flash, I will give you a hard time about that for the rest of your life! :-P

I don't know much about the ELO ratings, but think the idea of rating diplomacy players based on their competition is awesome, is there a way to do the latter without the former, or are people just complaining without offering suggestions?

I think winning is an important part of the original game, flash, so I don't really get how your style of play fits in or how you suggest rating such play. If you are on this site just as a hobby, that's fine, but doesn't seem like you are really contributing to the discussion on a rating system. You are coming off as some sort of hippie that thinks we should all rate ourselves or that walking away from a victory is worth three hugs while a solo is only two hugs... Okay, I'm poking fun a bit, but it's a pretty artsy fartsy lovefest of not judging and complaing about others who are interested in coming up with a valid system of rating players (sorry I couldn't word this response better).

Regarding the 6 vs 1 nature of diplomacy compared to the 1 vs 1 nature of chess, I assume/believe that is accounted for in the GR calculations by the amount of rating that are at risk in any single game.

In chess the risk/reward is a 1 to 1 ratio. If you win you get a portion of the other persons rating while if you lose they get a portion.

In diplomacy if you lose you lose a portion but if you win you win a portion from 6 people, not just one. I assume the portion is therefore a smaller portion than is used in the chess ELO calculations.

Regarding six low ranked players versus one high ranked players... If you are that much better you should be able to diplom well enough to avoid the six versus one by convincing one or two that you can help them along. Or you can just avoid playing six low ranked players.

One thing I could see doing would be to limit how much rating you risk based on the average rating in the game. For example, average rating + 50% would be the max at risk. Then a high ranked playing a bunch of low ranked wouldn't risk as much of their rating so wouldn't be "afraid" to play in such a game.

Again for clarify, an adjustment to the amount of rating at risk is the adjustment that makes the 1 vs 1 rating system potentially work for a 6 vs 1 game. You risk a smaller portion to win 6 times that smaller portion (adjusted by everyone's ratings of course.)

So I respectively disagree with the statements that it is seriously flawed.

The only real flaw I see is people putting too much meaning into it. It should just indicate if someone is experienced or not which is what we all really want to know. It might even be better if the ghost ratings weren't known beyond what 10% percentile range you were in. Top 10% of players versus 10-20% and so on.

MM:

Actually I expressed explicit agreement with TMG's more detailed comments above. I therefore can hardly be expected to supply an alternative ratings system for this site.

My position is that you cannot use a one-on-one rating method where there is a group playing a game and claim any accuracy. Worse still in some respects, it distorts the play. In Chess, you win or lose entirely due to your own play against an opponent: in Diplomacy, even Genghis would have trouble winning against six countries that ally from the start.

The temptation to attack a player simply because he has a very high rating has - in my observation at least - proved too great to resist; and

the temptation to set up games in order to increase GR ratings has become an unwelcome feature of this site for me.

I withdrew from all on-line competitions and will not now play linked games in here. If that offends anyone, I am sorry, that is not my intention, but I will not let a flawed ratings system come between me and a good game. I know when a game has gone well and can easily stop before reaching the line and shake hands. Indeed, that often happens in f-2-f - we even sometimes start a new game as soon as it is clear that a player has achieved a winnable position. In here though those games would not be counted as wins as you have to actually cross the line.

Alderian: I have to disagree. I used to play Chess. I was not that good but I know that I never ever lost a game because of anything other than my own weaknesses and my opponent's skill.

In Diplomacy, I have been in games more than once where I have been told after joining that I had to be removed first because I was the most dangerous player. To suggest that I could in some way overcome that attitude is patently absurd. If six players have already agreed to take someone out, they will take that person out.

Because of the nature of on-line Dip, it happens all too often that you or I can wander into a game and find a strong group of friends who behave in the way I have described. It does not have to happen often, just once and the rating is permanently affected.

So play anonymous games. That overcomes that attitude, does it not? Afraid of multi-accounts? Then play in anon games formed on the forum so you know who is in it.

I played in a masters game with Babak recently. My previous time playing him I allied with him and he quickly screwed me over. Everyone in this masters game was expecting me to not ally with him because it was Babak and he was feared as the most dangerous. I totally used that to my advantage and had a great alliance with him and ended in a three way draw with him and hellalt.

So if you play six people and they all want to take you out, that is their stupidity and even if they didn't know you were an experienced player it probably wouldn't have been a good game anyway.

Again, anonymous games should relieve that in any case.

flashman: "the temptation to set up games in order to increase GR ratings has become an unwelcome feature of this site for me."

When I see "ghost rating" games on the forum, I don't believe those are being setup to increase GR. I believe those are setup to put together a really good game with really good players. Or are you referring to other situations?

“1) people have used it to determine their alliances [however, that likely happens through other forms of notoriety in FtF tournaments as well]”

It might be because of who I am and/or who I play, but I have never experienced this. What I have experienced is that people look at *points* to determine alliances. In truth, any indicator of ability will be used by some to justify alliance decisions. MM has even tried to get me scuttling off to attack Jacob as Russia because at that time he was on a 100% win-rate with Russia.

This is not, therefore, a phenomenon peculiar to Ghost-Ratings.

“2) I feel that it is inaccurate. Chess is a game of 1v1. Barring some major life changing event a rank 2400 player does not lose to a rank 1200 player. Diplomacy is a game of 1v6. The ELO system of Chess does not apply. If we put the rank 500 - 505 players in a game with the rank 1 player, MM, and told them they get a dollar for every time they solo, but the 6 of them get a dollar every time MM loses, MM would never win. Yet he is ranked significantly better than they are.”

In that game, Madmarx would have an expected result of a little over ½. Given that those players are decidedly average, I don’t actually think that that is unreasonable. He does, after all, have a win rate of 30% and a draw rate of 43% to boot, as well as being on a peak in rating too.

Also, the paying people a dollar thing makes your argument false, because you are adding a biased incentive to take out Mad-marx. Without those incentives, Madmarx could expect to win a large number of games.

“3) More in accuracy. Chess is Chess. The same board, the same rules, the world over. This is not true of Diplomacy. There are variants. No, I am not talking about other maps, yet. Let's just talk standard map. There is the official variant, fleet Rome. There are numerous forms of press. There are the shift left and shift right variants. There is blind variant. And now there are the MULTITUDE of map variations (I thought is was at least 50). So, does playing one form of the game make you good at playing another form of the game? They all use one GR, but I would argue that being good at one does not make you good at another.”
Bluntly, this is nonsense. Being good at fleet rome and being good at classic diplomacy are very connected. Being good at Gunboat, Public Press or other maps are connected to being good at diplomacy somewhat. That is why there is weighting.

“4) Tournaments warp the GR. They do this by changing the meaning of the word "winning." Some of the tournaments of this site promote meta-gaming, some allow for it, and some do not. Plus the "point" structure of the tournament affects whether or not solo'ing in a game is better than some other out come.”

This point is fair. People will take odd results in late games in tournaments, and particularly with the GFDT, the point system was not matched by anything available on the site. I’m not sure it makes that much of a difference in truth. About 20 players play tournament games more or less exclusively. Some others play in a mixture. Most don’t play in them.

“An epiphany: why not multiply the current GR by the win percentage of a player to come at a final GR?”

The Ghost-Ratings do actually mean something more than the rank they place you in. They represent the relative results we would expect people to get on average in a WTA game.
What you suggest would mean that Ghost-Ratings no longer had any actual meaning. I think here you are attacking the wrong target, and should look towards the scoring system that you don’t like.

“I fully agree with the comment above about the inappropriacy of applying Chess style ELO ratings to a multi-player game. It is a seriously flawed approach.”


Had I cookie-cuttered precisely the same methodology of Elo-Ratings, your criticism would be valid, however, I have made changes for precisely some of the reasons you mention.
One of the key assumptions in the Elo rating system have been changed: that the best performing player in that particular game will win. Elo considered the abilities of the two players as two normal distribution curves, so that their skills would vary from game to game in line with the probabilities described. Then you take the odds of player A playing to a higher standard than player B, and call that the odds of player A winning. Vica-versa for player B. In other words, the player who plays the better game wins.

This does not apply to multi-player games, where one player can play better than any others, but still be unfortunate in some aspect and be teamed up on. Even if Player A plays brilliantly, if players B through G don’t listen to him, he’s stuffed. All he can do is give himself a much better chance of winning than any of the other players.

The expected result calculation does reflect this: it is based on ratios of the different abilities of players rather than the probability that a certain player plays the best game. Thus it does remain fair.

@Flash

two questions:
1. You, TMG, MM and so on were all in the League. How come you all complain about the way these games were played when it was all of you against each-other? Did someone force you to play in a certain way? No.

2. Whatever troubles you might have (in terms of people metagaming against you), surely you must realize the same happens to the other top players. How successfully you deal with it is what will make the difference and will be reflected in the GR. Why is that a problem?

3. We have anonymous games now. They sort this problem. Whoever wants games clear of metagaming and GR influence can stick to these. It sounds like a very cheap excuse to blame the system when there's an easy way to avoid it, doesn't it?

I think Flash, MM, and TMG are complaining about the leagues affecting their GR. they don't mind playing games where met-ing may occur, but don't want those games to affect their ratings.

Given that League games have fewer CDs and NMRs to affect the games, I'm not sure that they are particularly poor in terms of being warped by factors other than the skills of the players.

Yeah, and this is my point. They were all three in the same League - so I believe it is not so hard to realize that to have 7 people with 50% win ratio play 4 games together and still retain their win ratio... is simply absurd.

But League games come with collusion across games (meta gaming) to make certain players get their position at the end of the league. It does have an unnatural influence when these are counted into the GR.

The Bulgarian Open will sort this out :P

If so, this seems too much like USA College Football. To many people care who "the best" is to the point where they want to change things in ways which will result in worse overall games.

I don't care who is best. I just want good games. GR are a tool for finding like skilled players, not for the purpose of gaining GR, but for the purpose of playing good games.

For me, the accuracy of GR only matters to the point that I can tell good players from bad for the purposes of created good games. I think for the most part it is accurate enough for that. Does anyone disagree with that?

I think it is great for that purpose, but I also just look at a potential players stats and see how long they have been around, what type of games they play, and if they CD or not.

As I said, it is "a tool". One of many.

Ivo, in reverse:

3) GR is not part of Diplomacy proper, nor indeed is any other ratings system. I therefore reserve the right to feel that they don't belong here and ignore them;

2) I was not claiming that I was especially singled out for being ganged up on. I was merely pointing out that it has happened and when it does it leaves a permanent stain on the GRs (and points). I have an extreme view of the importance of fairness in games and exams (my profession);

1) We had problems in the League, more than one. None of those were related to GRs though. I chose to withdraw; I was not the only one. Again, my extreme views on fairness led me to my decision.

I will just say though that we had made a firm and clear agreement before one season regarding unbalanced country allocation and this agreement was reneged on for the final game ('no player would have to accept a country three times in a season').

For the next season, I had a disagreement with the Tournament Director. This was a much more serious matter from my point of view... but not GR related.

I think Tom knows my view well enough. I suppose it will have to be one of those 'agree to disagree things'.

I think we will meet at some stage in the future, either at Cambridge or in the Houses of Parliament, along with cg, to settle the issue of who really rocks. I have no doubt that such a game will be played in the true spirit of Diplomacy.

I prefer drawing over winning.

I agree, Alderian, GR games, for me, are all about finding quality games with good players that will (more likely) not NMR or CD. The assumption that GR games are meant to boot personal GR is incorrect, but I suppose a nice bonus. Same goes for high pot games, they are meant to be good games, and a possible huge bump in points is just a nice bonus. ALSO, GR games and high pot games, anon or non-anon, in my opinion/hope/ignorance, give me comfort in thinking people won't pull this six-on-one mentality previously described, that there will be no tall poppy syndrome for anyone based solely on their name/points/GR.

Draugnar, I'm stating leagues do affect GR negatively and I'm stating that I don't like Leagues, but I'm not stating I don't like Leagues because they affect my personal GR negatively. GR is what it is, I don't play games on this site just for my GR, otherwise I would never two-way draw when I could easily solo as it is obviously throwing away GR (though I don't make the habit of throwing away victories as much as flash).

Anyway, my main gripe with League play is that in my first league, the first game two players got a two-way drew the game. In game two, those two were eliminated (largedly because they both drew the first one) and I was part of a draw. In a later game, the initial game's two drawers both lied to me and took me out, mainly because of the standings in the league. This is all understandable in league play, but I really don't like entering a game in which you WILL lose it because of a past game, but even moreso I didn't like getting an apology from one of those guys saying he actually wanted to ally with me and not the other guy but had to take me out because of the league. I honestly believe that guy, it was not just him being nice, it was a statement of fact. I don't blame that guy, I blame the league, and I don't play league games any more because of that. Suppose it's true I don't like the affect being forced to lose a game before it starts affects my GR, but the GR is not my reason for not playing leagues.

Well damn. This is a long and intense thread. I'm sorry I didn't read it sooner.

Sorry but I don't know if I can just commit the time to read it all. But.... I read the first page, and my idea is this:


To me, winning IS everything, I wholeheartedly agree with that proposition.

It takes a wily diplomat indeed to pull off a solo in a WTA game, but the reasons why are manifold.

In a WTA game especially, but as a result in many games, people start talking about what kind of draw they want almost off the bat. You'll be hearing about draws in 1905, even earlier.

Most of this, honestly, is bunk. The game usually does not work out the way you think in 1905. So to arbitrarily look and see that since Russia has 7 SCs, and is leading, and England and France both have 5, that they should three way.

To me that is sort of... how to say it... ending the game before it's due time. It's like murdering a 15 year old lol... there's a lot of life yet to live. If you know what I mean.

Instead of that, I believe, in a perfect world, Russia would be saying sweet nothings in France's ear about how they will split England and take over the map together while England is telling France now is the time to strengthen their alliance and challenge Russia.

It's a balance and it's hard for the leader to solo.

With draws though, it's even HARDER for the leader to solo. Not saying that's good or bad, I'm just saying.

Think about it. I usually play to win. I don't really like to "go for" a draw, if you know what I mean. I draw if I'm forced to. I think of lot of us do.

But it's more complex than that, because when you're the leader, the number two who wants to "go for" the draw holds all the cards.

Why? Because if you can convince him you will try for a draw, he won't attack you. But if he smells a rat and thinks you're just going for the solo, well, you won't be soloing lol.

So it's a cat and mouse game. Sometimes you corner yourself SO much in the course of trying to convince the other players that you are willing to draw, that you end up being forced to draw anyway, there is simply no way you can fight your way out.

It's a tough break.

So I think the high number of draws is not necessarily due to the system per se. I think it is more because the people who want a draw hold all the diplomatic cards.

What 7 SC endgame power doesn't want to be in on the draw? For him it's the best option!

So... I don't know. It's tough stuff.

Ideally, people would basically try to get the draw down to as few people as possible, allowing for a stab by one of the larger powers who then takes the solo.

But it's all very complex. I think if you get the right group of players, you're going to see them all going for the win and only drawing an "aw man" kind of attitude.

Anyway... that's not to mention player fatigue.

I hate that. I hate when people draw just because they don't want to play anymore. That really bothers me. I think if you sign up for a dip game you should be prepared to fight it out to the bitter end... even if that end is in 1930.

...and takes until 2030 to finish IRL?

Hmmm, I hope TGM isn't so discouraged by this thread that he is refusing to issue any more Ghost-Ratings! ;-)

It hink he will view it as our desire is to help make the ghost ratings better by brainstorming possible ways. He's a good dude, so he'll take it right. although, he has been a bit pissy to me lately, slamming me in threads and all... <grin> JK, TGM!

I know, I'm just curious when the July ratings will be out, it is the 2nd afterall...

Geez. I was about to say it's a holiday weekend, but Ghost isn't from the states, so July 4th is just another day to him: the day we cocky, arrogant, conceited Americans declared our independence from that damn British King George III and made it stick!

“3) GR is not part of Diplomacy proper, nor indeed is any other ratings system. I therefore reserve the right to feel that they don't belong here and ignore them;”
No more than a league is part of football, a trophy a part of tennis or an Elo-rating a part of chess.

“I think Tom knows my view well enough. I suppose it will have to be one of those 'agree to disagree things'.”

First name terms, Roy? :P

Fortunately I don’t need to listen to you :)

“The assumption that GR games are meant to boot personal GR is incorrect”

Equally wrong is the assumption that they would.

“Geez. I was about to say it's a holiday weekend, but Ghost isn't from the states, so July 4th is just another day to him: the day we cocky, arrogant, conceited Americans declared our independence from that damn British King George III and made it stick!”

As you may have realised, I’m kinda a fan of America.