So with all of you time advocates, how do you handle a case like Tyran or Halt, who CD all of their games, leave the site, but then come back in 6 months? Should they get a fresh reliability because it has been 6 months so that they can do it again?

6 months? No, I stated at least a year, another suggested 2 years.
If possible, a formula that took this into account would be even better. If you stay active you can dig out, if you drop out of sight for this 1-2 years and come back...nope, you were not active and your rating has not improved

I think it should count less. I'm not talking about a hard cap of counting/not counting, I would hate that, but I think the weight of a CD should reduce with 5% each month for example. After 6 months, the weight of a CD would still be more than 70%, and in theory the CD never stops counting.

Usually everybody says that a CD is a terrible thing. And I agree with that !

But now, some players seem to want some sort of "forgiveness" in their statistics.
Statistics should show reality. Forgiveness should come for the heart.

For example, if I see that the PERMANENT record of CDs of some player is equal to x out of y games, I can talk to that player and if he gives me some decent explanation about his previous CDs, I can possibly choose to invite him in a password game in which I hope that player will not CD nor NMR. I could possibly give ONE chance to such a player and if he proves to be reliable in that game, I could then consider that that player is in fact reliable, although his statistics might be terrible. That's what we could call "forgiveness".

On the contrary, if I know that a statistic can be manipulated by acting such or such a way, by calculation or simply by letting time elapsing, I will just not look at that statistic. That's what I do, for example, with the current RR.


"I only believe in statistics that I doctored myself. "

That sentence was attributed to Winston Churchill, possibly by Nazi propaganda to depict him as a liar.
Anyway, it is very interesting : what can we believe in a statistic ?

I suppose my idea should be more based on the amount of games played properly after a CD, instead of time. That makes more sense. That way there's no quick fix, no leaving the site for a while until the RR is good enough, you can only become more reliable by evidently becoming more reliable.

Of course, that's next to the actual stats IMO. Ratings shouldn't get into the way of actual information, and I say this while it's probably against my favour. I don't care too much about this one one way or another, but I think there should be a difference made between a CD yesterday and one 5 years ago, while not removing any proof of the CD 5 years ago.

I do believe in forgiveness ...over time and after proving yourself once again reliable. But This applies (in my mind) to the rating only and the stats should remain that show you have missed phases, your ranking says you have made up for past misgivings but the record remains permanent. That's MY feeling as to how it should be.

Making up for this is only done by time and a showing that you have continued to not miss orders. Simply dropping out shows time but no correction, correction and time combined together is the key!

Then we agree.

Adding a time-factor (or even more complex: a counter for games finished after the CD) is not really possible at the moment, as I have not really the time to add this. Also this information is not available in the database, so everybody would start with a new clean rating without his history.

And using the simple and easy calculation CD/totalGames time is already a factor.
If you have a bad CDratio you can just improve this by playing games without going in CD (same for NMR). If you have played 8 games here and have 2 CDs your CDratio is 75%, but 5 years later you have played 100 games, and your CDratio is 98%. The other way round: if you managed not to CD in 100 games you have proven that you are a reliable player. 1 CD does not invalidate your history of 100 perfect games.

A more interesting question is: If we show a NMR-ratio and a CD-ratio on each ones profile do we need a gamecreation-setting for each one or do we combine these 2 into one reliability-rating? And if we make a combination how should we value these (CD vs NMR = 1:1 or 2:1)? CD in 2 out of 10 games is 75%, this number allone does not look that bad as it needs to IMHO.

@Tom - if you look back it will be difficult to implement *if* functions based on time as the system has not recorded *when* a CD occurred. Oli has said that it might be possible to program this according to in-game timestamps, but that it would require a lot of work. So while that system of penalising more recent CDs over older CDs is a good idea, it is actually a lot more work than it sounds.

^ninja'd by the man himself :)

Oli - I think to start with a game creation for both stats <at least to start with> would be a good idea. We can observe how players use this function and decide in a few months time whether or not to combine them into 1 reliability rating @ game creation or keep it at 2 settings.

Right now the impact of the CDs don't seem influenced by the amount of total games played, if they are 'not balanced', which should still not make a difference...

That needs to change, other than that it works for me right now.
The thing is that I have 6 CDs because of RL issues, not complaining, and that means 60% off my RR, no matter how much games I play, as long as I don't join open games which is something I don't want to do.

I don't care too much, but I think that's quite a flaw in the calculation.

6 CDs out of 100 games has the same reduction as 6 CDs out of 6 games as far as I can tell, which is unfair.

@ Oli :

You are of course right : players can simply improve their stats by not CDing nor NMRing.

I think that the NMR-ratio and the CD-ratio should be used independently. What's more, how could we combine them seriously ? First of all, a CD is much worse than an NMR, second, let's please think about it : how many Phases does it take to eliminate a CDed country that is not taken over by a new player? Sometimes, it could be MANY Phases. As a consequence, a CD should be penalized at least as much as the number of those virtually Missed Phases.

So yes, please, let's have a permanent record of Missed Phases and of CDs so that a precise NMR-ratio and CD-ratio can be available clearly on a player's profile.

Finally, in order to be even clearer, these ratios should be called "No NMR-ratio" and "No CD-ratio".

--------------------

@ steephie22 :
6 CDs out of 100 games would give you a No CD-ratio of 94% ; 6 CDs out of 6 games would give you a No CD-ratio of 0% !

--------------------

@ everybody :
I would please remind that a CD or an NMR can only be decently explained by a lack of access to the internet. Indeed, all a player has to do to not NMR is simply to open the page of his game : that is only one single mouse-click !

We also have several tools to avoid CDs and NMRs : Extends and Pauses have been so far easily granted automatically by the moderators team and there is also the possibility to find a sitter for your game.

NMRing and CDing should happen very rarely. Accidents in Real Life do happen but when a player manages to miss 5% of his Phases or, much worse, 5% of his games (unless he has played very few games, of course), then that simply shows that such a player is not reliable. Some even say that it is plain rude.

The way I understand it, the RR was designed in order to motivate players to take over CDed nations, which is possibly a good thing. But the RR doesn't show AT ALL a player's Reliablity because any CD can be so-called "balanced". And maybe that is part of the misunderstanding : the letters RR stand for *Reliability* Rating and as a consequence, some players sincerely think that they are reliable because they have, for example, 97 % even though they have had many CDs !

Can I ask if this RR is in regards to only Civil Disorder (Game Dropout) or does it apply to NMR's as well (missed orders)?
NMR's is terrible and affects a game adversely, this should (and maybe IS?) be taken into account as far as a reliability rating is concerned. A CD should be even worse, it means multiple missed orders and simply can not be stood for, recovering from CD should be VERY DIFFICULT to recover from! I'm reading here that we have an example of 6 cd's out of 100, that should be a horrendous "reliability" rating, if however it's 6 missed phases out of 100 games (X how many phases), that's not bad!

@ Tomahaha :

As it is shown here :
http://www.vdiplomacy.com/reliability.php?userID=3272

The exact calculation of the RR is :
RR = 100 − (100 * 2 * NoMoveReceived / TotalPhases) − 10 * (CDs - CD-takeovers)

In other words :
- each % of your ratio Missed Phases / Played Phases diminishes your RR by 2%.
- each of your CDs diminishes your RR by 10%.
- BUT each time you take over a CDed country, you erase statistically one of your CDs : and that is a SERIOUS problem because it makes the RR useless as far as measuring a player's reliability is concerned !

It might make sense up to that take over part! I understand rewarding people for taking over a game. But adding reliability to one who has EARNED an unreliable ranking makes no sense. A electronic medal or badge next to someones name would do wonders while not affecting reliability. This reward makes no sense in the least.

Indeed, Tom ! This whole conversation started with a player (no offense meant to him) who had :

- about 33 Missed Phases out of about 1785 Played Phases : that made him lose about 3.70 % on his RR. (I wrote "about" because we can only see today's numbers of Played and Missed Phases).
- 5 CDs : that made him lose 50% on his RR. Please note that the number of played games was not taken into account which is of course not statistically correct.
-> That player had a RR of about 47% a few days ago.

BUT :

That player has quickly taken over 5 CDed nations in less than 24 hours, and suddenly his RR jumped to 97 % !

Incidentally, I wonder why his 96.30% is rounded up to 97%.

And so, we have a player who is given a *Reliablity* Rating of 97% even though he has 5 CDs out of 79 finished games !

OK, how is this any different, see how silly this really is...

Let's say we have a support group for alcoholics.
We assign a some sort of reward system for going without drinking. Each time someone slips up, they get a mark against them.
But we want to reward those who volunteer to be support people.

We now have people who have demonstrated they have trouble staying sober rewarded for being support people. Why not give anyone that volunteers to help a six pack while at it?

@Tomaha: yes, the current system is not good, that's why I will change this.
This thread is about finding a new system. The old will not be used anymore.

The main purpose of the RR should be to have a tool to limit your games to very committed players (esp. for anon games).
Because of the fear to limit the available games too much and to give some incentive to actively take open spots in ongoing games I added the possibility to ballance these CDs. But as you all realized this is totally against the spirit of the feature.
And because not much people use a RR-setting in their games at all a sub optimal RR does not limit the possibilities of a player that much, but the current system even does not help the committed players too, so no one really gains anything.

Oli -

IMO, the best option would simply be the "non-CD" percentage and "non-NMR" percentage that has been discussed thus far. Personally, I'd say no time limit on it, it shows your entire history, that way, as you play more, your percentage goes back up if you are truly reliable.

At that point, creating two filters in game-creation would allow people to truly select people who are reliable to play in their game.

I would caution about reading too much into how people use the game creation filters though. I would wager that a fair number of the high quality games on here are invite only games, and as such, don't even bother with the filters. So while it might look like the filters aren't being used as much, that might be just because people are looking at stats and only inviting those who would meet their filters anyways.

@ drano019 : I fully agree with you +1


When I create a game, most of the time, I use the password option. Then, I don't bother about using filters because I invite players either because I know that they are reliable, or because I have found that their profile shows that they are reliable.

Of course, I have also sometimes created games without passwords, then the filters become very useful. And anyway, the filters are obviously most useful in anonymous games.

As a conclusion, yes, I think that simply showing most clearly the "non-CD" percentage and "non-NMR" percentage would be most efficient and of course two filters related to these parameters would finish the job.

Some finer questions:

How should we record a CD.
+1 CD / +1 games total
+1 CD / +0 games total

How should we record a country that goes in CD and later the player rejoins the game as no one else took over the country.
+0 CD / +1game played.
+1 CD / +1game played.
+2 CD / +1game played (in the extreme case he does this 2 times in the same game).

Oli,

For your first question, without hesitation, a CD should simply be recorded : +1 CD / +1 games total.
For example, a player has played 9 games with 0 CD : his "No CD ratio" is 9/9=100%. Then he plays another game in which he CDs : his "No CD ratio" becomes 9/10=90%.

For your second question, it may be less clear. My personal opinion is that if the player comes back into the game, then there is no CD, there were "only" NMRs.

I think a CD is a CD whether you come back or not.

I"m with Mapu.

First question: CD should be +1/+1
Second question: If you CD and come back, you still went into Civil Disorder. +1/+1 . If they CD again, +2/+1. If they're that unreliable, they deserve to have it shown.

I am of the opinion if someone returned from CD they would be NMR status for those turns, after all, it's the same thing and the extra NMR's is just as bad or worse to a rating (or should be ...my brain hurts looking at math formulas)
CD happens when you quit or miss so many turns in a row. If they came back that shows at least something doesn't it? Isn't that the idea here, and if they CD again...hit them with TWO CD's in one game! (all while NMR's accrue)

By default CDs happen when you miss 2 turns in a row.
But there are some optional settings that send your country in CD even for one single miss.

bump

In the last 2 years we've been assisting to fast RR recovery due to the take over +10% bonus.
Sometimes, even too fast.

If we're going to lose this bonus entirely I'm afraid that:
1) Open Games position will be much less appealing to find substitutes.
2) RR recovery will be really slow. Many players may find it more convenient to reset their bad condition it via creating a new account.


If we change the way that the RR is calculated, we should re-discuss the boudary conditions too, since the two arguments are bonded together: how many games per RR% we can play simultaneously and other factors like RR filters.

IMO, toss the RR% restrictions on games played. If people want to let shoddy players join their games, let them. People will learn to filter by non-NMR and non-CD rates and those shoddy players will be left in the dust. We shouldnt' be catering to those people anyways. We should be trying to develop a system that rewards people who ARE reliable, not one that caters to people who aren't.

Regarding Open Game positions be taken slower: What if we changed it so that if you fill in a CD country, the game does NOT go on your stats? That alone should encourage more people to try them out for fun. Adding in an electronic medal for those who do it a lot should also be a bonus.

Regarding RR recovery being slow...umm...that's the point. If you suck it up and CD and NMR a lot, your recovery SHOULD be slow. It shouldn't be forgotten in 2 weeks just because now you have more time, but oh wait, a month from now, you'll CD all your games again when you get busy. Again, why should we cater to unreliable people? As for resetting their bad condition via a new account...that's how we get them. Make the limitations on games played be tied to phase count, not RR. Make it so that if people keep resetting, they get punished by not being able to join many games because they're "new" and need to learn.