I am very interested as well, but I don't see any formula or description of how things are calculated.

A lot of it actually sounds like an MMORPG. You have a rating shield that only works when you charge it doing X. If you level up, you gain a 5% chance for your unit to win a one-on-one battle (no longer a bounce).

It could all very well be just an integrated GR system/formula. There may be some improvements, but it's nearly impossible to determine the validity/reliability since we do not know the calculations (unlike ghost's system where it was based on a mathematical system that has been tweaked).

I imagine its quite similar to the GR system. An interesting part of it is that your rating is more volatile for your first few games. Last time we had a debate about rating systems, that was something that was brought up.

The 'ratings shield' seems really hokey and stupid but, meh, their choice.

They explicitly said that they're not releasing the algorithm so that people can't game the system. Because fuck a sense of community and involvement, right?

If it's a good rating system, then it would be near impossible to "game the system." Most people wouldn't even care about the formula and algorithm because it is a lot of math, and secondly, even if you understand how it works, it doesn't mean that you can/will move up the rank. It's not like once you understand how it works, then you just win games. And really, that's the only way to gain rank anyway: to win games.

I don't particularly trust "blackbox" systems where we are just to suppose to trust that it works without knowing how it works.

Once Kestas installs the new features, I will push to see if we can integrate GR into our system, and we are not deleting everyone's points. :)

What is PlayDip?

If there worried about gaming the system then clearly its not a good algorithm right?

Like we all know how GR works and I don't think anyone can 'game' the system...

@nigee.

Cannot tell if serious so I'll bite. It's a much larger for profit diplomacy site. It's been around for only half a dozen years or so. The guts of the system was based on kestas's code so there was a little bit of tension there which, as far as I know, has been resolved.
I just have one game going on there with RL friends because they prefer the interface. I've been meaning to try get another game in there at some point just to branch out and 'see' some new faces.

GR is gameable by choosing your opponents wisely. But those same opponents won't chose you necessarily, so it takes their willingness to be patsies.

in what way? The only way to seriously bump your GR is by playing people who have higher GRs then you and lose.

Is playdiplomacy based off of Kestas's code? I know the way his silence is set up requires that anyone who uses his code give him access to any additional features/improvements should he ask for them.

@Fasces, I'd suggest vdip contacts theghostmaker or anyone else who has run the ghostratings and ask for their assistance making a similar system, the main difference in determining how to weigh each of your variants.

I am going to disagree with you here Draug. While what you said isn't false, it isn't realistic. To consistently gain rating, you have to either win/draw with those whose own rating are fairly high (which is obvious and why many of us choose to only play highly ranked GR games and such) or to consistently win games against lower ranked players. That's not really gaming the system. That's just winning.
Any valid rating system would look at those win/draw vs highly ranked and win vs lowly ranked and reward the player. So, that's just saying the system will reward you for winning.

As a high GR player, you look for really low GR players who have played plenty of games and just suck, then beat them in WTA.

But the point is it shouldn't reward someone for winning outside their bracket. You gonna count a Big Ten team beating some community college in basketball as worthy of praise and granting them a March Madness seat?

GR should look at the average of your six opponents. If you have a sufficient GR above that average, then you risk yours but have no chance to make more than you contribute.

That's easier said than done.

If you can consistently win in those WTA games, then you must be pretty damn good. If you gain significant rating from those, that means you must win a lot more than you lose because a win there has less of a gain than a win/draw in a highly ranked GR game.

Moreover, this is one of those practical issues and not a theoretical one. Until there is someone who can game the GR system legitimately and not cheat, it's not really an issue. I haven't seen a person who has/can do this without being a very good diplomacy player him/herself.

Agreed it is more theoretical. But arguably, Mad Marx could play nothing but newb games and increase his GR. Each game would just be such a small relative increase compared to a single win in a top 200 game.

"You gonna count a Big Ten team beating some community college in <football>?"

But consider the risk ... What would happen if Appalachian State were ever to beat Michigan?

Oh.

Sorry, I don't know college sprts worth shit. So Big Ten is football, eh? OK.

Like I said, you should not get a boost for *winning* but still take a hit for the loss. Please read *everything* I wrote about that.

For an official implementation of the GR here there should be a solution to different player-sized variants. It would be stupid if it could rate only classic games.

Ghostmaker's GR and subsequently the GR that Alderian and I looked at do work with all variants and with different player sizes. It's a general formula, not specific to the classic map.

"But arguably, Mad Marx could play nothing but newb games and increase his GR."
I would say he could win a large percent of them, but he will certainly lose a few, due to random chance and what not. Since it is against bad players he will suffer more.

When playing against bad people, you win a lot, and you lose every once in while. The relatively small increases from the wins balances out the big penalties from the losses.

When playing against better people, you lose a lot, and you win every once in while. The relatively small decreases from the losses balances out the big rewards from the wins.

@jmo, yes playdip is based off of webdip's code. But the license at the time did not require reciprocity. Presumably the playdip situation is what led to him switching to the license that requires reciprocity.

@everyone else, if a highly rated player plays a bunch of lowly rated players, he or she would have to win every game to gain rating over time. A single draw would cost what you gained in several solos. And even more so in a PPSC environment.

FZ, as I've said every time we end up discussing GR, one of the bigges flaws, IMHO, is the te GR system (unlike Elo) treats all losses as equal. I lose the same amount of GR whether I lose to 6 MMs are 6 nooby McNeverplayeds.

Your point still stands about him only getting relatively marginal gains each time though.

It's far easier to lose to 6 Nooby McNeverplayeds than it is to lose to 6 players in the top 100. Far easier.

"A lot of it actually sounds like an MMORPG. You have a rating shield that only works when you charge it doing X. If you level up, you gain a 5% chance for your unit to win a one-on-one battle (no longer a bounce)."

Gah!

That sounds terrible. A bounce should always be a bounce.

nobody wins alot, the best players on this site have maybe 30% solo rate.

There is no LIVE matches at PlayDip ..... !!!!

I don't know... Right now, if I lose a game, my GR drops 42 . If I solo against all noobs (100 GR each), I gain 34 , and if I 3-way draw all noobs I lose 17 . So, a solo and two three-way draws and I break even. Two solos and a loss, I gain 26 .

Things changes relatively quickly even if you are rated as highly as #50 GR, as you only lose 15.5 in a loss, so against all noobs you'd be creeping your GR up even with one win and two losses, while a three-way against all noobs would net you about a point...

I'm sorry nobody is good enough here to go 67% win rate even against all noobs.

Gen, that's a gunboat mentality. The only games I really play against noobs are live games, and I've won every live game I've played in several months. Admittedly, that's only about four, and I think some of them were PPSC, but if I can do that, there are players far better than me who could maintain a 67% win rate against bad players.