My number for instance is ~103.5

I finally found out where you got your name thucy. Th ancient greek writer.

That doesn't account for gunboats, though...

I agree. Better diplomats communicates frequently.

People who play a lot of gunboats don't have a high rating here. Clearly a bogus method.

*(s)

CLEARLY BULLSH*T! Obviously you didn't account for some small factor. This method is not 100% accurate. It does not perfectly measure a person's skill and therefore must be useless.

>Obviously you didn't account for some small factor.

At least 80% of my games are gunboats. I sincerely hope you don't qualify a factor of five difference as a "small factor".

Well, I average 10 messages / game. I must really suck! Communication, who needs it! :-P

"Presumably, the higher, the better diplomat!"
"People who play a lot of gunboats don't have a high rating here."

Doesn't refute, because Gunboats aren't Diplomacy.

Maybe not, but gunboats count as games and thus if you message 100 times per 'real' diplomacy game, yet you are like smiley or myself who play around 80% of the time gunboats, your message rate is very low.

Either way, usually the ones who talk the most are not the best diplomats but rather the trash talkers and sore losers who annoy everyne by spamming their inboxes with insults and endless rants

I am 203 messages per game. But I have only every played one gunboat.

typo... I have only *ever played one gunboat

Thucy is just giving a rule of thumb, he isn't proposing a new ranking system. Yes people play gunboat and yes this would affect your ThucyRank. However, it doesn't matter if this makes the number come out incorrectly because he was just noting a trick he came up with for judging how communicative players are and suggest that this would be a means of judging their diplomatic skills under the assumption that a more communicative player is a better diplomat. Just because it isn't sufficient for determining skill, doesn't mean it isn't helpful.

You can use the number to see if someone is ignoring you or just doesn't go online very often...

Thucy: does the number of messages in your profile actually include ongoing games? I was under the impression that the number only reflected completed games. Hold on, I will do a quick check...

Well, my number is probably around 15 and I am online at least three to four times hours day on the site, maybe more. And I am just saying that as a rule of thumb it is a very bad one. It does not even come close to the actual number of messages per game. If stats were separate for gunboat and regular games it would make sense, but right now it is just an erroneous formula which only produces useless information.

completed games only

I sent a new message in an active game, and my message count hasn't changed five minutes later. So either it updates more infrequently, or we should only be looking at completed games in the equation.

as I said it is completed games only. Otherwise it could be used as a tool to see if a rival is talking more to other players than you.

I put out a report that dealt with messages/game (not counting gunboats) and win/draw percentage as a ratio. You could always caluclate someone's msg's/game and win/draw pct and compare it to the linear or logarithmic fit to see if they fall above (indicating that they're stronger than the mean) or below the best fit line (indicating that they're weaker than the mean).

The equation for the linear fit (for those who want to know) is:
y = 0.3102x + 14.295 (r = 0.71)

The equation for the logarithmic fit is:
y = 18.6ln(x) - 37.9 (r = 0.75)

All you have to do is calculate their msg/game and input it for the x-value. This will give the predicted win/draw percentage and if their win/draw percent is higher than the calculated one, then they're stronger than the mean, whereas if their win/draw pct is lower than the calculated one, then they are weaker than the mean. The r-values indicate the strength of the fit (the logarithmic fit is a stronger predictor than the linear fit, but the linear is much easier to calculate).

So, with my msg/game of around 110, I should have a ~48% win&draw %age - and it's 51% - within 1 game. I'm on the curve :)

Cool deal. Is that with the linear or logarithmic fit?

Well, using the logaritmit fic and my average of 108 messages per game (but I played many gunboats) I should have a 49,18%, having 70% should put me a bit over the mean :-)

But in fact this statistic fit is pretty uselless, if you have a 300 msg/game rate you should have a 107% win/draw rate....

webDip sabermetrics! I love it!

Msg/game (non-gunboat) are low, 43, but a win/draw of 39%. Logarithmic fit is 32.24, so it looks like I'm above the mean (?).

Could you consider this variable to be a proxy for an overvalued/undervalued player? Similar to a measure of WAR (wins above replacement) in baseball, it measures a player's worth compared to a replacement, average player. Granted, I'm not sure how strong an indicator msg/game really is for player quality, but you could probably calculate a similar statistic for webDip players relative to a theoretical "average" player.

Interestingly, the faster you get eliminated in a game, the less opportunity you have to press the other powers, so your msg/game rate might be lower. Somebody who is still around in 1918 should have sent more messages than someone who was knocked out in 1906. It's kind of self-reinforcing... maybe if you were a better diplomat, you would stick around longer and have a higher msg/game rate.

BTW, I like the term ThucyRank. Does Thucydides get to trademark it, or Zarathustra?

@rdrivera: You're wrong about the 300 msg/game. Statistically you cant extrapolate beyond the range of the data because the best-fit lines might not hold true for those valuse. The statistics were calculated for ranges from [7, 234]. Thus, if you have fewer than 7 msg/game or more than 234 msg/game, then you really shouldn't use the formulas I've posted.

@2ndWhiteLine, yes, according to that, you're above the mean. In terms of overvalued/undervalued players, this is merely a predictor of where one should be given the typically expected outcome. There is a measure of expected error which I haven't calculated.