PROCEDURE TO MITIGATE ISSUES ARISING FROM LISTENER BARCODE MISPRINT KPFA LSB ELECTION 2009

PROCEDURE TO MITIGATE ISSUES ARISING FROM LISTENER BARCODE MISPRINT
KPFA LSB ELECTION 2009

Due to a print house mistake all listener ballots were coded with listener membership codes and not the anonymous barcode as directed. Here are some of the issues identified so far and proposed solutions. All feed back is welcomed.

ISSUE 1: The identity of the member voting is compromised as his/her membership code is the barcode.

ANALYSIS & PROPOSED SOLUTIONS
Is this a big problem?
- Member ship codes range from 3 to 6 digits and we are talking about a list of 20936 listeners that are valid voters with member ship codes. We have 2895 (plus 65 “problem” ballots) valid ballots. Unless you have access to the MEMSYS department and have almost Superhuman Powers to memorize these - I don’t think that this is a problem. In other words – having membership codes attached to the ballots for the purposes of judging if the ballot is indeed valid (which we will do on Saturday) does not appear to breach anonymity. On the other hand if after ballots are processed (barcodes scanned and ballots scanned) – someone with access to MEMSYS gets a hold of the ballots associated to the membership codes then this may compromise the voters identity – after the fact.

I propose that before we seal the ballots for posterity (after ballot barcodes/membership codes and the ballots themselves are scanned) we remove the barcodes by simply cutting off the corner of the barcode. Barcodes should be kept in the same place that the ballots are kept. This way we can assess later on that the same number of barcodes is associated to the number of ballots processed to make sure that only valid votes were used. The only risk in this case is that one could theoretically trace WHO actually voted but not HOW they voted.
- Here a reference about human memory and capacity to retain strings of digits and matrices. Exceptional Memory. KA Ericsson, WG Chase - American Scientist, 1982 - psy.cmu.edu, cited by 121 primary source journals – see Google Scholar search.








ISSUE 2: Duplicate and Triplicate Barcodes
- There are ballots which were issued the same barcode as they correspond to household membership numbers (approximately 3000 potential duplicates).
- There are reissued ballots that have the same number as the original barcode printed by the mailhouse because there is overlap between the reissued barcodes and the membership codes (10 duplicates due to membership number for 1 individual overlapping with the reissued ballot barcode of another individual, 5 triplicates due to membership number for a household of 2 individuals overlapping with the reissued ballot barcode of another individual)

ANALYSIS & PROPOSED SOLUTIONS
Is this a big problem?
- The list of live barcodes that will be issued to the barcode screener will allow for the acceptance of duplicates and triplicates for these cases ONLY. Additionally there will be an “OR” statement included in order to adjust for individuals who were reissued a ballot (reissued barcode) but ended up using their first ballot (membership number).
- The only issue here is that someone other than the valid voter photocopied one of these duplicate or triplicate ballots and submitted the ballot and their ballot barcode was screened BEFORE the valid ballot barcode. In order to adjust for this - the reject ballot box will be rescreened to identify potential valid ballots at risk of this occurring (all duplicates or triplicates).

ISSUE 3: Volunteers were issued a blank ballot
- There are 165 volunteer that did not get a reissued ballot. Their ballot had no barcode as they have no membership code.
- All ballots that were found with no barcodes were kept separate in the box of “problem” barcodes.

ANALYSIS & PROPOSED SOLUTIONS
Is this a big problem?
- 165 votes lost represents approximately 0.7% of the total number of valid listener voters.
- There are approximately 50 blank return ballots which represents approximately 1.6% of the total number of votes cast (including 65 “problem ballots” – this was 2958).
As an alternative to reissuing ballots to all volunteers – which would not be good practice as the election is over – we can assess the originality of the blank ballot (original printing marks are being identified today) and assume that these represent volunteers that cast their votes.