Why only single tie breaker criteria?

Forum rules
Please DO NOT POST BUGS on this forum. Please report (and vote) bugs on : https://boardgamearena.com/bugs
Post Reply
User avatar
ij_26
Posts: 5
Joined: 29 December 2023, 07:38

Why only single tie breaker criteria?

Post by ij_26 »

The tie breaker criteria is just number of development cards. So what then when more than one player scores the same 𝘢𝘯𝘥 they have the same number of cards at the end of game? I'm sure this criterion has been reached by many players in the history of BGA playing Splendor. Why didn't the developers go further to break tie further (say...opponents who have reserved fewest cards, number of chips etc, remaining end of game)? Because without further tie-breaking criterion, the player with the higher ELO always gets penalized and that's a horrible system.
Ceaseless
Posts: 338
Joined: 12 November 2022, 17:06

Re: Why only single tie breaker criteria?

Post by Ceaseless »

If both players played at a close enough level, then there's nothing wrong with a draw. If both players play at the same level when ratings say they shouldn't, then the higher rated player should lose rating to the lower rated player. Just adding things to a list to try and reduce draws isn't really worth it. Sometimes neither player stood out enough for the win over the other.

That said, I've never been a big fan of the tiebreaker system in place for Splendor, having fewer development cards being rewarded in the rules seems strange to me. I'd have rather given a tiebreaker edge to whoever went last, as first turn advantage seems like a valid natural edge in the game. Not because there would be no draws this game, though it would do that too, but because it would help balance the variance caused by turn order.

It's a Splendor issue though, I'm not aware of BGA having any impact on this. They're just playing by the rules of the game.
User avatar
ij_26
Posts: 5
Joined: 29 December 2023, 07:38

Re: Why only single tie breaker criteria?

Post by ij_26 »

Ceaseless wrote: 10 April 2024, 18:54 If both players play at the same level when ratings say they shouldn't,
So the the higher ELO player should get points because they are expected to play better?? That doesn't make sense.
Ceaseless wrote: 10 April 2024, 18:54 I'd have rather given a tiebreaker edge to whoever went last, as first turn advantage seems like a valid natural edge in the game.

It's a Splendor issue though, I'm not aware of BGA having any impact on this. They're just playing by the rules of the game.
Disagree with this as I think last player has a indeed a great advantage as the game ceases as soon as that player hits 15. First, 2nd, and so on can get to 15 first but last player still has chance to swing to 15+ if they were at 13-14.
Difference of opinion on this matter but oh well.

And while it is true that BGA has to emplace to rules set by game publisher/creator, the ELO is clearly the ball of BGA. The site managers that came up with the ELO math could have easily put in the program not to penalize higher rated player when there is a tie. To me is complete nonsense. This is the reason I gave up on playing Connect 4 where a stalemate(..err tie..) happens often. I won't going to the rabbit hole of that discussion as I'm aware there's a whole forum post on this issue but it seems not too many in community is bothered enough by BGA on this faulty 'in-case-of-tie-ding-the-higher-player'
Ceaseless
Posts: 338
Joined: 12 November 2022, 17:06

Re: Why only single tie breaker criteria?

Post by Ceaseless »

ij_26 wrote: 11 April 2024, 06:36
Ceaseless wrote: 10 April 2024, 18:54 If both players play at the same level when ratings say they shouldn't,
So the the higher ELO player should get points because they are expected to play better?? That doesn't make sense.
Ceaseless wrote: 10 April 2024, 18:54 I'd have rather given a tiebreaker edge to whoever went last, as first turn advantage seems like a valid natural edge in the game.

It's a Splendor issue though, I'm not aware of BGA having any impact on this. They're just playing by the rules of the game.
Disagree with this as I think last player has a indeed a great advantage as the game ceases as soon as that player hits 15. First, 2nd, and so on can get to 15 first but last player still has chance to swing to 15+ if they were at 13-14.
Difference of opinion on this matter but oh well.

And while it is true that BGA has to emplace to rules set by game publisher/creator, the ELO is clearly the ball of BGA. The site managers that came up with the ELO math could have easily put in the program not to penalize higher rated player when there is a tie. To me is complete nonsense. This is the reason I gave up on playing Connect 4 where a stalemate(..err tie..) happens often. I won't going to the rabbit hole of that discussion as I'm aware there's a whole forum post on this issue but it seems not too many in community is bothered enough by BGA on this faulty 'in-case-of-tie-ding-the-higher-player'
The higher rated player should lose points for a draw vs a lower rated player. They were supposed to play better, so the fact they apparently didn't play better is a fine justification for the loss of points.

This "advantage" you speak of isn't really a thing. Splendor rules are set so that everyone has the same number of turns. The reason the game ends if the last player hits 15 is because they pulled it off after everyone had their chance to and failed with the same number of turns. If the first turn player hit 15, they pulled it off having had one additional turn over everyone else, so the rules give everyone else that same final turn, since the game is based on how many turns it takes to reach 15+. The main difference is that the first turn player got access to the cards first, and has a lead in tempo over the chips, so they have better access to cards from the start.

Connect Four is not a particularly drawish game, lol. Not in its default 6x7 no extra opening rules. The first turn advantage in that game is brutal. If there was a tie, someone messed up. Higher rated player loses elo for a draw is the ideal. If there was no rating change for a draw, that would be grounds for something to change so it goes back to costing elo to draw. Fortunately, BGA got this one right, it's just basic elo. If you want to know what a draw costs you just do the math. Unless you tried beating up a much lower rated player and failed miserably with a draw though, the loss is going to be pretty tiny. I'm not aware of a forum thread on this either.

I'm not sure why you're talking about not playing games due to this. You haven't played many games yet, so all of your game ratings are really low, meaning any drawing issues should favor you or cause minimal change. Even losses don't do much atm. So unless you're just posting with an alt account I'm not following the train of thought with respect to impact here. Unless you are saying your reasons for not playing are not due to how it affects you.
User avatar
ij_26
Posts: 5
Joined: 29 December 2023, 07:38

Re: Why only single tie breaker criteria?

Post by ij_26 »

"So the the higher ELO player should get points because they are expected to play better?? That doesn't make sense." Replace "points" with "punish". Stupid typo on my part.
Ceaseless wrote: 11 April 2024, 13:27 The higher rated player should lose points for a draw vs a lower rated player. They were supposed to play better, so the fact they apparently didn't play better is a fine justification for the loss of points.
Still is nonsense. This is clearly an arbitrary opinion. They were "𝑠𝑢𝑝𝑝𝑜𝑠𝑒𝑑 𝑡𝑜" nothing. Humans are humans. And there's much luck/unluck involved in Splendor with its card shuffle. With both of these combined there's much unpredictable. There's numerous players here that will play with low ELO simply because they're new to BGA but master-level players in disguise, on the warpath and knocking off opponents, reaching 400+ELOs in less than 20 games. Those players ofcourse didn't draw but I'm making point to the unpredictability factor.
Ceaseless wrote: 11 April 2024, 13:27 If there was no rating change for a draw, that would be grounds for something to change so it goes back to costing elo to draw. Fortunately, BGA got this one right, it's just basic elo.

What are you talking about?? Grounds for 𝑤ℎ𝑎𝑡 to change? Why should there by 𝙖𝙣𝙮 change? BGA did NOT get it right. Tie is a tie is a tie unless you have the tiebreaker. If tie breaker is also tied then let ELOs be. Without any hits to the higher ELO.
Ceaseless wrote: 11 April 2024, 13:27 meaning any drawing issues should favor you or cause minimal change.
That's your opinion. Those are my ELOs and any loss of them due to draw, be it -1 or -15, I have issues with.

I will cease it with you here as we will disagree on most points.
Ceaseless
Posts: 338
Joined: 12 November 2022, 17:06

Re: Why only single tie breaker criteria?

Post by Ceaseless »

ij_26 wrote: 11 April 2024, 14:50 "So the the higher ELO player should get points because they are expected to play better?? That doesn't make sense." Replace "points" with "punish". Stupid typo on my part.
Ceaseless wrote: 11 April 2024, 13:27 The higher rated player should lose points for a draw vs a lower rated player. They were supposed to play better, so the fact they apparently didn't play better is a fine justification for the loss of points.
Still is nonsense. This is clearly an arbitrary opinion. They were "𝑠𝑢𝑝𝑝𝑜𝑠𝑒𝑑 𝑡𝑜" nothing. Humans are humans. And there's much luck/unluck involved in Splendor with its card shuffle. With both of these combined there's much unpredictable. There's numerous players here that will play with low ELO simply because they're new to BGA but master-level players in disguise, on the warpath and knocking off opponents, reaching 400+ELOs in less than 20 games. Those players ofcourse didn't draw but I'm making point to the unpredictability factor.
Ceaseless wrote: 11 April 2024, 13:27 If there was no rating change for a draw, that would be grounds for something to change so it goes back to costing elo to draw. Fortunately, BGA got this one right, it's just basic elo.

What are you talking about?? Grounds for 𝑤ℎ𝑎𝑡 to change? Why should there by 𝙖𝙣𝙮 change? BGA did NOT get it right. Tie is a tie is a tie unless you have the tiebreaker. If tie breaker is also tied then let ELOs be. Without any hits to the higher ELO.
Ceaseless wrote: 11 April 2024, 13:27 meaning any drawing issues should favor you or cause minimal change.
That's your opinion. Those are my ELOs and any loss of them due to draw, be it -1 or -15, I have issues with.

I will cease it with you here as we will disagree on most points.
Why yes... the higher rated player is meant to perform better, so a draw will cost them. If I was playing Connect Four and got paired with someone still working out the basics, idk, a 350 for example so I have numbers to plug into the formula, a draw would be them really overperforming and me really underperforming. I'd lose nearly 8 points if I were to somehow tie a 350, and I'd deserve it. Now that's an extreme case, if I were to face a more reasonable rating, maybe a 600, then we'd be talking a loss of around 3 points. It's not arbitrary, it's written in the formula how it works out, a quick look at the BGA forums and some basic math and you can crunch the numbers.

Ignoring your exaggeration of ratings of 400+, which are nothing too wild, as somehow being "master level" (BGA has a few random titles they throw around, though if you wanted something to call arbitrary, those would be it), players that are underrated due to being new are quickly moved up with their higher k factor speeding up the process. The players that face them 1-2 times before that are minimally affected, elo straightens this out easily with play. This luck factor is also being exaggerated by you. While luck affects the game, better players still win more often at a decent enough rate that players even make it to the 800s in Splendor. The truly high-luck games get nowhere near those numbers, even with high activity. A few unlucky games still get sorted out with continued play regardless, since when your rating is below your play level, you get more for your wins/less for losses until it balances out again (and the other way around for those who are temporarily overrated from a lucky streak.)

As for the "grounds for what to change" fortunately nothing needs to be changed with the current system with how it handles draws, but if for some reason BGA had made a horrible mistake and made draws no change, that would be something BGA would want to fix so it was working properly. Luckily, BGA nailed this one.

Why yes... the higher rated player is meant to perform better, that's how elo works, so a draw will cost them. If I was playing Connect Four and got paired with someone still working out the basics, idk, a 350 for example so I have numbers to plug into the formula, a draw would be them really overperforming and me really underperforming. I'd lose nearly 8 points if I were to somehow tie a 350, and I'd deserve it. Now that's an extreme case, if I were to face a more reasonable rating, maybe a 600, then we'd be talking a loss of around 3 points. It's not arbitrary, it's written in the formula how it works out, a quick look at the BGA forums and some basic math and you can crunch the numbers.

Ignoring your exaggeration of ratings of 400+, which are nothing too wild, as somehow being "master level" (BGA has a few random titles they throw around, though if you wanted something to call arbitrary, those would be it), players that are underrated due to being new are quickly moved up with their higher k factor speeding up the process. The players that face them 1-2 times before that are minimally affected, elo straightens this out easily with play. This luck factor is also being exaggerated by you. While luck affects the game, better players still win more often at a decent enough rate that players even make it to the 800s in Splendor. The truly high-luck games get nowhere near those numbers, even with high activity. A few unlucky games still get sorted out with continued play regardless, since when your rating is below your play level, you get more for your wins/less for losses until it balances out again (and the other way around for those who are temporarily overrated from a lucky streak.)

As for the "grounds for what to change" fortunately nothing needs to be changed with the current system with how it handles draws, but if for some reason BGA had made a horrible mistake and made draws no change, that would be something BGA would want to fix so it was working properly. Luckily, BGA nailed this one.
User avatar
ij_26
Posts: 5
Joined: 29 December 2023, 07:38

Re: Why only single tie breaker criteria?

Post by ij_26 »

Ceaseless wrote: 11 April 2024, 18:50 Why yes... the higher rated player is meant to perform better, so a draw will cost them. If I was playing Connect Four and got paired with someone still working out the basics, idk, a 350 for example so I have numbers to plug into the formula, a draw would be them really overperforming and me really underperforming. I'd lose nearly 8 points if I were to somehow tie a 350, and I'd deserve it. Now that's an extreme case, if I were to face a more reasonable rating, maybe a 600, then we'd be talking a loss of around 3 points. It's not arbitrary, it's written in the formula how it works out, a quick look at the BGA forums and some basic math and you can crunch the numbers.
Ceaseless wrote: 11 April 2024, 18:50 fortunately nothing needs to be changed with the current system with how it handles draws, but if for some reason BGA had made a horrible mistake and made draws no change, that would be something BGA would want to fix so it was working properly. Luckily, BGA nailed this one.
Your inflation and insistence of how BGA "nailed it" "or got it right"is plain sickening and suspect, leads me to think you're involved in the platform other than simply as a player. Draw is a draw. No one 'deserves' any hit for it. That's your opinion. Nothing logical about that.

The formula is broken. They could have coded the formula better or the platform overall (as they have done with letting every player stay at minimum 100) not to hit higher ELO players for draws or myriad of other examples. One of examples being in a multiplayer table, a higher ELO player still losing points to lower ELO opponents, even if they were winning because one of the players was ejected for time.

The platform is good for making boardgames accessible to many people worldwide at a low cost but that's where it ends. Leaving the ELO calculation to make hit higher level players on draws or numerous other instances is complacent on the platform's part.
Ceaseless
Posts: 338
Joined: 12 November 2022, 17:06

Re: Why only single tie breaker criteria?

Post by Ceaseless »

ij_26 wrote: 04 May 2024, 22:09
Ceaseless wrote: 11 April 2024, 18:50 Why yes... the higher rated player is meant to perform better, so a draw will cost them. If I was playing Connect Four and got paired with someone still working out the basics, idk, a 350 for example so I have numbers to plug into the formula, a draw would be them really overperforming and me really underperforming. I'd lose nearly 8 points if I were to somehow tie a 350, and I'd deserve it. Now that's an extreme case, if I were to face a more reasonable rating, maybe a 600, then we'd be talking a loss of around 3 points. It's not arbitrary, it's written in the formula how it works out, a quick look at the BGA forums and some basic math and you can crunch the numbers.
Ceaseless wrote: 11 April 2024, 18:50 fortunately nothing needs to be changed with the current system with how it handles draws, but if for some reason BGA had made a horrible mistake and made draws no change, that would be something BGA would want to fix so it was working properly. Luckily, BGA nailed this one.
Your inflation and insistence of how BGA "nailed it" "or got it right"is plain sickening and suspect, leads me to think you're involved in the platform other than simply as a player. Draw is a draw. No one 'deserves' any hit for it. That's your opinion. Nothing logical about that.

The formula is broken. They could have coded the formula better or the platform overall (as they have done with letting every player stay at minimum 100) not to hit higher ELO players for draws or myriad of other examples. One of examples being in a multiplayer table, a higher ELO player still losing points to lower ELO opponents, even if they were winning because one of the players was ejected for time.

The platform is good for making boardgames accessible to many people worldwide at a low cost but that's where it ends. Leaving the ELO calculation to make hit higher level players on draws or numerous other instances is complacent on the platform's part.
While I appreciate the compliment, you give me too much credit calling the drawing system my opinion, it's just widely implemented common sense. The basics of the elo system are not unique to BGA but are seen all over the place, draws for a large enough rating gap hitting the higher rated players are simply the norm. So while I'd love to take the credit, I'm nothing more than a positive observer. :D

BGA's rating system could be improved upon. They could delete the game progression percentage nonsense, reduce rating gains for winning vs vastly lower rated players to 0 instead of some tiny decimal to prevent the abuse by players of beating up absurdly lower rated players for free points, and if it's still a thing, remove the draws when a player quits, the quitter loses to everyone else, no outcome between any other players. The 100-rating floor is a joke, making 100 the new 1 was nothing special. At most it probably lets the rating abusers have a slightly easier time beating up players stuck at 100 since now they won't go any lower, adding to the incremental climbing they work on.
pjt33
Posts: 215
Joined: 05 April 2020, 15:35

Re: Why only single tie breaker criteria?

Post by pjt33 »

ij_26 wrote: 11 April 2024, 14:50 "So the the higher ELO player should get points because they are expected to play better?? That doesn't make sense." Replace "points" with "punish". Stupid typo on my part.
Your Elo rating is not a matter of punishments and rewards: it's an estimate of how well you play the game. If two people tie then they played equally well, so their ratings should move closer to each other. There are valid criticisms to the way BGA uses Elo, but they're issues with how well it works in estimating skill.
cowboy_dan
Posts: 71
Joined: 10 May 2015, 22:56

Re: Why only single tie breaker criteria?

Post by cowboy_dan »

Another way to think of it is that a draw is like 0.5 win and 0.5 loss. So, if you are 600 elo and opponent is 400, a win gains you 5 points and a loss costs you 15. One win and one draw is +5 -15 = -10. Half of that is -5. This example calculation backs up the intuitive explanations of previous posters, where doing just as well as someone with a different elo moves the two of you together (and a draw with an opponent of equal elo keeps things the same).
Post Reply

Return to “Splendor”