Uhh...

Jump to navigation Jump to search

What are you doing? I'm legit curious. --Revolverman (talk) 06:53, 9 July 2012 (UTC)

Revolverman (talk)06:53, 9 July 2012

something for his magic graphs i assume

il'Dictator Mikal06:55, 9 July 2012
 

Those are ballot files formatted for OpenSTV.

-- Nx / talk06:56, 9 July 2012

It works great. But I used the wrong type of election at first (alternate in stead of STV). So I redid it, with a surprising result. Now I've to look into the algorithm itself, I'm afraid...

larronsicut fur in nocte07:16, 9 July 2012
 
Edited by author.
Last edit: 07:06, 9 July 2012

Being lazy... Nx gave this link for calculating the results of the STV: https://instantstv.appspot.com/

You have to put the data into it via an url like

http://rationalwiki.org/w/index.php?title=User:LArron/mod01&action=raw

The first line is the number of seats and the number of candidates. I used this to get a ranking of the winners of the mod election:

  1. AD
  2. Genghis Khant
  3. Dick Turpis
  4. Armondikov
  5. Waiting for Godot
  6. Blue
  7. Weaseloid
  8. Ace McWicked
  9. Kels

So, Ace and Kels should be alternates...

Used the wrong voting system - STV turns to be quite interesting - there is a little inconsistency, most interesting from a theoretical point of view

larronsicut fur in nocte06:58, 9 July 2012

ahh, thanks for the enlightment!

Revolverman (talk)06:59, 9 July 2012
 

Hm, that's an interesting way of doing it, but I don't know how valid it is. It's once again doing something with STV that it's not designed to do.

-- Nx / talk07:01, 9 July 2012

So there have been several attempts at that this - all unhelpful.

AceThe Rep Grows Bigger07:04, 9 July 2012

I'm curious - I'm afraid that doesn't mean that I'm necessarily helpful... :-;

larronsicut fur in nocte07:08, 9 July 2012

You are helpful in many ways.

AceThe Rep Grows Bigger07:10, 9 July 2012
 

When you our ballots for a council of five seats, you get as the winners:

{Ace McWicked, AD, Armondikov, Blue, and Genghis Khant}

If you do so for a council of six seats, winners are:

{Ace McWicked, AD, Blue, Genghis Khant, WaitingforGodot, Weaseloid}

So Armondikov is in the smaller body, but not in the bigger (here you find Weaeloid and WaitingforGodot). At least this realization of the STV lacks the expected monotony when enlarging the number of winners: you would expect that every one who is voted into a council of n seats gets a place in a council of n+1 seats.

This monotony is necessary to determine a rank of the winners, which helps to determine moderators and alternatives.

larronsicut fur in nocte07:15, 9 July 2012

With more seats, the threshold becomes lower, and as a result more votes are transferred, which allows Weaseloid and WfG to win over Armondikov.

-- Nx / talk07:21, 9 July 2012
 
 

What is certain is that if you want seven mods, you need to count the ballots for seven seats. That's what STV can do.

Selecting the alternates by looking at who the last two eliminated candidates were is problematic, because of the random elimination in case of a tie (in December, there were many tied candidates, and the winning alternates depend on the random elimination). There was some discussion about this after the first by-elections, but it went nowhere.

-- Nx / talk07:10, 9 July 2012

I agree that a council of seven (C7 = {Ace McWicked, AD, Armondikov, Blue, Genghis Khant, WaitingforGodot, Weaseloid.) should be elected this way - the question for me was: could you take a council of nine, and then take those of the council of nine (C9 = {Ace McWicked, AD, Armondikov, Blue, DickTurpis, Genghis Khant, Reckless Noise Symphony, WaitingforGodot, Weaseloid}) which aren't in the council of seven as alternates: Alternates = C9 / C7 = {Reckless Noise, Dick Turpis}.

While this would work in our situation, it doesn't work in general with this realization of the STV. I don't know whether this is a quirk of the STV in general or just of the implementation.

Fact is: The STV is brilliant for elections in general, but such small elections as ours may not work as well...

larronsicut fur in nocte07:24, 9 July 2012

It wouldn't be fair to do it that way, because in C7 DickTurpis's and RNS's votes had to go to other, higher ranked candidates, in C9, the higher ranked candidates can afford to transfer their unneeded votes (which would otherwise be wasted, as they are sure to win without them) to lower ranked ones, like DickTurpis and RNS. If you determine the alternates using C9 / C7, you are in essence giving more weight to certain ballots.

-- Nx / talk07:28, 9 July 2012
  1. Giving unequal weight to ballots is something happening in any voting system. The question is how to keep the inequalities as small as possible
  2. For us, STV has certain disadvantages:
    1. the number of our ballots is small in comparison with the number of seats and of candidates, and therefore the random eliminations may have a deciding influence.
    2. The result isn't so clear that the cumbersome election process would be justifiable: it takes quite some time to fill in the ballot

I'd prefer for coming elections to use a simpler system - perhaps let each voter name three candidates. This would have resulted in {AD, Blue, Ace McWicked, Dick Turpis, Ghengis Khant, Armondikov, Weaseloid} as moderators and {Nutty Roux, Waiting for Godot} as alternates.

larronsicut fur in nocte07:55, 9 July 2012
 

BTW, I run the election for two places after deleting all the seven winners from the ballots (User:LARrron/alt02 ). The result for the alternates are Dick Turpis and Reckless Noise Symphony...

larronsicut fur in nocte08:54, 9 July 2012

Again, the problem is that Dick and RNS get votes from ballots where they were placed behind one of the seven winners. By deleting the seven winners, they get these votes, but in the original calculation, these votes go to the winners. That's why it's not fair, because those votes are counted twice, essentially.

-- Nx / talk09:01, 9 July 2012