[EM] Re: San Francisco IRV Ballots - District 9

[Eric Gorr]

Greg Dennis wrote:
> just FYI on the San Francisco data . . .
> 
> a vote for candidate (number-of-candidates + 1) indicates an "overvote," 
> which is a vote for more than one candidate at the same rank. in 
> District 9 there were only 7 candidates, so a vote for candidate "08" 
> indicates an overvote. SF counting rules dictate that you ignore the 
> overvoted rank and all subsequent ranks.
> 
> a vote for candidate (number-of-candidates + 2) is an "undervote," a 
> vote for no candidates at that rank. in District 9, a vote for candidate 
> "09" was an undervote. in the case of an undervote, SF rules dictate 
> that you promote all subsequent ranks. that is a vote for "blank 2 3" 
> should be treated as if the voter selected "2 3 blank".

Interesting.

This promotion of undervotes _DID NOT_ occur for the District 2, 3 and 9 
elections in the official results. If you take the ballots in these 
three elections and count all the first place votes which do not 
correspond to the undervote or overvote code, you get counts that match 
the official results.

This seems as if it is a scandal waiting to happen unless a check was 
performed to verify that following the rules for the undervote ballots 
would not have changed the outcome. If this check was not done, the 
incorrect winner could easily be selected in a close election where the 
undervote ballots would tip the election to someone else.

However, even if the check was performed, it still seems odd that they 
would not follow the rules and report the numbers based on that.

As a side note, I've updated my ballot archives with the SF ballot data. 
It can be found at:

http://www.ericgorr.net/library/tiki-index.php?page=SanFrancisco2004