ratsy wrote:

M57 wrote:

Xrayjay wrote:

Remember we (the users who care) are essentially an impotent Senate that talks a lot, living under a (thankfully benevolent) dictator who can pick and choose what part of the talking he listens to. Did Tom even weigh in once on the very lengthy discussion to get around to Option I? I'd go back through and wade amongst the posts looking, but there are just too many.

Option I did not enjoy full site consensus, not that anything could when there were a dozen+ ideas floated, but it was highly debated and in pseudo-voting was a favorite if not on the short list of many.  Full disclosure, when I come up with the idea, I was fully capable of being relentless in lobbying for it.

That said, CPs and Rankings (General, Colonel, etc.) are the top layer of the stats machine, and perhaps the more important part of that machine that needs an overhaul is its underpinnings.. the Global point system.  These things can exist separately .. I.e., you can pretty much couple any CP system with any GR system with a little tweaking here and there, so technically you don't have to do change/implement either or both at the same time, but in that the GR is fundamental and the CP system is much more arbitrary in nature, it makes sense to know that a long term GR plan is in place.  So for instance, I'd be curious to know if Tom is amenable to crossing over to a Trueskill-like system?  If so, we should wait for him to implement that, and then tweak the new CP system on top.  If not, then by all means.. we need to make noise about putting a more egalitarian CP system in place.

If I recall - that moot ended with Tom saying: he'd be happy to implement some kind of a trueskill like system, so long as someone was able to re-create it - and someone volunteered - sort of, but I bet it fell off the radar. 

Then we also had the idea to give out rankings over 5 categories - to reflect different strengths - and someone just needed to make the badges for that - which I considered, but got busy with RL. 

Yep - I recall similarly.

Championship points.

tom wrote:

With TrueSkill I'm not clear how it works with a game like WarGear where there is only one winner and the rest are losers. It seems more geared to a game outcome where the game results in a ranked of 1-n for an n-player game: http://research.microsoft.com/en-us/projects/trueskill/details.aspx

I don't know much about Gaussian approximations, but your link does say "The player with the highest performance is the winner; the player with the second highest performance is the first runner up, and so on. If two players' performances are very close together, then the TrueSkill ranking system considers the outcome between these two players a draw."

So perhaps everyone who loses could just be considered to tie for last.

M57 wrote:

Xrayjay wrote:

Remember we (the users who care) are essentially an impotent Senate that talks a lot, living under a (thankfully benevolent) dictator who can pick and choose what part of the talking he listens to. Did Tom even weigh in once on the very lengthy discussion to get around to Option I? I'd go back through and wade amongst the posts looking, but there are just too many.

Option I did not enjoy full site consensus, not that anything could when there were a dozen+ ideas floated, but it was highly debated and in pseudo-voting was a favorite if not on the short list of many.  Full disclosure, when I come up with the idea, I was fully capable of being relentless in lobbying for it.

That said, CPs and Rankings (General, Colonel, etc.) are the top layer of the stats machine, and perhaps the more important part of that machine that needs an overhaul is its underpinnings.. the Global point system.  These things can exist separately .. I.e., you can pretty much couple any CP system with any GR system with a little tweaking here and there, so technically you don't have to do change/implement either or both at the same time, but in that the GR is fundamental and the CP system is much more arbitrary in nature, it makes sense to know that a long term GR plan is in place.  So for instance, I'd be curious to know if Tom is amenable to crossing over to a Trueskill-like system?  If so, we should wait for him to implement that, and then tweak the new CP system on top.  If not, then by all means.. we need to make noise about putting a more egalitarian CP system in place.

Fair point. I believe about a year ago Tom did volunteer to run 1 or more options in parallel side by side to see what it would look like.

True skill (op K)  was at the top of the voting, (but technically difficult/possibly unworkable which is probably why it hasn't been done).

Option I came in 2nd and is probably the simplest one to try (which was generally the big argument in its favor the whole time).

 The conversation at that point got hijacked and later somehow moved to the suggestion box area where it died.

 

Here's the original list.  

Option A:

+1 CP to every player over 1500 for every player over 1500 (current threshold).  

It would make everyone over the mark "in the points" and would make 1st place on wgwf/Civil/etc. worth 120­+ CP's for first place.

 

Option B 

+ X% to each player over 1500 for each player over 1500.

 

Option C

Basic scale expansion extending the current threshold to 2500 or 3000 and pay out the top 20-30. (3000 would put the points above ALL current scores)

 

 

Option D

Regressive system such that the uber-popular boards are not too influential.

E.g., give the top 16 boards extra CPs linearly:

WGWF top player = 100

Colossal Crusade top player = 95

..90, 85, 80,..   ..30, 25, 20, 20, 20,..

 

Option E

Logarithmic scale (example):

# of plays	Top Rank CP	Min GR for top rank	# of ranks that earn CP
0-10	10	1250	10
10-100	20	1500	20
100-1000	30	1750	30
1000-10000	40	2000	40
10000-100000	50	2250	50
100000-1000000	60	2500	60

 

Option F

A system for factoring in difficulty, such as:

CP x Difficulty factor x Popularity factor

 

Option G

The current CP system with a log modifier to weight it.  A compromise between exponential growth & linear growth.  So once x games are played or players played, it'd be x2 or x3, or x4, etc. Most boards keep the 20/1500/10 that we currently have. Popular boards get 40 CP.  Very popular boards get 60 CP, ultra popular (Colossal, wgwf, etc.) gets 80CP,

 

Option H

Same as Option G but with a LOW log modifier to weight it.  1.5x, 2x, 2.5x

 

Option I

Simply calculate CPs by taking a player’s board rating and subtracting 1000.

If you have a 1750 rating on a board - that’s 750 CPs

negative ratings are thrown out - So no score is generated from a board ratings below 1000.

 

Option J

Expansion of Option I with Bonus CPs awarded to the top X players where each position gets an additional straight percentage of their current Rating for that board. For instance,

#1 = 100%, #2 = 95%, #3 = 90%, #4 = 85%, etc..

 

Option K

A Trueskill ‘like’ ranking algorithm that replaces both current GR & CP systems. (Possibly the most accurate, but complicated for non-math people to understand).

 

Search Option or CP if oyu want to see the hundreds of posts debating the topic. 

I have been notified by a user via  private message that a new user may be an NWO reincarnate. Is there any way an admin can check to see whether or not this is the case?

PM tom or email him at [email protected]