[EM] breakdown of Oakland mayor ballots

[Kristofer Munsterhjelm]

Bob Richard wrote:
>   On 11/13/2010 8:09 AM, Kristofer Munsterhjelm wrote:
>> Sand W wrote:
>>> Here are the results on an actual election: 
>>> http://www.demochoice.org/dcresults.php?poll=OakMayor&type=table 
>>> <http://www.demochoice.org/dcresults.php?poll=OakMayor&type=table>
>>>
>>> Perata (or maybe someone in his camp) accuses the other candidates of 
>>> "gaming the system" by promoting each other as 2nd choices.
>>>
>>> Some challengers tried to do that to IRV-leader Kriss Worthington 
>>> too, but he won by a landslide.
>>
>> Unfortunately, there isn't enough data here to check who would won 
>> under other methods, except for Plurality (where Don Perata would have 
>> won). To find out the social order for a Condorcet method, one would 
>> need the Condorcet matrix, and for most other methods, the raw ballot 
>> data itself.
> 
> Raw ballot data is here:
> 
>     http://www.acgov.org/rov/rcv/results/OaklandMayor/ballot_image.txt
>     http://www.acgov.org/rov/rcv/results/OaklandMayor/master_lookup.txt
>     http://www.acgov.org/rov/rcv/results/ballot_image_help.pdf
> 
> Here is the Condorcet matrix from these data files (entries are the 
> number preferring the row candidate over the column candidate). If your 
> email program doesn't handle HTML tables very well, this might be scrambled.

Alright, I think this should be the correct social ordering for Schulze 
based on your Condorcet matrix:

1.  Jean Quan (CW)
2.  Don Perata
3.  Rebecca Kaplan
4.  Joe Tuman
5.  Marcie Hodge
6.  Terence Candell
7.  Larry Lionel Young Jr.
8.  Don MacLeay
9.  Greg Harland
10. Arnold Fields
11. Write-in

It seems IRV got it right this time and elected the CW, and that 
Plurality would not have done so. The conclusion, of course, hinges on 
that the old program I used to calculate the ordering doesn't have any 
bugs that would alter the winner.

I'll have to transform the raw data before checking further.