1. The games you left are now included in the calculation of your reliability.
Each game you "Left" subtracts 10% from your rating. You can recover from this by taking an open spot in a "Open" game.
2. Your Rating is now Displayed as an American-grade (A-F) A=90+, B=80+ C=70+ D=60+ E=50+ F=49-.
Creating new games you can now set a minimum grade instead of a Rating.
3. Your Grade is displayed in game behind your points.

Awesome, I'm an 'A'.

As always everything is up for discussion and is likely to change sooner or later. I still need some experimentation on this topic.

I worry that joining games could cloud the measure of your reliability. I've been in a lot of games where someone CDs and then someone else joins and they CD. So my bias is that the kind of people who jump into Open games are similar to the kinds of people who CD in general.

hehe "American Grades" don't have E's though?!?!?!

No E..? Ouch... What are the typical American grades?
I just browsed wikipedia. We have 1-6 here in Germany... (or 15-0).

@gopher: If you taken an "open" spot and go in CD later you didn't gain anything... +1 and -1 is zero.

typical American would be:
A,B,C,D, or F

F stands for "fasiled" and A-D are passing grades in the american system. Typically, below 60% is a failure, and below 70% is below average with 75% being the true "average" grade.

*failed, nor fasiled

Oli- it is typically
A- 90+
B- 80+
C- 70+
D- 60+
F- >60

I thought Oli was an Italian

Oli is German, I think.

@butter, that would be F - <60. >60 means "greater than 60".

Nice Oli!

*<60

yes, I realized that as soon as I posted it, and was going to correct it, but got distracted by Mass Effect

Oli is German. This server is hosted in Germany. I checked.

Just curious, who has the most phases without missing any?

None of 1184 for me.

But theres probably people out there with more.

@Oli....so imagine that people who join open games also CD a fair amount, as a class of people. I care about the probability of someone CDing. If they CD 20% of the time, but join Open games for 25% of their games, then they would have a super-duper reliably rating, no? Such a person would be precisely the kind of person that I would want to avoid and my ability to spot them is being reduced.

That was my concern. A, B, C, D and F seems like a good idea.

@GOD....when did you join House Martell?

I have 1403

Very good, Oli. This harsh penalty for CDs is what we need and the 'recover-taking-over' rule looks good, too. I'm sure that, from now on, players will take more serously what not enough people say: a CD screws the game and it's a shame.
Also, I believe that the number of Open games will be somewhat reduced because of this.

Since you said everything is up for discussion, I'd like to comment a couple of aspects.

1) I'm afraid that penalty for a NMR stays uneffective.
I calculated that with the 1,432 phases I already played, I need to NMR 74 times before I fall in the "B" group. Even a single NMR can screw a game, so I think I shouldn't be authorized to miss so many phases only because I've been a good child during last 6 months. An upgrade in this part of the formula is still needed, IMHO.

2) Balancing CD with TakeOvers is pretty simple, and this is a good thing. But someone could be tempted to CD a hopeless game thinking he'll just TO the best game among the Open, to balance the loss. I propose that TO should be worth 0.5 CD.
First of all, DO NOT CD. Then, if ever, TO a couple of games to recover.

Yeah...'2' was basically what gopher said.

It's so weird. I was in the middle of posting about how little sense the current reliability rating makes and BAM! It's changed.

Still not completely convinced it's the right way to go. I'll just post what I had wanted write:

In my view there are 4 variables useful for calculating how reliable a player is, they are: ('#' stands for 'number of')
# phases missed (missing phases makes you less reliable)
# phases played (playing phases, i.e. not missing them, makes you more reliable)
# games left (leaving games makes you a very unreliable player)
# games finished (likewise, finish more games and you are more reliable)

So what would be a good measure? When deciding if you should play with someone or not you might want to know: "What is the chance this player will not miss any phases when we play?"
We can get a good average if we calculate that chance with:

chance of NOT missing a phase = 1 - (# phases missed / # games finished)

This would already give a decent reliability rating. If you have missed any phases, you can calculate your own now.
Examples: you've played 60 games and missed just 1 phase because sometimes your an idiot. Calculate chance = 1 - ( 1 / 60 ) = .983 so a rating of 98%.
But what if you're really unrealiable with 3 missed phases in 4 games? Chance = 1 - ( 3 / 4 ) = 0.25 so a rating of 25%. We can call it asshole rating too.
N.B. this rating can be a negative number too, for the persons that manage to miss more than one phase per game on average...

You could ask in the same way: "What is the chance this player will leave our game when we play?"

chance of NOT leaving = 1 - (# games left / # games finished)

It is possible to have two separate reliablity ratings on people's profiles. You can even show what chance they represent: not missing a phase and not leaving a game respectively.
Or you can combine these two chances by muliplying them. This is not completely fair, because every game that someone leaves will also 'cost' them two missed phases. So the # missed phases gets counted twice. And in the # games finished, the # games played is included, which skews the result some more.

A more balanced formula would be:
Reliability rating = ( 1 - ( # phases missed - 2 * # games left) / ( # games finished - # games left ) ) * ( 1 - # games left / # games finished )

Hope the notation is clear, i can upload a picture of it too.

Another example rating. If you missed 5 phases, left 1 game and finished 80 games the rating would be:
Rating = ( 1 - ( 5 - 2 * 1 ) / ( 80 - 1 ) ) * ( 1 - 1 / 80) = (1 - 3 / 79 ) * ( 1 - 1 / 80 ) = 0.95 = 95%

If you had missed 5 phases, left 1 game and only finished 10 games it would be:
Rating = ( 1 - ( 5 - 2 * 1 ) / ( 10 - 1 ) ) * ( 1 - 1 / 10) = (1 - 3 / 9 ) * ( 1 - 1 / 10 ) = 0.6 = 60%