Print

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

[ashersky]

Ahh, that's right. It has been a while. So in any case, if you can get a tie, the higher ranked player, in this case should lose points. If your scores are tie but the game isn't it should be a full win for the victor.

this actually happens in the rating system right now. I had this happen yesterday, where i lost 25 points for a tie with a player ~1500 points lower than me.

And i certainly felt it's not right. It was a game with no +buy and we traded provinces back and forth. He broke PPR, and i couldn't do anything but tie, since i was 2nd player. Why should i get penalized for this?

I agree with you, although for different reasons.

As mentioned, there are no ties in Dominion.  Either one player wins and the other loses, or they both win.  No other options.

In the case where you forced the joint victory as 2p, you both should have been given wins (as you rejoiced in your shared victory).  Your opponent would have gotten a large boost to his rating; you would have gotten a much smaller one.  All would be right with the world.

[dghunter79]

A tie is worse than a win. It's what you settle for.

If you play someone whose rank is lower than yours and you don't beat them, you should get a smaller number of points than if you win. If this person has a MUCH lower rank than you, this might be a negative score, and you'll lose points. Seems fair to me.

[Davio]

Okay, I found some other annoyance today.

I had "purchased" the final Dark Ages adventure with my coins as it was the only one I could buy with regular coins.
The adventure never worked, whenever I wanted to start the first game, the progress bar hung after 10% or so.

Today I tried again, but I got a message that I didn't have some needed cards. Great, thanks for letting me waste coins on an adventure I can't play.

[WanderingWinder]

In my opinion:

A rating system should never discourage play. If players are avoiding play because of the system, this is bad. Rating systems such as elo are always more inaccurate the further you get from each other,

I haven't actually seen any evidence of this. Certainly, it wouldn't surprise my that much, because stronger players tend to not play those much much weaker than them for a host of reasons apart from the possibility of losing rating points. Anyway, I mean to say, I haven't seen evidence that this is actually true in practice, nor especially that it *must* necessarily be true.

so you always end up with a system where advanced players avoid games with less advanced, or seek it out. Either is bad (in the latter case it means people avoid the closer ranks which are generally the best games) Either is also bad in particular for less advanced players because they are either avoided, or preyed upon.

The solution which I have seen work for this problem is to make it so that all games outside a certain range is rating free. This works quite well.

I disagree. I mean, I really have no interest in playing such a game. Well, that I don't have this interest now, either, is true as well, but I hardly think that this will help matters. And from a rating perspective, the primary purpose is to predict future outcomes and/or give an indication of playing strength - and throwing away data is not good for either of these purposes.

Beginners get to play against experts and learn from them, and the experts can experiment, be ambassadors for their game, and just have a good time. Yes, there are players who will under this system not play beginners but those are exactly the players you don't want to have playing your beginners. I would strongly recommend any game with a skilled rating system look at the extremes of the rating system, the inevitable difference that exists where experts should either avoid or seek games, and make those rating free.

Alternatively, they could just fix the system so that the expectancy curve matches with what you should actually get, such that games with big rating differences have the same overall long-run impact on ratings as any other game. To me, your 'solution' is like, upon seeing that cars traveling at current highway speeds are getting into too many fatality accidents, suggesting that we only allow bicycles on the highways.

[Jack Rudd]

I'm not really into rating systems, but can you give me examples of where a rating system that causes rating decrease when winning is used?

The ECF Grading system, as used for rating chess matches in England. Your grading performance for a single game is {Opponent's grade}+50 for a win, {Opponent's grade} for a draw, and {Opponent's grade}-50 for a loss, where {Opponent's grade} is your opponent's grade unless it differs from yours by more than 40, in which case it equals Your Grade+40 or Your Grade-40 for the purposes of that game. Your grade for the following season is then the average of all your grading performances of the single games you played this season.

Therefore, if the running average of your grading performance over the season is more than 10 points above your current grade, playing someone more than 40 points below you will depress it, even if you win. (This isn't a serious problem in actual play, because both playing more than 10 points away from your current grade over a whole season and playing players more than 40 points away from you are relatively uncommon occurrences.)

[LastFootnote]

The problem I have with gaining or losing points from a tie is that suddenly the higher and lower ranked players are playing games with different victory conditions. If you're lower-ranked, you're aiming for a win or a tie. If you have a higher rank, you're avoiding ties.

[Mic Qsenoch]

The problem I have with gaining or losing points from a tie is that suddenly the higher and lower ranked players are playing games with different victory conditions. If you're lower-ranked, you're aiming for a win or a tie. If you have a higher rank, you're avoiding ties.

The points given out are Win > Tie > Loss for both players. It's the relative values that actually matter here. There's incentive for both players to try for a tie if they think that's the best outcome they can achieve, and to try for wins if they think that outcome is more likely. I guess in practice people have a stronger aversion to losing points (so maybe the psychological aspect is more what you're talking about). But I think the proper incentives are still in place based on the relative values.

More importantly, as has been pointed out elsewhere, if a higher ranked player repeatedly ties someone who is ranked below them then their ratings should become closer. The ties indicate that their skills are similar. There's no reason to ignore the information gained from a tie game.

[Kirian]

Ahh, that's right. It has been a while. So in any case, if you can get a tie, the higher ranked player, in this case should lose points. If your scores are tie but the game isn't it should be a full win for the victor.

this actually happens in the rating system right now. I had this happen yesterday, where i lost 25 points for a tie with a player ~1500 points lower than me.

And i certainly felt it's not right. It was a game with no +buy and we traded provinces back and forth. He broke PPR, and i couldn't do anything but tie, since i was 2nd player. Why should i get penalized for this?

I agree with you, although for different reasons.

As mentioned, there are no ties in Dominion.  Either one player wins and the other loses, or they both win.  No other options.

In the case where you forced the joint victory as 2p, you both should have been given wins (as you rejoiced in your shared victory).  Your opponent would have gotten a large boost to his rating; you would have gotten a much smaller one.  All would be right with the world.

Look, everyone here and elsewhere knows that "rejoice in your shared victory" is just a cute way of saying "tie." Both players do not actually win in competitive play.  If nothing else, this would encourage everyone to play for ties.

You're just playing silly buggers with words.

[eHalcyon]

And we should note again that if ties counted as mutual victories for the purpose of ranking, you could game the system so easily by just tying repeatedly against the same player. Both your ranks would inflate.

[LastFootnote]

Look, everyone here and elsewhere knows that "rejoice in your shared victory" is just a cute way of saying "tie." Both players do not actually win in competitive play.  If nothing else, this would encourage everyone to play for ties.

You're just playing silly buggers with words.

"Shared victory" is quite meaningful if the game has at least three players and two of them tie for first place.

[Phelddagrif]

In my opinion:

A rating system should never discourage play. If players are avoiding play because of the system, this is bad. Rating systems such as elo are always more inaccurate the further you get from each other,

I haven't actually seen any evidence of this. Certainly, it wouldn't surprise my that much, because stronger players tend to not play those much much weaker than them for a host of reasons apart from the possibility of losing rating points. Anyway, I mean to say, I haven't seen evidence that this is actually true in practice, nor especially that it *must* necessarily be true.

The skill based rating systems rely on assumptions about game skill that are simply not, in general, true. For example they rely on transitivity of the skill between players. Fortunately, these assumptions are close enough to true to make it work. But, the further between skill levels the more those assumptions break down.

so you always end up with a system where advanced players avoid games with less advanced, or seek it out. Either is bad (in the latter case it means people avoid the closer ranks which are generally the best games) Either is also bad in particular for less advanced players because they are either avoided, or preyed upon.

The solution which I have seen work for this problem is to make it so that all games outside a certain range is rating free. This works quite well.

I disagree. I mean, I really have no interest in playing such a game. Well, that I don't have this interest now, either, is true as well, but I hardly think that this will help matters. And from a rating perspective, the primary purpose is to predict future outcomes and/or give an indication of playing strength - and throwing away data is not good for either of these purposes.

If you aren't interested in playing beginners now you are not in the player base I am culling. There is no reason you should either, the best games are going to be in general against closer opponents. And, if you agree with my premise (which you don't, though i believe if you look into it you will find it is well documented) that the systems are inherently bad he further players get from each other you are throwing away data rather than misusing it. Discarding outliers is perfectly fine in data analysis if done properly.

Beginners get to play against experts and learn from them, and the experts can experiment, be ambassadors for their game, and just have a good time. Yes, there are players who will under this system not play beginners but those are exactly the players you don't want to have playing your beginners. I would strongly recommend any game with a skilled rating system look at the extremes of the rating system, the inevitable difference that exists where experts should either avoid or seek games, and make those rating free.

Alternatively, they could just fix the system so that the expectancy curve matches with what you should actually get, such that games with big rating differences have the same overall long-run impact on ratings as any other game. To me, your 'solution' is like, upon seeing that cars traveling at current highway speeds are getting into too many fatality accidents, suggesting that we only allow bicycles on the highways.

Fixing this problem is not easy, I would guess not even possible since the skill set for any deep game, and the player base for that game is constantly shifting. Certainly goko isn't going to do it. The solution I propose has been used on several games just fine, and it is not radical; many ladder systems require competitors to be close in rank for a ladder match, otherwise it is off ladder.

[ashersky]

Ahh, that's right. It has been a while. So in any case, if you can get a tie, the higher ranked player, in this case should lose points. If your scores are tie but the game isn't it should be a full win for the victor.

this actually happens in the rating system right now. I had this happen yesterday, where i lost 25 points for a tie with a player ~1500 points lower than me.

And i certainly felt it's not right. It was a game with no +buy and we traded provinces back and forth. He broke PPR, and i couldn't do anything but tie, since i was 2nd player. Why should i get penalized for this?

I agree with you, although for different reasons.

As mentioned, there are no ties in Dominion.  Either one player wins and the other loses, or they both win.  No other options.

In the case where you forced the joint victory as 2p, you both should have been given wins (as you rejoiced in your shared victory).  Your opponent would have gotten a large boost to his rating; you would have gotten a much smaller one.  All would be right with the world.

Look, everyone here and elsewhere knows that "rejoice in your shared victory" is just a cute way of saying "tie." Both players do not actually win in competitive play.  If nothing else, this would encourage everyone to play for ties.

You're just playing silly buggers with words.

Rules are rules, man.  What do official tournaments do with shared victories?  Play another game?

How does chess treat the stalemate?  That's the analogy here, isn't it?

I still think it is Win > Shared Victory > zero > Loss.

Think of it as half a win.  It's a viable option for advanced players to salvage a game.  It helps combat 1p advantage.  It's built in to the game for a reason.  I don't think DXV cared a bit how your feelings were affected by the words victory, defeat, or tie, so calling something the game creator purposefully included in a competitive game "silly buggers with words" is insulting to the man.

[yudantaiteki]

Look, everyone here and elsewhere knows that "rejoice in your shared victory" is just a cute way of saying "tie."  Both players do not actually win in competitive play.

That would depend on the rules for the competitive environment.  Dominion has no rules for tournaments or any structure beyond a single game (with the exception of the rule that the loser goes first next game).  It is certainly not impossible to create a league or tournament structure where a "tie" is counted as a win for both players -- depending on the structure there may be no actual difference between a tie and a double-win, but it might make a difference.

Even in ISOdom it would make a difference; let's say we're playing to 3 games and I'm down 1-2.  If we tie the next game then we play a fifth game, but if it's considered a double win then my opponent advances and I'm knocked out.

[Kirian]

How does chess treat the stalemate?  That's the analogy here, isn't it?

Yes.  And in chess, a stalemate is a forced draw.

Think of it as half a win.

In other words, what just about every other competitive environment calls a draw or a tie.

It is certainly not impossible to create a league or tournament structure where a "tie" is counted as a win for both players -- depending on the structure there may be no actual difference between a tie and a double-win, but it might make a difference.

It is also not impossible to make a tournament structure where each match involves five games, and then a single die roll determines who actually wins.  You can do it, but no one would.

[ashersky]

Think of it as half a win.

In other words, what just about every other competitive environment calls a draw or a tie.

But if a win is worth a 50 point jump in ranking, isn't half a win worth 25?

The argument you might be making that I'm misunderstanding is that half a win is also a half a loss, and so should be -25 points (in the case where loss would net you -50).  Is that what it is?

[WanderingWinder]

How does chess treat the stalemate?  That's the analogy here, isn't it?

Yes.  And in chess, a stalemate is a forced draw.an immediate draw by rule.

Made that more precise for you.

See, people use 'stalemate' to mean all kinds of draws, when indeed it's only a very specific thing - the side to move is unable to without being in check.

This message has been brought to you by the coalition for chess pedantry and is not meant in any way to attack or defend either of your positions on the issue at hand.

[WanderingWinder]

Think of it as half a win.

In other words, what just about every other competitive environment calls a draw or a tie.

But if a win is worth a 50 point jump in ranking, isn't half a win worth 25?

The argument you might be making that I'm misunderstanding is that half a win is also a half a loss, and so should be -25 points (in the case where loss would net you -50).  Is that what it is?

I don't understand the semantics. A 50 point jump in WHAT ranking?

Generally, you see it on a 0 to 1 scale. A loss is 0% of a win - so 0. A win is an entire win - so 1. A draw or half a win or rejoicing in shared victory all mean the same thing - whatever you want to call it, it's 0.5.

So......

[ashersky]

Think of it as half a win.

In other words, what just about every other competitive environment calls a draw or a tie.

But if a win is worth a 50 point jump in ranking, isn't half a win worth 25?

The argument you might be making that I'm misunderstanding is that half a win is also a half a loss, and so should be -25 points (in the case where loss would net you -50).  Is that what it is?

I don't understand the semantics. A 50 point jump in WHAT ranking?

Generally, you see it on a 0 to 1 scale. A loss is 0% of a win - so 0. A win is an entire win - so 1. A draw or half a win or rejoicing in shared victory all mean the same thing - whatever you want to call it, it's 0.5.

So......

If I manage to beat you on Goko (say you misclick every turn or something), my 5200 ranking goes to 5400 and your 9001 ranking drops to 8501.  If you beat me (say you played with your eyes closed), your ranking goes to 9003 and mine goes to 5190.

If we share a victory, I think one of two things is fair:

Option 1: WW to 9002 and ash to 5300
Option 2: WW to 8751 and ash to 5195.

Option one is a shared victory (half a win each).  Option 2 is a shared loss (half a loss each).  What I think would be unfair is for our shared victory to result in WW to 8501 and ash to 5400 just because you should beat me 999/1000 times.

[Kirian]

Ah, I see your problem!  You are attempting to directly equate "victory" with "rating increase."  Instead you should equate "rating increase" with "the rating system now considers player A more likely to defeat player B than it did before this game."

Here's how a normal rating system works under these circumstances.  Let's take two players, A and B, with ratings 4000 and 6000.  Player B should, on average, win more games--let's say 60% because, hey, that makes the numbers look like they make sense.

So these two players play one game.  If player B wins, he's won 100% of that game, which is better than the expected 60%.  His rating goes up.  Player A did worse than expected (0% vs 40%) and so her rating goes down.  If player A wins, she did much better than expected (100% vs 40%), so her rating increases quite a bit more, and player B's rating decreases quite a bit more.

Now, take a game where players A and B draw.  Player A did better than expected (50% vs. 40%) and has earned a small rating increase.  Player B did worse than expected (50% vs. 60%), and gets a small rating decrease.

Within this sort of rating system, that's what a tie does.  "Half a win" is equivalent to "Within this schema, won 50%, which is [better|worse] than expected."

[eHalcyon]

Think of it as half a win.

In other words, what just about every other competitive environment calls a draw or a tie.

But if a win is worth a 50 point jump in ranking, isn't half a win worth 25?

The argument you might be making that I'm misunderstanding is that half a win is also a half a loss, and so should be -25 points (in the case where loss would net you -50).  Is that what it is?

I don't understand the semantics. A 50 point jump in WHAT ranking?

Generally, you see it on a 0 to 1 scale. A loss is 0% of a win - so 0. A win is an entire win - so 1. A draw or half a win or rejoicing in shared victory all mean the same thing - whatever you want to call it, it's 0.5.

So......

If I manage to beat you on Goko (say you misclick every turn or something), my 5200 ranking goes to 5400 and your 9001 ranking drops to 8501.  If you beat me (say you played with your eyes closed), your ranking goes to 9003 and mine goes to 5190.

If we share a victory, I think one of two things is fair:

Option 1: WW to 9002 and ash to 5300
Option 2: WW to 8751 and ash to 5195.

Option one is a shared victory (half a win each).  Option 2 is a shared loss (half a loss each).  What I think would be unfair is for our shared victory to result in WW to 8501 and ash to 5400 just because you should beat me 999/1000 times.

Did you ignore my note that you can't have both players' scores increase on ties?  Consider two players who repeatedly play each other and it just so happens that they both tie repeatedly.  What do you think should happen?  By your logic, both of their ratings should increase.  If they tie a large number of games, their scores will become extremely large.

But this is not good for the system.  The system learns nothing about the skills of the players, save that they both seem to be very close in skill.  This info is not correctly represented by increasing both their ratings to astronomical levels.

If the two players started with different ratings and they tie, you should not expect both ratings to go up.  Their ratings should get closer together; whoever was expected to win should lose rating and whoever was expected to lose should gain.



And I want to point out that "half a win" is not a good way to look at it.  A tie is rather the midway between a win and a loss.

In your example...

you win: +200 your rating, -500 WW's rating
WW wins: -10 your rating, +2 WW's rating

tie op1: +100 your rating, +1 WW's rating
tie op2: -5 your rating, -250 WW's rating


But the logical result in this analogy is actually:

(+200-10)/2 = 95
(-500+2)/2 = -249

tie: +95 your rating, -249 WW's rating
WW to 8752, ash to 5295.

[Davio]

My main grievance with Goko is that it's not pro friendly at all, in any way.

• The rating system is bad
• The UI is bad, mistakes are easily made
• There are still bugs with certain cards
• The lobby is horrible, making it next to impossible to find a good match
• There is no "same starting hands" requirement or something similar
• There is no point tracker, this makes it unfair for those who don't have a browser add-on
• They don't understand basic math
• Having players with different card sets makes the ratings hard to justify

Whenever I play at Goko, I just feel out of place. I always think "this site isn't meant for me", but out of some sense of loyalty I keep playing Dominion there because it's the only place I can. I don't think this will last though, at some point I'll have had enough and finally learn Innovation.

[yudantaiteki]

Yeah, I play with a group of college friends most weekends; we bought a goko account immediately but we seem to play less and less each week and we've talked about trying to play Innovation as well (which we've all played before; I bought it as a wedding present for two of them).

The primary annoyance for us is that you can't make a random deck with certain criteria, so we often just have to manually make the deck to get Dark Ages in or remove a few cards that annoy people like Swindler.  I'm the only one in the group who played random matches on Isotropic, but I haven't bought my own goko account yet because of Goko's ridiculous lobby design and puzzling lack of automatch.

I wouldn't necessarily say it's not "pro friendly", I would just say it's not friendly in general.

[yudantaiteki]

It is certainly not impossible to create a league or tournament structure where a "tie" is counted as a win for both players -- depending on the structure there may be no actual difference between a tie and a double-win, but it might make a difference.

It is also not impossible to make a tournament structure where each match involves five games, and then a single die roll determines who actually wins.  You can do it, but no one would.

The Dominion rules do not mention die rolls.  The Dominion rules do mention shared victories.

[ragingduckd]

My main grievance with Goko is that it's not pro friendly at all, in any way.

I expect this is mostly a matter of their business model.  "Oh, you've already bought all the cards?  Gee... we'd like to help, but right now we're focusing on customers we can still make money from."

My personal pet peeve: https://getsatisfaction.com/goko/topics/rejoin_delay_after_being_kicked#reply_11963152

[Davio]

Well, it's not that I blame them in any way, they're running a business of course.

It's just that what it ended up being is miles from what I'm looking for in an online Dominion implementation. And their business model doesn't appeal to me: "So you physically have all the sets? We don't care, just buy them all again"
Maybe a monthly (or yearly) license fee which just gives you all the cards could have convinced me, in its current form it doesn't look like I'll be paying soon.

I understand I'm not part of their target audience, but I'm probably not alone and it's so sad to see so many good players leave the arena.