MM, with respect, you're making an unwarranted assumption: that someone would prefer a 2-5-0 record to a 0-0-7 record. I tend to play pretty conservatively, and would probably prefer the latter to the former - and that has diddly to do with my GR, or the points, or anything else - it's just my preference.

Thus, I think the issue is less an issue of GR/Points influencing choice than of players preferring to draw rather than risk a loss. Obviously, I feel that is a perfectly rational strategy; equally obviously there are those who disagree, some quite strongly.

Also, if your thesis is that TGM is contemplating leaving the site due to the GRs, asking him to redo all the calculations isn't likely to make him want to stick around ;)

@stratgos

TMG!=TGM

@Strat, I think he meant TheMasterGamer, not Ghost...

But I'm with you on the win versus draw thing.

Win if you can.
If you can't, be part of the Draw.
Otherwise Survive.
And last but not least, take out the a-hole who took you out.

That means 7 draws are worth way more than 2 wins, cause all the rest are a loss of one form or another.

I'm not trying to question an individuals preference to draw 7 out of 7 versus win 2 and lose 5. I'm not trying to question how an individual chooses to play this game. I am trying to question how the best method of ranking people on this site (which most agree is the GR) might be further fine-tuned to more accurately account for how amazingly much more difficult it is to solo a WTA game rather than draw a WTA game.

I suppose this raises the whole question of if the GR affects how people play compared to if the GR merely documents what people have done. There's lots of good stuff to discuss here!

For example, TMG has told me I play the game "wrong", that this game is not called "Allies", or something to that affect, that I should be more focused on winning than drawing. I'm confident that TMG's view of this game is that winning is by far the main point of this game, and I think he has a very valid point regardless of whether or not I could actually win more and regardless of whether or not changing the GR formula would actually change my way of playing (but merely change how I am ranked to more accurately describe my inability to solo, especially as compared to TMG).

Winning is all that matters. Draws are only to stop solos.

rlumley, how do you propose your view might manifest itself in fine-tuning the GR?

Rlumley your view goes further to strengthen madmarxes point that gr must be finetuned to reflect how hard it is to solo in wta. In fact, I would go as far as to say that your observation is redundant since mm's point acknowledges it, besides the point since this thread is about gr not strategy, and reflects a little bit your lack of understanding of what goes on in this site via a via madmarxs experienced view!

Madmarx, I have played dozens of WTA games this past month and couldn't agree with you more - GR should reward solos much higher than it does with draws. But then again, that might influence strategy and make it even harder to solo!

The key is in finding a balance. I know the GR is based on a formula used in Chess rankings, but then stalemates in chess are rare compared to 3 and 4 way draws in Diplomacy, so perhaps it does need a little tweaking to account for the greater occurance of a draw and the lesser occurance of a win, especially in WTA. Wins are a bit more common in PPSC because of the survivor getting something out of it.

Actually GR is based on a statistical method used to assign players probabilities of victory in chess, but it's not based on the actual chess formula itself. The dip formula is all Ghost's original work.

my experience from two german platforms like this and lots of Email diplomacy is to benefit solo´s by not dividing the points between the winner and the survivors.
the option of winner takes all is the true gameform if it´s about your system of dividing points.
I understand your fellow "TGM" by saying this game is not called allies. But is it really important to find another why of showing who is the best player or one of the best? The points behind a name for me are showing how the player cares about his gameplay and "how long" he plays on this website.

Well, the points behind the name mean absolutely nothing. You have 55, I have 65 as of this writing. And total points mean little as there are plenty of ways to pad points by playing in PPSC games and going for the strong second all the time, promising the leader you'll help him get the win. The GR is something entirely different, however, as it doesn't reward (or shouldn't) that kind of play.

And GR rewards a great performance against a superior competitor more than a great performance against an inferior opponent and punishes conversely. Lose, but to a top guy, take less of a hit. Lose to a noob, get hit hard.

"rlumley, how do you propose your view might manifest itself in fine-tuning the GR?"

Well since draws are only a bunch of people all losing at the same time, I'd suggest that you don't get any points for draws...

How does that work in a zero sum system when suddenly everyone has no points? Does that mean nobody loses points either? That's hardly fair to the two or three way legit draw where the rest of the players (maybe some even better) have been completely eliminated and a rela stalemate line has been reached.

I would also point out, rlumley, that you have twice as many draws as wins. So your GR would be significantly lower if draws didn't improve your lot.

Please note, this whole discussion may be completely worthless in practice, as I think the only real problems that this thread likely points out is with accurately ranking a handful of players. For the large majority of people on this site, I think the ranking is likely extremely accurate. That said, I still think it's an interesting discussion and worth thinking about a bit, if you like that kind of thing.

I get what MadMarx is saying. More often than not, I play for the win, at the expense of being assured a draw. I thin my stats (9 wins 6 draws) reflectst this. However, if I play that way and don't win I get a survive. Now, obviously a draw is better than a survive, so I'm shooting myself in the foot. Obviously you can make an argument that playing that way means I'm not that good of a player, and my GhostRanking (about 230) can either confirm or deny that depending on your scope. However, I think what MadMarx is saying that with the current system, playing like I do doesn't give the player enough credit for the occassional win, when compared to the possible draws if played the other way.

@ MadMarx - I either got your point or was so far off the ball as to be ridiculous.

Or, LJ, that playing for the draw gives people, including myself, too much credit.

Well, to me a three way draw is worth 1/3rd of a win, a four way 1/4th, etc. So to adjust the distribution, divide the numbers lost and gained by the number of people in the draw. Get in a 6 way draw and you get next to nothing. Pull off a stalemate with one other player and you get a nice chunk, but not nearly as nice as if you had won it.

Well, I don´t think that just playing for the win should be rewarded. Like stratagos said, no one can say if it´s better to have 2 wins and 5 survivals or defeats or 7 draws. It depends on how you see and play the game, there isn't a right or wrong here.

So, you can point that the way GR is calculated affect the way people play, favoring draws, but if you change it to reward less the draw, you will also affect the gameplay, because people will try the win at all costs (maybe even metagaming, like "I let you win this game and you let me win the next one....") so I am not sure why this will be better overall.

Draugnar, that's exactly how the GR is currently set up, so that's a "vote" to change nothing. Draws are pretty common, so I'm not so sure a win should equal the sum of its parts, the sum of its draws, as a solo is "exponentially" more difficult, at least conceptually. For me, I'm MUCH more proud of a WTA solo than a set of three three-way WTA draws.

As useful a tool as the ghostrating is I'm not sure that the points awarded, however they are devided via wins or draws, is going to effect how people play the game. In all honesty I have yet to play in a game where the winner, be it myself or another, has won without at least one of the other six players acting like a blinkered fool and not noticing the blatently obvious solo until the victor's tanks were rolling through his capital (perhaps a slight exaggeration, but it is true to say I have not yet seen a win that was easily avoidable if others had been alert to it). In short I would suggest that a draw arising from a hard fought game has a higher morale value than a win that owes more to Germany entering his moves after a weekend's heavy drinking than any great tactical nouse on the part of the victor.

Still, it is clear to even the casual observer that my experience of the most hard fought games between Diplomacy greats is somewhat limited. If I am talking out of my arse feel free to point it out :).

Can someone post the laymans version of the maths behind the GR?

Draugnar - The super simplified version is that you compare everyone's ratings, and this produces an expected result between 0 and 1 (i.e. if everyone had the same rating, then the expected result for each person is 1/7, but if someone is higher/lower, then their expected result changes to reflect this). This is then compared to the actual result, and if you did better, your rating goes up, and if you did worse, then your rating goes down. So if your expected result in a game was 1/5 (since you had a higher rating) and you were part of a 3-way draw, then your result was 1/3, so your rating would increase. For WTA, you get either 1 (for a win), 1/people (for a draw) or 0 (for a loss). For PPSC, you get 18/34 for a win (as I recall), 1/people (for a draw) or SCs remaining/34 for a survive/defeat.

I’ve heard that from TMG that he is planning to leave. He is not the first to decide to leave having been frustrated by the tendency to draw, Noodlebug did so after being asked if he wanted a draw a few turns into a WTA game.
Noodlebug was prior to the Ghost-Rating system, but post points. I think that the issue would not be Ghost-Rating system rather than the existence of a rating system making evaluations of the result people have achieved. Before points, the ratings (Puppet, Mastermind etc.) were done on the basis of raw win count. It was a terrible system because it promoted a “scattergun” approach to diplomacy. Eventually the more nuanced reward system of PPSC was used as a means to keep people from abandoning games without having the small, tight community of pre-0.7
The relative values of different results are difficult to assess, however first of all, for a rating system, the game should be zero-sum. A perfect game should end in a draw, and if that is negative sum, that means that a group of players all playing perfect diplomacy could lose rating. Similarly, if the draw is considered positive-sum, then players could “fix” games to gain rating indefinitely.
Draws see all players equal, according to the rules. That means that an n-way must give a result of 1/n. Now, the only way we could increase the value of a win would be to say that a win was worth more than 1, and then say that a defeat and/or survival when there is a solo is worth worse than nothing. This seems to me to be potentially absurd: if defeat is worse than nothing, people who go out in year 3 are then loosing points because of the other 6 players’ choices after they are out of the game, but if we go for just survival being worse than nothing, then there could be an absurd race to be defeated. You could try a kind of continuum between the two, of course.
Interestingly the idea of having survival + a continuum would really penalise killing off people only to come second.
Note that Rait and TMG have both topped the Ghost-Rating lists.

However, I would warn you, people’s actions are not based necessarily on what benefits them in G-Rating, even if that is their aim. Perception is what matters, and there are many (e.g. you are better off playing stronger players in G-Rating)


To give you an idea of G-Rating and rewards, if, at the point of the last Ghost Rating, MM had played numbers 2-7 in a game, these would have been the changes in ratings he’d have for various results:
Win: +191.669
2-Way: +73.9294
3-Way: +34.68286
4-Way: +15.05958
5-Way: +3.285619
6-Way: -4.56369
7-Way: -10.1703
Defeat: -43.8102
You should note that it is not linear from 0, and that 3 3-ways are not worth a win, indeed, with the reduction in deltas that comes from gaining rating, you gain more from one solo than you do from 6 three ways.

Here is a list of Win Percents, with a min qualification of 40 games, 25% wins.
UserID,Wins,Games,Win Percent
950,61,119,51.3%
6454,51,100,51.0%
3149,19,41,46.3%
15832,25,56,44.6%
9013,18,42,42.9%
877,19,50,38.0%
1824,22,60,36.7%
5945,96,274,35.0%
5265,22,63,34.9%
3922,18,52,34.6%
20789,23,67,34.3%
13159,16,47,34.0%
16965,17,50,34.0%
2803,18,53,34.0%
8737,19,56,33.9%
19520,24,71,33.8%
18322,19,57,33.3%
964,29,88,33.0%
17706,15,46,32.6%
4792,13,40,32.5%
3514,34,105,32.4%
9215,15,47,31.9%
11454,13,41,31.7%
1474,13,42,31.0%
20297,13,42,31.0%
13890,25,81,30.9%
2061,27,88,30.7%
19730,14,47,29.8%
5210,24,81,29.6%
8919,13,44,29.5%
1332,23,78,29.5%
5167,12,41,29.3%
11329,14,48,29.2%
11338,14,48,29.2%
2560,36,124,29.0%
4010,15,52,28.8%
4793,16,56,28.6%
3660,12,42,28.6%
1990,13,46,28.3%
10008,32,115,27.8%
1132,13,47,27.7%
9124,11,40,27.5%
928,24,88,27.3%
2735,25,92,27.2%
222,52,192,27.1%
1705,18,67,26.9%
5965,26,97,26.8%
4102,16,60,26.7%
4029,18,68,26.5%
842,14,53,26.4%
1207,15,57,26.3%
5385,15,57,26.3%
10873,16,61,26.2%
6718,17,65,26.2%
1967,33,127,26.0%
2037,14,54,25.9%
8019,27,105,25.7%
3474,44,173,25.4%
1778,24,96,25.0%
11508,10,40,25.0%
21717,10,40,25.0%

I love the game of diplomacy because of its nuances, which in turn make it difficult to score in this sense.

I would say ranking players is more akin to voting for MVP than a standings system. In that view GR would be similar to slugging average--a good indicator, but not definitive. We can all agree that a win is better than a draw, much like we'd all agree that a homerun is better than a single. But a player is more than his quantifiable attributes.

Also, I'd say that a win as Italy is better than a win as England, much like a win against a top player is better than a win against me.

Essentially, when comparing players I'd like to do like comparing favorite players---using endless data to support an endless argument. Sure Willie Mays is better than Willie Randolph, but Mickey Mantle will require some debate. So when comparing diplomacy players, we can look at all their games and let the arguing ensue

I was hoping you'd reply, Ghost, and you did not disappoint, thank you for your input.

I got a PM asking me to reply- hadn't noticed the thread before that. ;)

@TheGhostmaker's reference to losing points in a draw.

http://www.diplom.org/~diparch/resources/strategy/articles/odd.htm

I'm inclined to feel the same way as Edi Birsan. No well played game will end in a 6 or 7-way draw.

But they should end in 3 or 4 way draws