Amen. In face-to-face games, you don't have any excuse. PBM and suchlike, NMRs can trigger CD and you don't use that person again. Seems this version is attracting a lot of uninteresting or uncommited players. I've been in a few games now where the end game gets rushed into an 'easy win' because someone has decided it has taken too long.

News flash: this is a methodical game and if you aren't willing to devote a month or so, then pack up and go back to flash pong or whatever.

I discourage this. Most NMR's are people who have either quit, or have gone on a school overnight trip or vacation sometime, like me. In the case of quitters, you won't need it because they very likely will not be playing anymore. Also, this will make the innocent overnight trip and vacation guys look very unreliable.

But they are unreliable! Anyone who due to exceptional circumstances misses a few turns in all their games, but otherwise has a good record won't see too much damage to their NMR%, but people who do this sort of thing every weekend or every couple of weeks, disrupting all their games repeatedly, they will quite rightly get a high NMR%. That's pretty much my definition of unreliable!

If you're committing to a daily game then you are committing. Noodlebug hits it on the head.

A sane option would be to allow all the current players to vote on pausing the game and require a unianimous vote to pause. Then the game would be held until there was a unanimous vote to resume.

I agree with the general consensus here. I think we are all guilty of missing a turn or two due to commitments, but I think that if people are frequently doing it, they shouldn't commit to a game.

How about you need more than one in a row or a certain % per game for it to count?

great!

Bumping this thread in response to a thread posted by Mahatma. This might solve the same problem you're referring to, Mahatma, but objectively rather than subjectively.

Jim, I like your suggestion a lot. The only downsides are that it is not as immediate and that it wouldn't capture players who give up without abandoning.

True, this plan would do very little to fix a game that someone had already inconvenienced. Players who gave up without abandoning the game would receive an NMR with every passing round, though, so the plan would give them a reason not to do that.

The "abandon game" button is a good idea, although it, too, should come with some kind of penalty to your reliability rating. Discouragement of people who just walk away from games without even having the decency to click that button, then, could be intensified with a variation on Alex's suggestion: what if the first missed round in a row counted as a single NMR for purposes of calculating NMR%, the second in a row counted as 2 NMR's, the third as 4 NMR's (and being booted from the game) (or some equivalent weighting scheme)? That way, small errors and unreliable play would be slightly discouraged, and very disruptive play would be decisively punished.

Not a bad suggestion, but this problem should be covered with a new scoring system.

NMR's would have the side-effect of encouraging players to abandon accounts, and restart, after a vacation or such, because their NMR may be too high to play with anyone. They can bypass this by creating a new account and restarting with full credibility. Of course, the scoring system, has a little of this problem too... except...

With a scoring system, you can work off your demerits, and I doubt there will be a negative point system, and I imagine it will ultimately be culmulative, so starting a new account wouldn't be more advantageous score-wise than continuing to play on your original account.

As well, NMR's allow some leeway, which is obviously acceptable, but every move missed biases a game, each move missed should be penalized, no matter if it's the move that hits the 10% mark, or if it's the first move missed. With a scoring system, the offending player would feel the damage immediately, and there would be short-term incentive to stay with a game. If they absolutely HAVE to leave, they can do so at the loss of some of their score, but that can be earned back, where as a bad NMR reputation is harder to work off than it is to create a new account.

I think I agree that NMR% isn't the only relevant measure of reliability, and that a points system would do a better job of encouraging people not to start new accounts.

However, I can see a major drawback to a simple additive/subtractive points system, too. If Al has been playing perfectly reliably for a month and built up only 100 points -- and Bob has been playing for a year, built up 1000 points, then lost 800 of them recently due to a streak of negligence -- then I'd much rather play with Al. But Bob would have more points.

Maybe the answer is to keep two statistics about each player. NMR% on the one hand, and number of moves ever submitted on the other. Players who start new games could choose how stringently to discriminate against newbies and (separately) how stringently to discriminate against unreliable players. Therefore, you'd probably wind up playing most of your games against people who are more or less i) as new to phpdip as you and ii) as reliable as you.

That's not a bad outcome. And you'd be discouraged from starting a new account, because it would mean starting over, playing against new users and rarely encountering the people you know, whereas simply playing a game or two on your old account would repair some or all of the damage to your reliability score and return you to grace.

I agree with your point that every missed move should in principle be penalized, but I don't see how a points system handles that any better or any worse than an NMR% system. (Either way, it's possible to incur a penalty that just happens not to drop you below a penalized minimum threshold.)

The more thought I give the problem, the more I think it's important to penalize consecutive NMR's, dropouts, and other unreliable stunts more severely than isolated ones. If a points system is implemented, maybe the number of points you forfeit for an unreliable stunt could depend on how long it's been since your last unreliable stunt.

I was reading a bit about Diplomacy, and I saw that the standard rules are a bit unbalanced, for example, it's quite difficult for Italy to win a game.

So I'd like to know from you: what's the most balanced variable you know?

i'm assuming you mean "variant". and it's probably Fleet Rome. it's the same as the regular game, but Italy starts with a fleet in Rome instead of an army, so they can take Tunis AND Greece. that gives them a slightly stronger position. unfortunately, you can't play the Lepanto then :(

Of course you can still play the Lepanto.
Just a slight variation, same end result however. =D

And yes, I just realized that a points system would not differentiate between an unreliable player and a new player. But the point system works on the individual player, and not on the game creator. The purpose, is to deter players from being unreliable, it is not to allow users to create games and discriminate between other players.

If you have a higher score, you should not be playing someone with a much lower score than you, because either they are more unreliable, they are newer, or they are not as skilled. That is the only discrimination, and while it means that perhaps new players might have to deal with unreliable players, hopefully, the point system and a competitive atmosphere would deter players from being unreliable, and instead focus on playing the game with the intention of raising one's score.

I would not penalize consecutive NMR's more than isolated incidents, because if it's consecutive, it is likely that the player has taken a break due to real life issues, or vacation, in which case, they should not be penalized harshly, because obviously real life comes before phpDiplomacy. The other possibility is that the player has quit phpDiplomacy, in which case, further penalization would have no real effect, and might in fact be a major discouragement to any second thoughts the quitter might have.

A multiplicative effect on NMR's is large enough, there is no need for exponential growth.

The idea I've been going with for the points system is one where you have to place a certain number of your points into a game "pot" as a bet. The number of points you have to bet is chosen by the game creator.

When the game is done, or you are defeated, you get a certain amount of points back, perhaps more or perhaps less depending on your performance. If you go into civil disorder you don't get any points back.

This means that anyone who goes into civil disorder in high-stake games won't be able to join high-stake games for long. The winners can join higher stake games, those who are average float around in the middle, and people who go into civil disorder do the worst.

So when someone joins they have a small number of points and can only play in low-stake games with people who may be bad players or may go into civil disorder.
If you often go into civil disorder you don't progress, if you win against new/bad players you progress to higher stake games, and generally you end up playing people at around your skill level. If you go into civil disorder often you'll only end up warring people who are new, bad players, or go into civil disorder often, if you're an average player you'll end up in games with average players. If you're a good player you'll end up in games that only good players can get into.

It also means that if you have a certain number of points you can either join a whole bunch of smaller games, or one or two high-stake games, so it also provides a way of stopping people playing in more games than they should. If they're playing in too many games they'll probably be playing in low-stake games, so it won't matter if they can't dedicate a lot of time to playing in them.


That's the idea anyway

I like the gambling element very much. (Would you stand to win points if you won with a crappy power like Italy?)

I hadn't understood that when you guys talk about a "points system", you don't just mean points based on behavior, but points based on success in games. I guess I don't like the idea of being stuck with unreliable players simply because I'm a bad or new player.

And the converse of that concern is that if the system leaves it unclear whether a player with high points is a successful player or a reliable one, I don't have a sure way of excluding unreliable players from the games I play.

I guess we've had two different intentions in mind all along: Kestas and Chris are interested in implementing a system that puts evenly-matched players in the same games (which I agree is desireable); my own priority (and that of several other posters, in this thread and in the one started by Mahatma) has been avoiding playing with likely dropouts.

It'll be interesting to see whether the gambling system plays out the way you expect. Seems plausible. Will there also be free games, so that even before I build up many credits, I can play in up to 10 games?

* that should have read, "Would you stand to win more points if you won with a crappy power like Italy?"