

Invite players.  Good ones, preferably.  Then, take the pool of invited players and divide them evenly into groups.  The groups should try to space out the best players.  Each player in the group would play each other player, a certain number of times.  At the end of the day, there'd be a tournament-wide cut.  The cut would be everyone who "won" their group, followed by some "wildcards" for the remaining slots.  Then after this, you'd have a second day, and maybe a third, with the same format, new groups, and a final cut.  Once this final cut was made, it would go to a standard elimination tournament on the last day.  I like the idea of players waiting around to find out if they made the cut rather than going to straight up elimination at first. 

Let's say there's 16 players remaining after the final cut.  You could start with a best of five (first to win three games), between each player to advance to play the next one, culminating in a best of seven final.

[Archetype]

All I know is, if this actually got off the ground, I would definitely want to compete in it.

[chwhite]

It may take too long, but I'd love to see a tournament with pool play: you play some round robin prelim brackets, and then the top 2 players in each bracket (to guard against seeding mistakes where the two best players are in the same bracket) make it to either playoff round robins, or double-elimination, or something like that.  I say playoff round robins because the more games the better, but again I understand if that isn't feasible for large groups.  Starting with pool play would definitely be best I think.

Prelim matches should probably be best-of-3, raised to best-of-5 (or 7) in the finals.  Either 2p or 3p is fine (4p is not), if it's actually possible to set it up so you play both that would be ever better.

As for choosing the kingdoms, my preference would actually be to do it something like duplicate bridge.  The TD gets DXV or someone else who's really good at Dominion and isn't playing to come up with a whole bunch of kingdoms which are interesting, not too luck-based, have multiple paths, aren't going to take an hour to play, run the whole gamut of published cards.  And then the kingdoms are announced just before play, no time to think about them beforehand.  You could have the prelim players rotate tables (which have the kingdoms out there to begin with), for ease of setup.

[popsofctown]

Make it two player.

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

This qualifies as a variant, but is a huge step in the competitive direction

[popsofctown]

I don't like designed kingdoms.  If Donald likes engines , he might design engine boards, but WW has a career of making the best of BM kingdoms.  The tournament should be objective in nature, kingdom selection should be objective.  ( Familiar can still get an objective ban, I'm sure statistics back up its variance.)

Course I just supported a variant last post, so I'm not being consistent at all <_>. The answer to the OP is yes , there is definitely a way to pull it off, even without going variant (banning familiar isn't a variant)

[ftl]

Does it need to be that complicated?

Two-player Dominion with random kingdoms is pretty skill-based as it is. If you make it, say, best 4 out of 7, you can just do straight random kingdoms and trust that skill will prevail over luck of the draw in a series like that. I mean, that's the easy way to deal with luck - just increase sample size. Then you can use whatever competitive structures are already in use for other 2-player games. Use a smaller sample size for lower-level tournaments or rounds, a larger sample size for semis and finals, or something.

[popsofctown]

I'm not sure Swiss can guarantee a trusty winner in 16 hours without kingdom restriction .  I'm being totally lazy and not doing the math though..

[ednever]

I was thinking about this a few months ago.

I dont think it needs to be as ambitious as its made out to be. Ignoring format for a moment, imagine something like this:

The World Series of Dominion

A series of satellite tournaments that can do whatever they want. These tournaments would have to raise enough money to get their winner to the Championship and (maybe) pay there way. There could be online tournaments doing the same thing.

The Championship would have a high entry fee ($500?) and likely be based in Vegas (cheap flights from most places in the US, cheap accommodation, lots of relatively inexpensive venues)
100 players gets you a pool of $50,000. Assume $10,000 to cover costs and marketing. Winner gets $20,000, with another $20,000 divided among the runners up.

The organizer would be taking a little risk the first year, and even if successful wouldn't make much if any money. But, if you could build the brand and grow it from year to year you could hope for an explosion. Poker was a niche until it wasn't.

Problem is I'm not sure how entertaining Dominion is as a spectator event - which is what you really need to make big purses.

Ed

[DWetzel]

A somewhat "simple" solution for early rounds, of a sort:

Duplicate Dominion.

Have a set kingdom, and play it at all tables.  This is probably best with 2 players, but can be adapted somewhat for 3p and 4p setups.

Compare scores not against your opponent at the table, but against the people that are playing at other tables in your position.  Rotate opponents after each round (and new kingdoms, of course).

(It's not as elegant a setup as it is in duplicate bridge due to the vagaries of shuffle luck in Dominion, but over a decent sample of games this will somewhat even out, and everyone understands there's a decent luck element in Dominion anyway.)

[ftl]

Comparing scores is a bad idea. Dominion scores do not correspond well to how well you did in a game. You can win 1-0 by a landslide, or you could lose 100-0 in a close game that could have gone either way.

[^_^_^_^]

You also are all forgetting one thing- RULES. Tourny Rules. Like for instance, what happens if someone makes a hand jester or something that appears to be trying to give advice? What if a parent walks up to the board their kid is playing at and yells YOWZA! or NO!!!!!!!!!! when he makes a bad decision? What should kids be doing when they need help with a situation (rule clarification, etc.). What if I have to go to the restroom in the middle of my game?

[popsofctown]

How many kids play Dominion?
Are you a youth yourself, ^_^_^_^?

[DWetzel]

Aside from the fact that the "100-0 and could have gone either way" bit is a rather edge case, let me elaborate a bit.

It's not as though we're adding up the raw scores for each board and going from there.  Each setup (let's say we play 10 kingdoms for this) would make up 10% of each player's total score.  Let's say there are 22 total players, and thus 11 matchups.  You'd get 1 matchpoint for each person playing your direction that you beat, and 1/2 point for each that you tied, for a maximum on each kingdom of 10 matchpoints.  If you lose 1000-0, well, guess what, the worst that you can do is get 0/10 on that kingdom.

The bad part about this is when you get into KC/Goons setups where someone has the incentive to not end the game even though they're up 500-20.  I'm not sure of the best way to resolve that, but it's not any more luck based than whether you're sitting 1st or 4th when the Sea Hag with no trashers comes up.

[ftl]

You also are all forgetting one thing- RULES. Tourny Rules. Like for instance, what happens if someone makes a hand jester or something that appears to be trying to give advice? What if a parent walks up to the board their kid is playing at and yells YOWZA! or NO!!!!!!!!!! when he makes a bad decision? What should kids be doing when they need help with a situation (rule clarification, etc.). What if I have to go to the restroom in the middle of my game?

That's not Dominion-specific. Do the same thing that people do for any other 2-player competitive game.

[Axe Knight]

You also are all forgetting one thing- RULES. Tourny Rules. Like for instance, what happens if someone makes a hand jester or something that appears to be trying to give advice? What if a parent walks up to the board their kid is playing at and yells YOWZA! or NO!!!!!!!!!! when he makes a bad decision? What should kids be doing when they need help with a situation (rule clarification, etc.). What if I have to go to the restroom in the middle of my game?

Most of this is decorum, and would be of course decided before events began.  Somehow, millions of tournaments of all sorts of things go on without having to worry about this stuff much. 

[blueblimp]

Aside from the fact that the "100-0 and could have gone either way" bit is a rather edge case, let me elaborate a bit.

It's not as though we're adding up the raw scores for each board and going from there.  Each setup (let's say we play 10 kingdoms for this) would make up 10% of each player's total score.  Let's say there are 22 total players, and thus 11 matchups.  You'd get 1 matchpoint for each person playing your direction that you beat, and 1/2 point for each that you tied, for a maximum on each kingdom of 10 matchpoints.  If you lose 1000-0, well, guess what, the worst that you can do is get 0/10 on that kingdom.

The bad part about this is when you get into KC/Goons setups where someone has the incentive to not end the game even though they're up 500-20.  I'm not sure of the best way to resolve that, but it's not any more luck based than whether you're sitting 1st or 4th when the Sea Hag with no trashers comes up.

This encourages collusion between opponents to rack up large scores. Doesn't even need to be blatant either: curser on the board? Why don't we ignore that, it'll be better for both of us.

Edit: Also, 100-0 is not an edge case at all. KC-Bridge available with enablers? Neither player will buy any VP until they hit KC-KC-Bridge-Bridge-Bridge, and you will get a final score of 57-3 or thereabouts, even if the other player could have won next turn.

[ftl]

In an engine game where the game ends on piles, your score is fundamentally not related to how far ahead of your opponent you are. With +buy, you end the game on piles as soon as you are able to do so with a win, and buy, say, 1 point more than your opponent when you do so.

Don't count scores for anything. That way lie stupid rulings which are designed for smithy-money games and break down whenever the games get clever. The only meaningful bit of information that you get out of a game is who won; you don't get any more fine-grained information than that.

[Davio]

200 people in 2 days is impossible without luck.

It's also pretty simple to understand that a game of Dominion between two equally skilled players will be decided by luck 100% of the time, otherwise they wouldn't be of equal skill, get it?

This is true because luck plays an inherent factor in Dominion. There is shuffling, hence luck.
It's not true in chess where I'm pretty sure there is no luck. Well, you could be lucky to be white but you can just play 2 games to solve that.

If there is a 1% skill difference between 2 players, the part that luck plays is still close to 100%.

So you really can't throw luck out of the equation no matter how hard you try. This means you have to accept that it's a factor and by the end of two days you're not really crowning the luckiest overall, but you are crowning the luckiest of the best players.

[blueblimp]

It's also pretty simple to understand that a game of Dominion between two equally skilled players will be decided by luck 100% of the time, otherwise they wouldn't be of equal skill, get it?

This isn't accounting for playing well or badly. Otherwise, in games with no luck, two equally-skilled players would see the same game outcome every time, which obviously isn't true.

[popsofctown]

200 people in 2 days is impossible without luck.

It's also pretty simple to understand that a game of Dominion between two equally skilled players will be decided by luck 100% of the time, otherwise they wouldn't be of equal skill, get it?

This is true because luck plays an inherent factor in Dominion. There is shuffling, hence luck.
It's not true in chess where I'm pretty sure there is no luck. Well, you could be lucky to be white but you can just play 2 games to solve that.

If there is a 1% skill difference between 2 players, the part that luck plays is still close to 100%.

So you really can't throw luck out of the equation no matter how hard you try. This means you have to accept that it's a factor and by the end of two days you're not really crowning the luckiest overall, but you are crowning the luckiest of the best players.

Chess has luck.

You can argue this from two angles

1) How else do you explain best of 7 sets between players of different strengths being anything but 4 straight wins?

2) The future possibilities of a game are too many to be analyzed and calculated.  Thus, the chess player skips some analyses, and at some point, just guesses that the knight takes bishop retaken by bishop pawn push pawn push pawn push rook e4 rook f6 queen g6 is good for him.  Neither player actually looked at that chain of possibilities, but one player chooses to take the first step.  He bases that first step on other reasons, he probably analyzed some other equally long chains, and liked them, but neither player looked at this one.

The line of reasoning makes sense to me and I buy it.  At some point there are decisionmaking criteria that are, in actuality, arbitrary to successful victory in the game.  If they are arbitrary to victory they function randomly.



Think of it this way.  What if in a contest, I gave you a square grid of 100 division problems.  The divisors are all primes between 17 and 31 and you only have to produce the remainder, and your pencil writes quickly.  The numerators are all ten digit numbers.  Prize is five hundred dollars, you have to beat ftl.  I give you 40% as much time as it takes an intelligent person to solve all the division problems correctly.

Both you and ftl will solve 40% of the problems, probably the first 40%, because why not.  Then in a negligible amount of time you will guess the other 60% because each remainder you right gives you roughly a 4% chance of another right answer.  This game clearly has luck, you didn't really work all the problems.  The problems you two had time to work will be a brute contest of mathematical skill, and the best man will probably win, but one player may luck out.

Now if I instead rename half the problems "shuffle luck", and hide the numerators with black highlighter, leaving functionally random numerators, have I changed how much luck this game has?  No, not really, you're still going to guess on 60% of the problems.

Outcomes that can't be foreseen exactly are indistinguishable from outcomes that we don't have the time to analyze and then foretell is what I'm getting at.  It makes sense to me.  It's hard though.

This is why chess requires some sample size.  Not much but some.

[polonkus]

Aside from the fact that the "100-0 and could have gone either way" bit is a rather edge case, let me elaborate a bit.

It's not as though we're adding up the raw scores for each board and going from there.  Each setup (let's say we play 10 kingdoms for this) would make up 10% of each player's total score.  Let's say there are 22 total players, and thus 11 matchups.  You'd get 1 matchpoint for each person playing your direction that you beat, and 1/2 point for each that you tied, for a maximum on each kingdom of 10 matchpoints.  If you lose 1000-0, well, guess what, the worst that you can do is get 0/10 on that kingdom.

The bad part about this is when you get into KC/Goons setups where someone has the incentive to not end the game even though they're up 500-20.  I'm not sure of the best way to resolve that, but it's not any more luck based than whether you're sitting 1st or 4th when the Sea Hag with no trashers comes up.

This encourages collusion between opponents to rack up large scores. Doesn't even need to be blatant either: curser on the board? Why don't we ignore that, it'll be better for both of us.

Edit: Also, 100-0 is not an edge case at all. KC-Bridge available with enablers? Neither player will buy any VP until they hit KC-KC-Bridge-Bridge-Bridge, and you will get a final score of 57-3 or thereabouts, even if the other player could have won next turn.

I believe that this type of collusion actually did happen at a past tournament which took into account relative scores.
Anyone remember the details?

[DWetzel]

Aside from the fact that the "100-0 and could have gone either way" bit is a rather edge case, let me elaborate a bit.

It's not as though we're adding up the raw scores for each board and going from there.  Each setup (let's say we play 10 kingdoms for this) would make up 10% of each player's total score.  Let's say there are 22 total players, and thus 11 matchups.  You'd get 1 matchpoint for each person playing your direction that you beat, and 1/2 point for each that you tied, for a maximum on each kingdom of 10 matchpoints.  If you lose 1000-0, well, guess what, the worst that you can do is get 0/10 on that kingdom.

The bad part about this is when you get into KC/Goons setups where someone has the incentive to not end the game even though they're up 500-20.  I'm not sure of the best way to resolve that, but it's not any more luck based than whether you're sitting 1st or 4th when the Sea Hag with no trashers comes up.

This encourages collusion between opponents to rack up large scores. Doesn't even need to be blatant either: curser on the board? Why don't we ignore that, it'll be better for both of us.

Edit: Also, 100-0 is not an edge case at all. KC-Bridge available with enablers? Neither player will buy any VP until they hit KC-KC-Bridge-Bridge-Bridge, and you will get a final score of 57-3 or thereabouts, even if the other player could have won next turn.

1. Depends on how the scores are done, if it's point DIFFERENCE (which was my possibly poorly illustrated intent), no such collusion possibility exists.  What's good for P1 is bad for P2.

2. While your point is possible, in such a case, the person who realized that their opponent is about to go off and decides to buy a province and lose 54-9 instead will be rewarded, relative to people with equal luck. 

And there's just no way around the shuffle and seating luck, regardless of the format, so even discussing that is pretty pointless.  The only way to minimize that is to get more games in.

[Kirian]

So, I'm tempted to simply say "No."  But hey, this is all just speculation, right?

a) Luck should be minimized, and it should be a very good (but obviously not perfect) indicator of players' relative skills - at the very least, the very best players should reliably place near the top. [i.e. if you were to run the tournament again, you'd see a lot of the same people in the final rounds]

This automatically means a Swiss system with multiple games per matchup.

b) You have 2 weekend days, of 8 hours at most each day.
c) It should be able to accommodate a player pool of at least 200 people.

Again, the only way to do both of these and have any sort of reliable outcome is Swiss.  I still don't think you can get a truly reliable outcome from 200 players in 16 hours though.  20, maybe, but I think you really need 24.

- How many players per game?

Two.  Is this really even a question any longer?

- Should players play a single game, two games (each going first), or a longer match (3, 5, 7)?

Longer.  I'd advocate some variant of the tennis system I discussed in a different thread.  Players start first in order A-B-B-A-A-B, with tie games being replayed.  First to three wins and winning by two takes the match; a 3-3 finish is a tie.  Each match is therefore around 2 hours.

- If multiple games, should the Kingdom be changed or remain the same?

Changed each time, but all tables should play the same set of six kingdoms each round.  Kingdoms would be randomized by the TO and distributed as games are played.

- Should it be single-elim, double-elim, swiss with points, etc?
- How will players be matched up?
- How do you deal with ties?

Swiss.  Start with random matchups for round one. Use FIDE tiebreakers where needed.

- How should time be managed?

Time controls are difficult; only time spent making real decisions should be counted against a competitor.  Ideally, the matches would be over in two hours; at ninety minutes, judges would be asked to determine if players are deliberately playing slowly.

- How will Kingdom cards be selected?
- Should players know the Kingdom cards beforehand for any particular round?
- Should players have the same starting hand?
- Should there be any banned cards, or perhaps a veto system?

Randomly, no, yes, no.  A year ago I would have answered yes to the last question, but I've changed my stance there.  The singular exception might be Possession due to its likely effect on game length, not game play.

[ftl]

2. While your point is possible, in such a case, the person who realized that their opponent is about to go off and decides to buy a province and lose 54-9 instead will be rewarded, relative to people with equal luck. 

That rewards suboptimal playing-for-second-place decks. A Big Money deck will always get a province or two before the engine takes off and buys the rest. Doesn't mean the BM deck was closer to winning than the engine deck which had zero points because it was building up to the mega-turn that the opponent got to first.

In a lot of games, the relevant measure is "winning by X turns" - not "winning by X points". If I'm one turn away from having my deck fire and buy everything when my opponent's deck fires and buys everything, that's a really close, really well-played game - we differed by only one turn, but that one-turn difference could be an arbitrarily number of points, anywhere from one point difference to MEGA-GOONS-TURN, where that point difference means little to nothing. Whereas in a deck where a BM+X strategy competes against a mega-turn engine, the BM+X will get a few provinces, making the score look close even if the engine was the optimal strategy by far.

Here's another example - Ambassador games! Whoever wins the ambassador war probably gives their opponent all 10 curses by the end of the game. Doesn't matter whether it was close or not close, whoever wins will give out all 10.

Oh, here's a BM+X example - duchy-dancing! I could buy the last province and lose by a point. OR, I could buy a duchy, with one province remaining, since that gives me a chance to win if on the next turn, my opponent buys an estate and I have a chance to grab the last province. The latter SHOULD be clearly the right play, since it gives you a chance to pull out a win. And yet, buying the province is likely to lead to "lose by 1", while playing to win is likely to have you lose by 10 (if the opponent buys the last province on the next turn). Rewards people who end the game with a loss as long as it's a loss by not very many points, rather than rewarding people that play for the chance to win.


...just don't do it. Points in Dominion are NOT  a valid indicator of quality of play; while they may seem to be such at first glance, it's not actually true.

[blueblimp]

As far as reducing the luck in a match of Dominion goes, there's always the doubling cube option.

[carstimon]

Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

I hope you don't mind me going back a page to disagree with this.
I agree that mirror matches are boring, but I think this implementation has problems.  For instance, chapel.  Chapel is just sooo game changing, even if a board has multiple options.  This will practically mean that chapel isn't used.  Yeah, I mean there's board when it's an interesting decision whether or not to chapel, but I think this will just kill many interesting games.
Same goes for other power cards: KC, hunting party, torturer where chainable.

[DWetzel]

2. While your point is possible, in such a case, the person who realized that their opponent is about to go off and decides to buy a province and lose 54-9 instead will be rewarded, relative to people with equal luck. 

That rewards suboptimal playing-for-second-place decks. A Big Money deck will always get a province or two before the engine takes off and buys the rest. Doesn't mean the BM deck was closer to winning than the engine deck which had zero points because it was building up to the mega-turn that the opponent got to first.

In a lot of games, the relevant measure is "winning by X turns" - not "winning by X points". If I'm one turn away from having my deck fire and buy everything when my opponent's deck fires and buys everything, that's a really close, really well-played game - we differed by only one turn, but that one-turn difference could be an arbitrarily number of points, anywhere from one point difference to MEGA-GOONS-TURN, where that point difference means little to nothing. Whereas in a deck where a BM+X strategy competes against a mega-turn engine, the BM+X will get a few provinces, making the score look close even if the engine was the optimal strategy by far.

Here's another example - Ambassador games! Whoever wins the ambassador war probably gives their opponent all 10 curses by the end of the game. Doesn't matter whether it was close or not close, whoever wins will give out all 10.

Oh, here's a BM+X example - duchy-dancing! I could buy the last province and lose by a point. OR, I could buy a duchy, with one province remaining, since that gives me a chance to win if on the next turn, my opponent buys an estate and I have a chance to grab the last province. The latter SHOULD be clearly the right play, since it gives you a chance to pull out a win. And yet, buying the province is likely to lead to "lose by 1", while playing to win is likely to have you lose by 10 (if the opponent buys the last province on the next turn). Rewards people who end the game with a loss as long as it's a loss by not very many points, rather than rewarding people that play for the chance to win.


...just don't do it. Points in Dominion are NOT  a valid indicator of quality of play; while they may seem to be such at first glance, it's not actually true.

Neither, really, are win/loss in a lot of cases either (see the "who happens to draw the KC/Bridge combo first in an otherwise mirrored match"), but we use that.  At least with this method you're more fundamentally comparing apples to apples.  I don't deny that it creates a somewhat different "winning" heuristic, I just deny that that's an inherently bad thing.

[blueblimp]

Make it two player.

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

One other thing about this... assuming the two players don't write down the same card, is the rule against buying/gaining it actually enforced during the game? Because that wouldn't be much like Dominion anymore. But if you didn't enforce the ban, then the choice would be a bit meaningless.

[Jive Junkie]

Thanks for the great ideas, everyone!

Wait, is this really a hypothetical?  Just in case it is not, and you happen to be a billionaire (or your friend is), and you're posting this as part of your planning of such a tourney... then please consider this my sign-up.

Even though I'm way early, I'm proud to be the first to register for your tournament, sir.  Anything I can do to help organize it?

No comment. 

Make it two player.

This is something I'd be conflicted about. Multiplayer definitely occupies some of the space of "Dominion", but it definitely has increased variance and will take longer to complete games. I think I would be for acknowledging its existence and the fact that there is skill in adapting your play for more players. Perhaps have multiplayer take up 25% of the swiss rounds (2 matches out of 8, for example)?

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

That sounds like a really fun variant, but I wonder if it may stray too far from "pure" Dominion for most people.

As a random aside, how often does the mirror match come into play with the highest-ranked players (say 40+)? Among those, how often is it a fairly exact mirror match, and how often do they vary by 1-2 support cards?

Randomly, no, yes, no.  A year ago I would have answered yes to the last question, but I've changed my stance there.  The singular exception might be Possession due to its likely effect on game length, not game play.

Hmm, along those lines, would it make sense to try to limit the number of Colony games as well? How about other time-heavy cards? (I don't know what they would be off the top of my head)

Edit: Also, 100-0 is not an edge case at all. KC-Bridge available with enablers? Neither player will buy any VP until they hit KC-KC-Bridge-Bridge-Bridge, and you will get a final score of 57-3 or thereabouts, even if the other player could have won next turn.

So points as a tie-breaker seems like it could give inaccurate results. How about opponent record as a tie-breaker? I know some tournaments do that when making the swiss cutoff for the next round / playoffs.

200 people in 2 days is impossible without luck.

It's also pretty simple to understand that a game of Dominion between two equally skilled players will be decided by luck 100% of the time, otherwise they wouldn't be of equal skill, get it?

This is true because luck plays an inherent factor in Dominion. There is shuffling, hence luck.
It's not true in chess where I'm pretty sure there is no luck. Well, you could be lucky to be white but you can just play 2 games to solve that.

If there is a 1% skill difference between 2 players, the part that luck plays is still close to 100%.

So you really can't throw luck out of the equation no matter how hard you try. This means you have to accept that it's a factor and by the end of two days you're not really crowning the luckiest overall, but you are crowning the luckiest of the best players.

I think that would be an acceptable outcome. Poker, for instance, never really crowns the best player, but a combination of the best play + best luck. You still see a lot of the same faces at the final table, so striving for that should be good enough for Dominion.

And someone else made the point that sometimes the better player overall may not be the better player on any given match. Maybe that combination of cards fails to inspire him as much as it does the lesser player. Maybe he overlooks some weird corner case card interaction that proves to be the difference. Maybe he's distracted because his girlfriend broke up with him due to jealousy over his true love (Dominion).

Do you guys think a 20 hour tourney length would help? I could see 2 10-hour days being feasible.

[Kirian]

So points as a tie-breaker seems like it could give inaccurate results. How about opponent record as a tie-breaker? I know some tournaments do that when making the swiss cutoff for the next round / playoffs.

In a Swiss system, you can go through a bunch of different tiebreakers if you need to.  http://en.wikipedia.org/wiki/Tie-breaking_in_Swiss-system_tournaments

FIDE uses a specific set of tie-breaker rules which includes rules for dropouts and byes; I used them in the IsoDom tournament.  There's no need to ever, ever use game points as a tiebreaker in Dominion.  It's a bad, bad, bad, bad idea.

Do you guys think a 20 hour tourney length would help? I could see 2 10-hour days being feasible.

20 hours might do it.  It's enough for eight solid rounds, which segregates up to 256 via either elimination or Swiss.

[^_^_^_^]

You also are all forgetting one thing- RULES. Tourny Rules. Like for instance, what happens if someone makes a hand jester or something that appears to be trying to give advice? What if a parent walks up to the board their kid is playing at and yells YOWZA! or NO!!!!!!!!!! when he makes a bad decision? What should kids be doing when they need help with a situation (rule clarification, etc.). What if I have to go to the restroom in the middle of my game?

Most of this is decorum, and would be of course decided before events began.  Somehow, millions of tournaments of all sorts of things go on without having to worry about this stuff much. 

Well yeah, lots of stuff is decided before events begin but as Dominion gets bigger more people play it. And, there are vods on youtube that also show very young kids playing Dominion.

@pops yes I am a 'youth' as you call it.

[Mole5000]

2. While your point is possible, in such a case, the person who realized that their opponent is about to go off and decides to buy a province and lose 54-9 instead will be rewarded, relative to people with equal luck. 

That rewards suboptimal playing-for-second-place decks. A Big Money deck will always get a province or two before the engine takes off and buys the rest. Doesn't mean the BM deck was closer to winning than the engine deck which had zero points because it was building up to the mega-turn that the opponent got to first.

In a lot of games, the relevant measure is "winning by X turns" - not "winning by X points". If I'm one turn away from having my deck fire and buy everything when my opponent's deck fires and buys everything, that's a really close, really well-played game - we differed by only one turn, but that one-turn difference could be an arbitrarily number of points, anywhere from one point difference to MEGA-GOONS-TURN, where that point difference means little to nothing. Whereas in a deck where a BM+X strategy competes against a mega-turn engine, the BM+X will get a few provinces, making the score look close even if the engine was the optimal strategy by far.

Here's another example - Ambassador games! Whoever wins the ambassador war probably gives their opponent all 10 curses by the end of the game. Doesn't matter whether it was close or not close, whoever wins will give out all 10.

Oh, here's a BM+X example - duchy-dancing! I could buy the last province and lose by a point. OR, I could buy a duchy, with one province remaining, since that gives me a chance to win if on the next turn, my opponent buys an estate and I have a chance to grab the last province. The latter SHOULD be clearly the right play, since it gives you a chance to pull out a win. And yet, buying the province is likely to lead to "lose by 1", while playing to win is likely to have you lose by 10 (if the opponent buys the last province on the next turn). Rewards people who end the game with a loss as long as it's a loss by not very many points, rather than rewarding people that play for the chance to win.


...just don't do it. Points in Dominion are NOT  a valid indicator of quality of play; while they may seem to be such at first glance, it's not actually true.

Neither, really, are win/loss in a lot of cases either (see the "who happens to draw the KC/Bridge combo first in an otherwise mirrored match"), but we use that.  At least with this method you're more fundamentally comparing apples to apples.  I don't deny that it creates a somewhat different "winning" heuristic, I just deny that that's an inherently bad thing.

So the correct solution is to reject the idea that you can determine the 'quality' of a win and just accept that you only have win/lose/draw to work off rather than coming up with tournament rules that would totally distort play and in no way reflect the reality of the situation.

[Wolphmaniac]

Now I'm curious; does anyone know the highest stakes ever for a Magic tournament?  I have no idea what goes on in that scene but I'm curious to know now.

I also figure that the upper limit for prize money in a Dominion tournament is probably equal to 1/10th of whatever the biggest Magic tournament was.

[werothegreat]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

[DStu]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

You could just not have independent boards, but distribute the 170 cards (at which number are we at the moment) over all tables. That would give you random boards on each table, which just happen to not be independent over tables. But I don't see why this should matter. So you would only need about 1/2 (large) expansion per table, so ~50 boxes in total. +some Base cards. +maybe Ruins/Shelters, there probably aren't enough.  Don't want to calculate how many you will need.

Small problems with BlackMarket here, maybe just don't use it, or take an extra set per BM (so you have 10 extra of each card, which you can distirbute among the BMs.

[werothegreat]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

You could just not have independent boards, but distribute the 170 cards (at which number are we at the moment) over all tables. That would give you random boards on each table, which just happen to not be independent over tables. But I don't see why this should matter. So you would only need about 1/2 (large) expansion per table, so ~50 boxes in total. +some Base cards. +maybe Ruins/Shelters, there probably aren't enough.  Don't want to calculate how many you will need.

Small problems with BlackMarket here, maybe just don't use it, or take an extra set per BM (so you have 10 extra of each card, which you can distirbute among the BMs.

But then you're not being purely random.  Also, there are currently 187 kingdom cards (192 if you count Promos).

[Wolphmaniac]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

You could just not have independent boards, but distribute the 170 cards (at which number are we at the moment) over all tables. That would give you random boards on each table, which just happen to not be independent over tables. But I don't see why this should matter. So you would only need about 1/2 (large) expansion per table, so ~50 boxes in total. +some Base cards. +maybe Ruins/Shelters, there probably aren't enough.  Don't want to calculate how many you will need.

Small problems with BlackMarket here, maybe just don't use it, or take an extra set per BM (so you have 10 extra of each card, which you can distirbute among the BMs.

But then you're not being purely random.  Also, there are currently 187 kingdom cards (192 if you count Promos).

This discussion is relevant for a realistic scenario but it is moot given the parameters of the scenario proposed by the OP.  The tournament is being financed by a billionaire and has a $10 million purse.  The billionaire can easily afford to procure all of the necessary sets.  Not to mention procuring the physical space to host the tournament.

[DStu]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

You could just not have independent boards, but distribute the 170 cards (at which number are we at the moment) over all tables. That would give you random boards on each table, which just happen to not be independent over tables. But I don't see why this should matter. So you would only need about 1/2 (large) expansion per table, so ~50 boxes in total. +some Base cards. +maybe Ruins/Shelters, there probably aren't enough.  Don't want to calculate how many you will need.

Small problems with BlackMarket here, maybe just don't use it, or take an extra set per BM (so you have 10 extra of each card, which you can distirbute among the BMs.

But then you're not being purely random.  Also, there are currently 187 kingdom cards (192 if you count Promos).

Every single table gets a completely random set of cards. Each set has with probability 1/(n chose 10).  Sets between the tables are not independent, but that does not matter, because there is no interaction across tables.

Edit: Look at this like this. You have say 2 tables and only base game. You have 25 cards, perfectly shuffled. So now you give the first 10 cards to the first table, the second 10 cards to the second table. Without knowing the 10 first cards, the second 10 are a perfectly fine random set. The first 10 card are also without knowing the second 10 cards. So if the second table wouldn't exist, the first one would not even doubt that their cards are uniform random. And if the first table does not exists, the second one would not doubt.
So of course they now exists, and because of this the cards on both tables combined are not as they would be if you give both tables their cards indepently of each other. But there are no interactions across tables, so you don't mind if the 20cards have the proper distribution. All you need to know is if the first 10cards have, and the second 10 cards have, without caring for the correlations between these sets. And as we have shown above, each 10 for their own are perfectly fine.

Unless there is some interaction over tables that I don't see.

[Mole5000]

Unless there is some interaction over tables that I don't see.

What is the chance that both tables play a game with Chapel in it?

Whether or not that is an actual problem I don't know.

[DStu]

Unless there is some interaction over tables that I don't see.

What is the chance that both tables play a game with Chapel in it?

Zero, but..

Whether or not that is an actual problem I don't know.

... I don't see why it should.

[Jive Junkie]

Sidenote - you do realize that in order for 200 people to compete in such a tournament, you'll need at least 100 new copies of the Base and each expansion, right?

You could just not have independent boards, but distribute the 170 cards (at which number are we at the moment) over all tables. That would give you random boards on each table, which just happen to not be independent over tables. But I don't see why this should matter. So you would only need about 1/2 (large) expansion per table, so ~50 boxes in total. +some Base cards. +maybe Ruins/Shelters, there probably aren't enough.  Don't want to calculate how many you will need.

Small problems with BlackMarket here, maybe just don't use it, or take an extra set per BM (so you have 10 extra of each card, which you can distirbute among the BMs.

But then you're not being purely random.  Also, there are currently 187 kingdom cards (192 if you count Promos).

This discussion is relevant for a realistic scenario but it is moot given the parameters of the scenario proposed by the OP.  The tournament is being financed by a billionaire and has a $10 million purse.  The billionaire can easily afford to procure all of the necessary sets.  Not to mention procuring the physical space to host the tournament.

This is indeed what I had in mind - I wanted the discussion to be focused on the tournament itself, not on the funding to make it happen.

That being said, one solution to the problem of needing at least 100 of each set: require each competitor to bring all the sets! Since this is a big money tourney with presumably top players, they should already have all or most of the sets in their possession, or at least access enough to borrow them. And even if they don't, they're playing for millions - you can think of owning the complete game as part of the of entrance fee.

[DWetzel]

2. While your point is possible, in such a case, the person who realized that their opponent is about to go off and decides to buy a province and lose 54-9 instead will be rewarded, relative to people with equal luck. 

That rewards suboptimal playing-for-second-place decks. A Big Money deck will always get a province or two before the engine takes off and buys the rest. Doesn't mean the BM deck was closer to winning than the engine deck which had zero points because it was building up to the mega-turn that the opponent got to first.

In a lot of games, the relevant measure is "winning by X turns" - not "winning by X points". If I'm one turn away from having my deck fire and buy everything when my opponent's deck fires and buys everything, that's a really close, really well-played game - we differed by only one turn, but that one-turn difference could be an arbitrarily number of points, anywhere from one point difference to MEGA-GOONS-TURN, where that point difference means little to nothing. Whereas in a deck where a BM+X strategy competes against a mega-turn engine, the BM+X will get a few provinces, making the score look close even if the engine was the optimal strategy by far.

Here's another example - Ambassador games! Whoever wins the ambassador war probably gives their opponent all 10 curses by the end of the game. Doesn't matter whether it was close or not close, whoever wins will give out all 10.

Oh, here's a BM+X example - duchy-dancing! I could buy the last province and lose by a point. OR, I could buy a duchy, with one province remaining, since that gives me a chance to win if on the next turn, my opponent buys an estate and I have a chance to grab the last province. The latter SHOULD be clearly the right play, since it gives you a chance to pull out a win. And yet, buying the province is likely to lead to "lose by 1", while playing to win is likely to have you lose by 10 (if the opponent buys the last province on the next turn). Rewards people who end the game with a loss as long as it's a loss by not very many points, rather than rewarding people that play for the chance to win.


...just don't do it. Points in Dominion are NOT  a valid indicator of quality of play; while they may seem to be such at first glance, it's not actually true.

Neither, really, are win/loss in a lot of cases either (see the "who happens to draw the KC/Bridge combo first in an otherwise mirrored match"), but we use that.  At least with this method you're more fundamentally comparing apples to apples.  I don't deny that it creates a somewhat different "winning" heuristic, I just deny that that's an inherently bad thing.

So the correct solution is to reject the idea that you can determine the 'quality' of a win and just accept that you only have win/lose/draw to work off rather than coming up with tournament rules that would totally distort play and in no way reflect the reality of the situation.

No, not really.  Win/lose/draw compares you directly to your OPPONENT, who faced a rather different situation than you (seating, if nothing else, which as we all know matters a fair amount).  Comparing to other seat2's (for instance) on the same kingdom who faced the same starting situation you did is a more valid comparison.  If P1 is winning 60% on a particular kingdom, is it the least bit fair to P2 to get, effectively, 20% of a "loss" merely by sitting in the wrong seat?  Remember, one of the stated goals here is to minimize the luck comparison without stretching the game out over a huge sample space.

A possible compromise: weight wins according to what the "field" does (I assume everyone's on board with "everyone plays the same kingdom sets" as a reasonable rule).  If P1 is winning 60% on a particular kingdom, reward the wins for P2 on that kingdom more heavily than you would the wins for P1.  Details left to the reader.

[antony]

Duplicate Dominion sounds like a cool idea but again I think that that would (in practice) be a huge mess to organize.  Still, if we do that then I would score it as follows.
Still organize a full Swiss (with some deep cut after day 1).  On each round, each player plays two games against his opponent, the two games are independent and scored separately (more data points :-)).  If I win a game then I score x points, and my opponent scores 2-x points, where x is (1 - the percentage of players in the same starting position as myself, not including myself, who won their game).  For example, if I won as P2 and nobody else did, I score 2 and my opponent 0.  If I won as P1 and everybody else did, then I score 1 and my opponent too (as there was no "merit" in winning this game).  If I won as P1 and 50% of the other P1's did, then I score 1.5 and my opponent 0.5.  You get the idea.

[gamesou]

Make it two player.

Contentious suggestion:  Goal: Try to eliminate mirror games where players execute the same obvious strategy.  These games are either decided either by the luck of the draw or tactics, but I think the tactical decisions aren't usually that hard or interesting, and so are not good player skill differentators.  Implementation:  Use rejection sampling on kingdom sets.  Both players see the randomly drawn kingdoms, get 5 (2, 3?) minutes to analyze it, and then privately write down a kingdom card that the opponent can't buy/gain.  If they write down the same card, start over.

I think that this is an awesome idea that could be pushed even further. Something like this: both players rank all cards from the supply (including treasures and victory cards) from "best" to "worst". We compare the lists and search for the highest disagreement ; your opponent cannot gain that card from your list on his turns. I just tried the idea on random kingdoms and that's quite exciting, especially where to rank Province on an engine board is a tough question !