UPDATE: One thing this letter does not clarify is that while spoiled ballots might represent a voter education issue, they do not adversely impact the tally. In fact, the more overvotes and duplicate votes the scanners catch, the higher the spoil rate will be. Yet voter intent will be better represented in the end. My intent here is only to emphasize the need for voter education with RCV. Also, overvotes and duplicate votes found on cast ballots will not necessarily all have the effect of denying a reading of the voters' intent. In some cases the voters' valid higher rank choices count toward a winning candidate and the lower marks are never examined.
Hello Portland Charter Commission,
To introduce myself, I am a resident of Massachusetts with some familiarity with election administration and election methods. Currently I serve as Chairman of the Board of Registrars of Voters in Haverhill, Massachusetts. However, this communication represents no official position of the Board. I have studied various election methods and have presented lectures on election methods through Northern Essex Community College, a local institution. In the past I have supported local RCV/IRV efforts. For example, the Massachusetts state HAVA plan contains a clause contributed by me, stating that the state will “develop voting system standards requiring, as part of certification process, that the system demonstrate the ability to support a representative set of possible future ballot procedure changes, including instant runoff voting, as feasible, with an upgrade cost that is substantially less than the cost of complete system replacement”. More recently my familiarity with the experience of some cities has persuaded me that the problems posed by RCV are more serious than I initially believed. In this letter I would like to provide information regarding just a few points.
I recently wrote a blog article addressing the question of voter difficulties using RCV ballots, "Valid ballot – What does it mean for IRV?" . A few excerpts:
IRV is sometimes criticized as being too complicated for voters to understand. To prove that this is not a problem, proponents like to point out that the percentages of invalid ballots for IRV elections are usually low, usually between 0% and 1%. So what does it mean to have an invalid ballot in an IRV election?
Strictly speaking, the answer lies in the rules for each locality. However, it is common for a valid IRV ballot to be judged effectively by one rule: there must be no overvote in the first ranking of each race.
Overall, about 316 Aspen voters (168 + 148) had some kind of trouble with a ballot, 12.4% of them. It would be more useful to look at those statistics, rather than just say that only 23 ballots were invalid. We need to look at a more complete picture when assessing how well IRV is working for the voters.
For brevity I have left out details here, but the entire article may be found at the URL below. The spoiled ballot rate for Aspen was 6.6%, much higher than usual. Still, 5.8% of the ballots came through with overvote and duplicate vote errors, which potentially cause some marks to not be counted. Voter intent is lost, because vote marks after the overvotes and duplicates are ignored. Yet these ballots are still declared "valid".
Aspen hired a consultant to scan and machine tally their May 2009 IRV election, nearly doubling the cost of the election. The election day tally was later found to be in error, but only after the election had been certified . Aspen is quite likely to become the next jurisdiction to drop IRV, this November. There is ongoing litigation against the city to obtain ballot images for an audit. Also the local District Attorney is investigating many issues including legality of the new election rules and poor execution of the election last year. In 2007 Aspen voters voted simply for "IRV", but the details of the actually enacted election rules were more complicated and problematic.
In Minneapolis about 4.1% of the voters spoiled ballots (1888/45968). Again that would be quite high for a regular election, but it seems typical for RCV. In addition, 6.4% of the cast ballots (2958/45968) had some kind of error. I have communicated to FairVote members that this represents a stark contrast to their repeated claims that Minneapolis had only 1 invalid ballot.
Minneapolis programmed their ballot scanners to count all rank levels in the RCV races, whereas Aspen had them counting only the top one or two. If RCV is adopted in Portland I would highly recommend the Minneapolis approach, since it returns all overvoted ballots to the voters in the precincts for correction. Unfortunately no federally certified ballot scanner is able to detect duplicate votes, a problem unique to RCV. San Francisco spent additional funds for special programming to detect duplicates, although I am not familiar with how they handled certification. RCV is not cheap.
Another Minneapolis best practice is hand counting of ballots. This avoids the machine certification issue and increases public trust. Hand tallying RCV takes longer, but consultants should be available to recommend an optimal procedure and maybe assist with election worker training. There is no technology solution for RCV that matches the level of certification and public trust that hand counting affords.
My first recommendation for a ballot question on adoption of RCV is to develop and explain the exact RCV rules and procedures in full, so that Portland voters know before they vote exactly what they would be getting. To summarize my recommendations for Portland, if RCV is adopted:
Plan for voter education ahead of every RCV election.
Program optical scanners to catch as many overvote errors as possible.
Since RCV must be centrally counted, chain of custody for ballot delivery must be top notch and transparent.
Tally by hand, with a "fresh" group of well trained election workers, not the workers who have been at the polls all day.
The charter should provide for runoff elections in the event majority is not achieved.
 Valid ballot - What does it mean for IRV? - IRV Factcheck Factcheck – June 11, 2010
 Error found in IRV vote tabulation – Aspen Times May 29, 2009