Sunday, September 19, 2010

Are IRV elections outcomes publicly-verifiable?

OK - county and statewide mandatory county and statewide IRV elections are here. Our own State Board of Elections knows that we don't have software that can count the votes.   We have some idea how the paper ballots will be counted, but not how the DRE votes will be counted.  So whatever schemes they have to count the votes, will the outcomes be verifiable? What are the criteria for deciding?

Check out Kathy Dopp's paper:

11 Fundamentals of Auditing Elections

All of the following procedures are necessary to avoid certifying incorrect election outcomes:
1. Use independently-auditable voting systems — voter-marked paper ballots.
70 of North Carolina's 100 counties use voter-marked paper ballots for election-day voting.  Many use them for Early Voting.  All 100 counties use paper ballots for Absentee By Mail voting.  But unfortunately, almost 40% of NC ballots are cast on DRE touchscreen voting machines.  Although they do have a thermal paper trail, not all voters check the thermal paper trail which can fail to print out up to 9% of the votes.

2. Publicly report all vote tallies (audit units) used to tally overall election results prior to randomly selecting a sample for auditing.
Verified Voting advocates in NC fought very hard to come up with a method to randomly select a sample for auditing.  But since NC law requires the audits before the results of the canvass are certified, and there is no way to tally IRV in a reliable way, this won't happen in the Court of Appeals race. 
3. Allow meaningful public observation and participation in manual audits and in security and transportation procedures for electoral records, including ballots. Prohibit ballot and electoral record access between the time of public posting and manual audits.
How will the IRV races be audited?  Ballots will be moved and the first round will be counted, canvassed, certified and audited before the 2nd and 3rd column votes will be counted.  Will there be audits of the 2nd and 3rd column votes, and if so, when?  Will the same race be audited across the entire state AFTER there is a winner declared?  Or will this race just be an example of "faith-based" voting?  
4. Use sampling methods designed to allow the public to verify that the risk of certifying any incorrect initial election outcome is less than 1%, based on within audit unit upper margin error bounds and the overall reported contest margin. Assume that outcome-altering vote fraud could be well-hidden in the smallest number of audit units possible. If maximum within audit unit margin error is assumed to be less than the upper margin error bound, then allow candidates to select discretionary units for manual auditing in addition to the random sample.
There is no way to do a meaningful hand-to-eye audit of the IRV race without counting the whole thing by hand from Murphy to Manteo. 
5. Use publicly-verifiable random selection methods, preferably weighted by upper margin error bounds of audit units rather than using a uniform sampling distribution.
6. Conduct polling place and jurisdiction-wide reconciliation of printed, used, unused, and spoiled ballots with absentee ballot and polling place records.
We really won't know which ballots are spoiled in the 2nd or 3rd columns by voter confusion, since our voting machines can't count the 2nd or 3rd columns, much less find over-votes.  The 2009 Minneapolis IRV election had a spoiled ballot rate 300% higher than the rate in their 2005 General Election.  And ABM voter won't even have anyone to ask questions of while they cast their votes - and if they make a mistake (which they are 3 times more likely to do) they won't be able to get a new ballot.  
7. Provide timely public access to electoral records necessary to evaluate the accuracy and integrity of reconciliation and manual auditing processes.
We don't provide public access to those records anyway.  We can't tell how many over-votes or under-votes we have even in the 1st column votes unless we run around to each and every precinct and get the totals off the tapes. 
8. Use voter intent as the standard for manual counts during audits.
9. Publicly report any discrepancies found during the audit and use the manual counts to correct the initial reported results.
10. Expand the sample size, perhaps to a full recount, or certify the election using an algorithm based on the premises of the sampling method. Treat any missing paper ballots as discrepancies when deciding whether to expand or certify.
11. Complete the audit prior to certifying election results.
Does the state you live in follow all these procedures?

North Carolina voters were sure we could do this with single column votes, but it remains to be seen if the State Board of Elections will adopt procedures that can be verified or even observed by candidates and their representatives, or even interested members of the public!

No comments:

Post a Comment