(Duplicated from another forum thread)
Several issues need to be addressed, as I see it.
1. A fair clan ranking system which does not penalise smaller clans or rank sheer quantity of wins.
2. Fair guidelines for challenge match-ups (within which collusion is either eliminated or no longer advantageous for those clans which practise it).
3. A fair individual rating system (within which sandbagging is either eliminated or no longer advantageous for those players who practise it).
I doubt whether any simple solution will meet all these issues; a solution would have to be thoroughly well-thought out, openly discussed, probably tested for a year and possibly tweaked later on, in order to meet with wide-spread user acceptance.
Item 1. I favor ranking clans based on one or the other of the following criteria:
1a. win ratio (not bulk number of wins), OR
1b. net average rating change for a whole clan, This requires careful explanation to make people understand and accept it, but would in principle be easy to code; the point being that net rating change creates a level playing field for clans with many high-rated players compared with clans which have lower-rated players: net rating change encourages players to play for a win because what counts is winning games, not whether the ratings of the players are high or low.
(Substantially similar to Robbie's point 1 above.)
Item 2. I favor limiting the number of challenges between any two clans involving the same players which count towards the clan standings (that is, multiple challenges could be played with the same players, but only 2 or 3 would be counted in the standings). (Substantially similar to Robbie's point 2 above.)
Item 2a. I favor a maximum ratings differential between clan challenge opponents. A max., of 200 has been suggested. (Substantially similar to Robbie's point 3 above.)
Item 2b. I favor a minimum number of challenges or completed challenge games per season; we can hardly dole out a medal to a 2-man clan which plays one challenge and wins both games; sure they'll have a 100% score, but see shortcircuit's point below regarding a proposed bonus system.
Item 3. I favor keeping separate individual ratings for clan games, tournament games, and regular (non-clan non-tournament games). The point being that if individual's ratings within a clan are linked to the whole clan's ranking (see item 1b. above), then sandbagging drops not only the individual's rating, but also that of his clan as well (automatically)--thereby rendering sandbagging no longer advantageous in any attempt to manipulate clan standings.
Item 3a. Same as Robbie's point 5 above -- provisional ratings to be used initially, until the reformed clan system is up and running smoothly.
Item 3b. I favor tracking individual's ratings continuously. That is, clan games should show the individual's ratings at the time each challenge was issued (frozen in time), and should should show their current ratings as well -- this will help clan captains to spot sandbaggers and refuse challenges from opponents with suspicious rating swings. Tracking individuals' ratings might also help to shorten the period of provisional ratings.
Separating individuals' clan ratings from tournament ratings, and tracking individuals' ratings, would, I believe, meet shortcircuit's concern regarding players who enter tournaments and then resign games en mass; his point is valid. It seems to me that the most efficient way to enforce it is to encode this in the individuals' ratings and clan rankings themselves, track ratings, and observe a maximum ratings differential. I don't see how stopping players from resigning games is to be enforced; refusal to accept resignations before move 30 (or any other number) won't work because players have been known to make fatal blunders in the opening.
I agree with shortcircuit that a minimum clan size makes sense. Exact numbers yet to be discussed. How many clans are there, in fact, smaller than 5? How active are they? A maximum clan size might also be considered.
"There has to be some merit in playing 20 vs 20 challenge compared to a 2 vs 2 challenge." It is certainly more work for the clan captain to arrange a 20:20 match, but I'm not convinced it produces better chess matches. There is a precedent for using bonus systems as tie-breakers, so there may be merit to using a bonus system if, at the end of a season, several clans have identical win ratios or identical net ratings change scores (or whatever else it comes to as a ranking system).
The reform of the system should, in so far as possible, incorporate changes which automatically render such practises as sandbagging, collusion, and spurious challenges irrelevant, without human intervention to rectify infractions retroactively, whereas banning people and annulling points are both labor intensive and open to dispute.