|
|
| Author |
Thread Statistics | Show CCP posts - 23 post(s) |
|
CCP Veritas
C C P
C C P Alliance
694
   |
Posted - 2013.02.21 13:37:00 -
[1] - Quote
PalkAn4ik wrote:I was trying to look it up and having no luck. What is the Big O notation you got for that algorithm?
Since it operates on all possible sets of candidates, it grows O(numCandidates choose numSeats)
CCP Veritas - Senior Programmer - EVE Software |
|
|
CCP Veritas
C C P
C C P Alliance
695
|
Posted - 2013.02.21 16:05:00 -
[2] - Quote
Sgurd Battersea wrote:going up to 5 would be better.
People are free to only put in 5 if they wish. Heck, they can only vote for one if that's all the preference they have. The only downside is that they might disenfranchise themselves if noone in their small set of candidates end up having enough support.
CCP Veritas - Senior Programmer - EVE Software |
|
|
CCP Veritas
C C P
C C P Alliance
696
|
Posted - 2013.02.21 16:29:00 -
[3] - Quote
Malcanis wrote:Two step wrote:I wrote a blog post about what this means for wormhole candidates. It is critical to make sure that all candidates ask their supporters to list *all* wormhole candidates at the top of their ballots. Silly Two Step, this change is meant to prevent voting blocs from gaining more influence!
What Two step is coordinating is identically equivalent to having a primary, except it takes less coordination and is done during the election instead of prior. It has the added benefit of spare "wormhole party" support (as in, leftover votes that aren't enough to elect a "wormhole" candidate) potentially transferring to secondary preferences. The "wormhole party" doesn't magically gain more votes because of the procedural difference - if they account for 2/14 of the vote they'll get 2 seats, if they account for 1/14 they'll get 1 seat, ect.
CCP Veritas - Senior Programmer - EVE Software |
|
|
CCP Veritas
C C P
C C P Alliance
697
|
Posted - 2013.02.22 23:47:00 -
[4] - Quote
Poetic Stanziel wrote:Trebor Daehdoow wrote:One system that isn't feasible is Schulze-STV. While it is a very good counting method, its computational complexity explodes as the number of candidates and seats goes up. A back-of-the-envelope calculation indicates that a Schulze-STV election with 14 seats and 28 candidates would take over 9 years to compute on a decent PC. Schulze runs in polynomial time, not exponential time. The Schulze algorithm is a widest path problem. It has a running time of N^3, where N is the number of candidates. It's quite an efficient algorithm for all candidate sizes. You're way off base here, Trebor. I realize though that your programming experience is circa-1981.
Schulze-STV is different than regular Schulze. They're similar in that they're both path-based, but Schulze-STV is based on all possible group outcomes, making the graph size grow combinatorially.
CCP Veritas - Senior Programmer - EVE Software |
|
|
|
| |
|