double no

no in triplicate

Inside of what??!?

I think that CDs should go away over time. There has to be a way to get your RR back up, without playing a thousand games. I can only use myself as an example, and I did CD in some live games when I first joined the site. No question. My last CD occurred in Dec 2013. But I may not be allowed to join a tourney that I'm deserved to play in because of infractions 20 months ago. No matter what anyone thinks of me, if you play with me, you will agree that I am very involved. I think a change needs to happen.

What do others think? Is this something the powers that be would ever consider?

Given my record is ruined by CD's from between 48 and 66 months ago, I'm pretty strongly in favour of this, too. It was brought up earlier today, and will be discussed, but no guarantees can be made.

How about after a month of 100% no NMRs, no CDs, an entire CD is knocked off? We need to make sure that there is some kind of accountability still

I think after a year from last CD as long as there have been no more infractions, CDs should go away.

god no a CD ruins a game not playing for a month shouldn't erase one this is why we let the feature request thread die smh

I wonder how hard it would be to implement a system that calculates based on the last year's worth of games only.

I completely agree, they ought to expire over time. It reflects too that people's commitment and real life can change.

Yes they should eventual go away in my opinion. Maybe it should be weighted some how but I'm not sure of how the coding works or how complicated it would be. but something like all moves from the past 3 months account for 50%, moves from 3-6 months ago account for 25%, moves from 6 months to 1 year account for 13%, moves from 1-2 years account for 7%, moves from 2-5 years count for 5%

Obviously the time frames and percentage could be tweaked somehow but this is just a rough example.

I'm in favour of having some kind of a decay in RR. It seems like it's the best way of allowing players to reform their record (it should be possible for players to improve - or worsen - their RR, based on recent performance).

If we go this route, it's very important that players not be allowed to improve their RR by not playing.

My gut feeling is that something like "all games in the last 12 months, or the last 30 games, whichever is more games" would be a good rule for inclusion. There are other ways to do it too (and there are disagreements about exactly how to do this each time it's brought up).

My feeling is that the current system:

1) Isn't easily understandable (what does 90% RR *mean*?)
2) Doesn't react quickly enough to changes in player reliability

Well HR and I are two extremely active players here, and I'm sure there are others that have this same issue. We need a fix. I also agree that RR should not change based on non-playing. Something needs to change

#fuckghug

@ssorenn, I actually agree with you; I was just being obnoxious.

I don't think anything should ever be completely forgiven though as much as drowned out. I'm in favor of some sort of time based scaling like Caballo suggests.

I was being obnoxious too. :))

The base certainly has some problems... like not playing for 3 months results in what? Also, I think that for new(ish) players all games should be taken into account equally. Maybe like the first 20 games all games count equally then if a player has more that 20 games use some type of weighted scale. Also, to make it easier the scale could simple be based on number of games not when they occurred. So last X games count 50% and so on. This might make it both easier to implement and also make it so your RR doesn't change unless you actually play more games.

Maybe a good place to start is to see what the average CD rate is in those new games requiring a RR to get into. It seems arbitrary to deem that a 98% or 95% RR is any better than say a 75% RR. Maybe in practice 98=95=75. Seems like an easy analysis to determine the CD rate among those games with RR rating and do a simple regression to see where the RR cutoff really starts to effect game play. Perhaps a mere 50% RR cutoff effectively eliminates those who are most apt to CD and ruin games. We just don't know.

Caballo's system, while in nature seems like what we're going for, allows you to play far less games recently but still raise your RR too much. I propose games that are more recent are just weighted more heavily.

In terms of RR, the present matters much more than the past. A user should be able to quickly increase their RR by playing a few games well. If they CD, though, their RR should dramatically drop. A weighting function with exponential decay should handle this well.

In terms of what the user says, a percentage is not good. What does 90% mean? It certainly doesn't mean they will CD in 10% of their games. All it means is that they are to some unknown degree more reliable than anyone with a lower RR. A percentage that doesn't actually measure a percentage is odd.

I would propose one of the following:

1) A simple labeling system (ie Very Unreliable, Unreliable, Reliable, Very Reliable) that corresponds to whatever back-end algorithm you choose. This has the advantage of being very clear in meaning. There is the potential downside of locking players in low RR hell (seriously, will anyone play with someone labeled Very Unreliable?) but if we have a decent exponential decaying algorithm, they should only be trapped for 1 or 2 games and they will quickly learn to CD again.

2) Using Percentile instead of Percentages. No one really cares how reliable someone is, they really care how reliable they are *compared to other players*. Percentiles is a natural solution to this. Someone is in the 80th Percentile? OK, that's probably pretty good. 20 Percentile? Better watch out. The only issue with this (besides people needing to understand percentiles) is that if everyone is super reliable (or unreliable) the percentiles might get a bit out of whack.

so am i!
@ssorenn, fun hashtag
#fuckghug
#Snowden2016
#FreeTheSlaves1860

@ATC

Any decaying function should absolutely be based solely on number of games, not by time.

I don't know where the percentage comes from.

On Vdip, and we use the same formula Vdip uses, it's just Reliability = RNN, where NN is the number. No idea where the percentage comes from. It is not a (straight) percentage. It is the average percentage you don't NMR/CD to the third power.

Exactly, which is why it's odd to call it a percentage.

Exponentially Weighted Moving Average is the term I was looking for.

Which is odd but also a non-issue. It's not actually a percentage. Whatever. That's semantics. As a Diplomacy player I could care less about semantics.

WFW

quite sure the semantics could care less about you too

Just as a quick example:

RR = 0.5 (G1+(1-alpha)G2+(1-alpha)^2(G3)+(1-alpha)^3(G4)+...)/(1+(1-alpha)+(1-alpha)^2+(1-alpha)^3)+...)

If we set alpha to 0.5 (because why not) and assume a game Gn is either completed in full (1) or CD (0) we get the following:

G1 0 0
G2 1 0.5
G3 1 0.75
G4 1 0.875
G5 1 0.938
G6 1 0.969
G7 1 0.984
G8 1 0.922
G9 1 0.966
G10 1 0.988
G11 1 0.999
G12 1 1.0
G13 1 1.0
G14 0 0.5
G15 1 0.75
G16 1 0.875
G17 1 0.937
G18 1 0.969

So here we see that while you are quickly forgiven for a CD, you are harshly punished as well. This may be a bit extreme, but it is easy to calibrate with alpha. There could be a second weighting for NMR.

How about just a simple second factor called RR100 which is the RR for the last 100 games played? Or RR250?

If you can play 250 games with 100% RR, then people can use that for the tourneys instead of the standard RR (which is basically RR-Infinity).