Rules Discussions

Previous topic - Next topic

TheDetour

Definitely think that the 7th game should be the end of the match.  People do have real lives and there's nothing constructive about playing 10+ games of Dominion just to break a tie.

That being said, if I enter, I certainly will not complain if I am victim of this.

jsh

Quote from: TheDetour on 12 August 2017, 05:51:24 AM
Definitely think that the 7th game should be the end of the match.  People do have real lives and there's nothing constructive about playing 10+ games of Dominion just to break a tie.

That being said, if I enter, I certainly will not complain if I am victim of this.

Agreed 100%, speaking as someone likely to play lots if rounds and thus potentially tiebreak...

markus

I fully agree with drsteelhammer's reasoning. (I was about to post the same).

Regarding the fewer turns over the whole match: a usual 2 player game ends on the turn of the winning player - and I wouldn't resign on my own turn in that case. So after 6 games with equal number of starts and wins, the number of turns taken will be the same. And if game 7 is a draw, they also took the same number of turns. So the criterion will usually not break a tie.

AdamH

Quote from: drsteelhammer on 12 August 2017, 05:40:31 AM
Funnily, last year when we held a similar tournament I actually included your proposal Adam for the Quarterfinals and beyond. I chose to cut because I think it's actually less fair.

I don't have any data for this, but I think it's known to most people who played on a high level that 1st player wins more often than the second player, and even if the rules were changed that all current ties would be awarded to the second player, first player would still be favourable. (Wins 1p>Wins 2p +Ties)

I think that would still be true if the both players adapted to this rule change. Now, playing an 8th game instead of awarding the second player a win takes even more of their chances away to win the match. So this rule tries to give the coin flip who starts the game the least amount of weight, given a limited timeframe. My personal favourite would be to add two games every time until one set doesn't end 1-1, but I don't think we can expect anyone to play that much.

Personally, I'm not too invested about this, I just would like the disadvantaged person to have the smallest disadvantage possible. And I think playing an 8th game slightly increases that disadvantage over the current rule.

To DVX: Regarding your other tiebreakers, I don't quite follow why those would be better tiebreakers. The only thing I can think of is to avoid confusion about the Dominion rules? We have roughly a 52-46-2 % situation on our hands (maybe even worse for the second player). Why no just give this small possibility of a tie to the player who's being disadvantaged in that very game instead of picking a metric that didn't have anything to do with the match?

This logic is a slippery slope, though. Dominion is an inherently unfair game because there is variance, along with the first player advantage. Why do you want to try and correct some of it by using data that you don't have (and probably doesn't exist) but you don't want to enforce identical starting hands?

It's not the place of the tournament organizer to attempt to balance out unfairness that they see in the game by changing the game rules, it's the place of the tournament organizer to get in the way of the game rules as little as possible in determining a winner. The whole "coin flip" argument you make doesn't hold because games of Dominion aren't coin flips, they are opportunities for a skilled player to beat a less skilled player, which is kind of the whole point of the tournament. Playing tiebreaker games is the best way to provide "fairness" in this capacity.

Even DXV agrees that the best tiebreaker in terms of "fairness" (which I think means having the more skilled player win more matches, in the context of this conversation) is my suggestion. It seems that if people are so concerned with their time and want to put a hard cap on the matches at seven games, the best solution is to have the tournament rules reflect what keeps the most in-game integrity intact and allow people to shorten matches (DXV has a list of suggested tiebreakers they could agree on!) if they want to.

Let's assume 256 entries to the tournament, so there will be 256 matches played. Let's be really conservative and assume that half of them will end 3-3 after six games. So we have 128 matches that go to a 7th game. If 2% of those games end in ties, we'll have about 2.5 extra games of Dominion played, throughout the course of the entire tournament, meaning that the average player will play 0.02 extra games of Dominion using my tiebreaker rule as opposed to something that caps the match at 7 games. The impact of this method is much lower than it originally appears, but you end up with a much better structure as a result.

Donald X.

Quote from: drsteelhammer on 12 August 2017, 05:40:31 AM
To DVX: Regarding your other tiebreakers, I don't quite follow why those would be better tiebreakers. The only thing I can think of is to avoid confusion about the Dominion rules? We have roughly a 52-46-2 % situation on our hands (maybe even worse for the second player). Why no just give this small possibility of a tie to the player who's being disadvantaged in that very game instead of picking a metric that didn't have anything to do with the match?
Aside from the board (which yes does matter), the advantage of going first is that maybe you will get an extra turn. But we know the game is a tie, and that was already used as a tiebreaker; no-one got an extra turn.

This is an old thing, from people arguing about tiebreakers for single games of Dominion. The example I used back then was, getting to go first is like finding a lottery ticket on the ground. When we tie, your lottery ticket didn't pay off. Some people still like to punish the person for having found that ticket. I do not.

Donald X.

Incidentally you could split off all this discussion into a "Rules Discussion" thread if you want to keep the Rules thread itself pristine.

Donald X.

In constructed Magic: The Gathering, going first is an advantage (at some points in history, a gigantic one). Their giant tournaments that pay good money do not use who went first as a ranking tiebreaker and never have. What they use is:

1. Match points
2. Opponents' match-win percentage
3. Game-win percentage
4. Opponents' game-win percentage

#1 is just how well you're doing, don't be fooled. #2 is a nice tiebreaker, did you play against people who are doing well. There's more data there in Swiss than in single-elim but it's still something. After that they devalue draws in case that helps.

One thing to consider is, what will feel satisfying in the finals. Here it is, the finals. It goes to a 7th game and it's a tie. People are obviously going to want to see an 8th game played. But no, we need to wrap things up, 7 games is plenty, someone argues somehow. What tiebreaker could you possibly use there that wouldn't feel awful and stupid? I put it to you that the regular Dominion tiebreaker of fewest turns is the only thing even remotely acceptable. Anything else and it's, why didn't you have a real championship. There is more attention paid to the finals than to the semi-finals, more to the semi-finals than to round one. Still if it's obviously bad in the finals, why is it acceptable earlier? Those earlier games also decide who wins the championship. The finals match-up could have easily been the semi-finals instead due to how the pairings fell out. If there's a dark horse, a new online player that seeding doesn't place appropriately by skill level, the finals match-up could have happened instead in round one.

I do think it would be reasonable to let the players decide what to do after the 7th game. If both want to play the 8th game, they play it. If neither does, they use the tiebreakers (turns, match record, ranking; you could throw in opponents' match record before ranking). If just one does, that player wins the "they wanted it more" tiebreaker. I like that tiebreaker because it's right in line with the kind of person who will do well in the tournament anyway: the kind that plays a ton of Dominion. If you don't have time for an 8th game, I don't see how you had time to play enough to be good enough to make the finals.

jsh

A lot of games have shorter matches early in the tournament (often because there is a time crunch to finish in a weekend or something). Given that, it seems acceptable to me that the finals could have different rules. I agree you'd probably want to "play it out" there.

This is all up to the TO, of course.

heron

I'm with drsteelhammer on this. It's already a disadvantage to be the second player in the 7th match, why don't we try to give them a slight leg up? And possibly save some time while we're at it? The rules of dominion aren't sacred or anything; if everyone agrees to play a different way that's fine. Of course in this situation not everyone agrees, so that's a bit of an issue, and while we can't really tell how everyone feels because most people aren't commenting, right now it seems we are reasonably divided on the issue.

The only convincing argument against the current rule that I could think of is if it were common for there to be kingdoms where it was much easier to tie than win. I don't think that is common at all though.

I can see where Adam is coming from though. His argument of "we should follow the rules of the game" makes plenty of sense, but I think Adam and I just prioritized our values differently.

Regarding two players agreeing on a different tiebreaker: I don't really have a problem with it, but maybe someone would try to complain if they do this and lose because of it, so make sure to document the agreement.

Although I usually agree with things Donald says, his tiebreaker suggestions make no sense to me. I could be misunderstanding something or being dumb though.

1. fewest turns over the match
Suppose that both players fight to the end in all of their matches, and every end the game if it would cause them to lose (which is an even better idea than usual with this tiebreaker). Then the only way for this tiebreaker to come into effect is if 2 ties occur when it player A started the game, and 0 ties when player B started; or 3 and 1 (to understand this, consider: If there are no ties, the number of games player A wins when player B starts is equal to the number of games that player B wins when player A starts). In which case, the player who tied a lot when they went second will win the tiebreak. So... this is a really convoluted variant of the current rule, which usually won't even come into effect, and randomly punishes people who don't understand the rule and resign during their turn.

2. most wins over the tournament, not counting game 7
This doesn't really work since each match ends at 3.5. People who tied once and got a score of 3.5 instead of 4 are randomly punished.
I guess you can go with least losses instead, which works sort of well. Benefits the better seeded player.

3. highest rated player wins
I mean ok I guess. This would be kind of disheartening to the lower rated player though, and punishes good players who don't play online a lot (like Adam, it seems?)
I mean they are already hurt by the seeding so maybe that's ok but idk.

So uh those are my thoughts. Option 1 seems bad unless I made a math mistake, which is possible; options 2 and 3 help the higher seeded player which I don't really like.

drsteelhammer

To Adam: I didn't mean the game is supposed to be a coin flip, I meant the starting player is a coin flip for the 7th game. The rule that is set in place now tries to give the 2nd player the biggest space to outplay their opponent. Also, it doesn't really change the rules of the game. We could call the score 3,5-3,5 and the game wouldn't be changed at all, just the scoring would.

A comparison would be European football: In elimination rounds, there are 2 games, one at Stadium A and one at Stadium B. If they both score the same amount of goals, the first tiebreaker is to award the team that scored more goals in the other's stadium (I think that's a bad rule, it's just to show how other games change the scoring without changing the game).

I do agree we wont get the variance out of dominion, but starting player is a lever we have to pull into one direction or another, so we can try and do our best here.

I also agree with you this discussion will probaly never matter for the tournament, but that doesn't really make the decision easier (only easier to ignore)

to DVX: Analogies are a difficult to not be misleading, so you'll have to endure me poking yours a little. Firstly, this is  zero-sum game, not one person against an unkown entity that sells lottery tickets. So there are two people who look at the same money in their pockets, but you give one of them a dagger and tell him to get the other person's money out of theirs. And this current rule basically tries to give the second player a few seconds to run away.

Also, you know a lot about the game, so you will know that there is more to starting than having a 50% chance of having one more turn. You also have good chances to curse them a shuffle earlier; attack them earlier (Militia-Silver mirror openings have a great chance of making the 2nd player very sad). There are also contested piles, it's a lot harder to end up with 2 Saunas when you're starting the game.

So my made up data was the most conservative estimate I dared to make, it looks a lot more grim honestly.

Donald X.

Quote from: heron on 12 August 2017, 05:41:35 PM
1. fewest turns over the match
Suppose that both players fight to the end in all of their matches, and every end the game if it would cause them to lose (which is an even better idea than usual with this tiebreaker). Then the only way for this tiebreaker to come into effect is if 2 ties occur when it player A started the game, and 0 ties when player B started; or 3 and 1 (to understand this, consider: If there are no ties, the number of games player A wins when player B starts is equal to the number of games that player B wins when player A starts). In which case, the player who tied a lot when they went second will win the tiebreak. So... this is a really convoluted variant of the current rule, which usually won't even come into effect, and randomly punishes people who don't understand the rule and resign during their turn.

2. most wins over the tournament, not counting game 7
This doesn't really work since each match ends at 3.5. People who tied once and got a score of 3.5 instead of 4 are randomly punished.
I guess you can go with least losses instead, which works sort of well. Benefits the better seeded player.
Having not really read the rules, and I still haven't, I was under the impression that you always played the full 6 games. If you stop once someone wins 3.5 or 4, and why wouldn't you, then yes, it's not much of a proposal.

Donald X.

Quote from: drsteelhammer on 12 August 2017, 06:01:57 PM
to DVX: Analogies are a difficult to not be misleading, so you'll have to endure me poking yours a little. Firstly, this is  zero-sum game, not one person against an unkown entity that sells lottery tickets. So there are two people who look at the same money in their pockets, but you give one of them a dagger and tell him to get the other person's money out of theirs. And this current rule basically tries to give the second player a few seconds to run away.
I still like mine better!

Going first means maybe you will get an extra turn. If you didn't then you didn't. The extra turn tends to make you win rather than tie (especially when tying means you lose, so you will avoid ending it in what will not actually be a tie). No tiebreaker rule comes into effect when there's no tie.

Where the tiebreaker rule comes in is where the players had equal turns. In that situation I don't need to reward the player who went second.

Quote from: drsteelhammer on 12 August 2017, 06:01:57 PM
Also, you know a lot about the game, so you will know that there is more to starting than having a 50% chance of having one more turn. You also have good chances to curse them a shuffle earlier; attack them earlier (Militia-Silver mirror openings have a great chance of making the 2nd player very sad). There are also contested piles, it's a lot harder to end up with 2 Saunas when you're starting the game.
In fact I said "Aside from the board (which yes does matter)." By "aside from the board" I meant "obviously the exact cards in the match can vary the starting player advantage, and by "(which yes does matter)" I meant "no really, I actually understand this." The board may reward the first player via an attack they draw on turn 3 or whatever. It also may not! It may not do that at all.

This particular topic, as applied to Dominion games rather than matches, has come up before! I have endlessly considered it and argued about it. So I am pretty sure: I am that guy who does not want a tiebreaker based on turn order. I hate it. I successfully got it out of the game itself back when. If there were official tournament rules they would not possibly include a turn-order tiebreaker, and your tournament would be unsanctioned if it had it.

And I mean, that's fine; I don't mind people running unsanctioned tournaments with rules I hate. I am just here in this thread because I like to solve problems, and this problem was presented, of how to have the fairest possible tiebreaker when you can't actually play another game.

AdamH

Quote from: drsteelhammer on 12 August 2017, 06:01:57 PM
I do agree we wont get the variance out of dominion, but starting player is a lever we have to pull into one direction or another, so we can try and do our best here.

But the designer of the game had the chance to pull that lever and he decided not to. Now that's part of the game. You can pull lots of other levers too and the designer of the game decided not to pull those levers either. The game rules state these levers should not be pulled, so why pull them?

Quote from: drsteelhammer on 12 August 2017, 06:01:57 PM
I also agree with you this discussion will probaly never matter for the tournament, but that doesn't really make the decision easier (only easier to ignore)

That's not what I was trying to say -- I'm saying the only downside to my suggestion (more time) is so small, and the upside is that you don't have to change the game rules. The upside seems far better than the downside here, I feel like in a game with ties you're never going to get the downside smaller than this.

markus

I would be all up for playing the final with an even number of games until one player has won more. If it takes 20 games, it's going to be legendary.  8)
And I wouldn't mind that also in earlier rounds, although I understand that you want to have some upper bound.

But I don't see how playing an 8th game only after a draw is fairer than the current rules. If you could choose to go first and have to win, or to go second and a draw suffices, who would want to go second?

With an odd number of games, one person will have to go first more often than the other. Ideally, you want the choice of first player in the last game to not affect the win probabilities ex ante. Having the rule that a 3.5-3.5 is enough for the second player makes this choice a bit less relevant.

Donald X.

Quote from: markus on 12 August 2017, 11:50:21 PM
But I don't see how playing an 8th game only after a draw is fairer than the current rules. If you could choose to go first and have to win, or to go second and a draw suffices, who would want to go second?
Given a choice between two bad things, people can pick the one they prefer, and it may even be that everyone will pick the same one. That doesn't mean that offering people that choice is as good as it gets.