Recommendations for RCV Implementation

As the Bloomington City Council finalizes the Ranked Choice Voting (RCV) ordinance and decides how to conduct RCV elections, here are recommendations based on the experiences of other cities already using RCV. 

Number of Rankings

Research on the implementation of RCV has found that allowing up to six rankings gives voters the opportunity to express their full preference and minimizes ballots that may get exhausted for lack of ranking one of the top two finishers. 

The number of candidates a voter can rank in an election differs among the communities using RCV. Minneapolis and St. Louis Park have thus far only allowed 3 rankings. St. Paul allows up to 6 rankings. In some jurisdictions across the country, there are as many rankings as there are candidates.

We recommend the ordinance allow up to 6 rankings.

The Tabulation Process

Bloomington has the same voting equipment as Minneapolis and St. Louis Park, and can take advantage of the same counting process, which is significantly faster and less expensive than a hand-count process. 

The spreadsheet tabulation process currently used in Minneapolis and St. Louis Park is fully transparent and highly accurate; there have been no errors found in the past three election cycles. In reporting the results, the round by round re-allocations are reported publicly on the city elections website. The process relies on the exported cast vote record from the Hennepin County voting machines.

St. Paul does a hand count of the ballots because the machines they use do not produce an exportable cast vote record. Their process has also been highly transparent and accurate, but no more so than the Minneapolis method. And the hand count process is significantly more costly and delays the reporting of election results by days. 

Therefore, we recommend Bloomington use the same process as the other RCV cities within Hennepin County - the spreadsheet process - until the Universal Tabulator is adopted which would automate the process. (The RCV cities in Hennepin County are currently exploring the possibility of adopting the Universal Tabulator.)  

Write-In Candidates

Bloomington should follow the process it uses now, which is to record votes for write-in candidates, but not require that that work be done before declaring winners. 

The ordinances for all RCV cities in MN - Minneapolis, St. Louis Park, St. Paul and Minnetonka - require write-in candidates to pre-register before election day in order to have votes for them recorded as individuals. Pre-registration gives election officials time to establish the identity and eligibility of the write-in candidates. The rule also helps ensure write-in shenanigans don’t slow down the tabulation process.

If there is a significant write-in campaign, the write-in candidate is likely to be registered. If not, write-in candidates are typically eliminated as a group in the first round of counting because they have not earned adequate support. To be clear, all write-in votes are counted and reported in the official certification process, but they have never been reported as part of the public election results.

Because of the desire for timely election results, we recommend the ordinance to do what the other RCV cities’ ordinances allow which is to require write-in candidates to file a request to have votes for them counted during the tabulation process. This best matches the city’s current process of reporting votes for write-in candidates, i.e., the votes for each write-in candidate are recorded at the end and included in the election results reporting. Administrative rules separate from the ordinance can include this requirement. 

Recounts

Standard practice in the Minnesota RCV ordinances is to provide taxpayer-funded recounts according to winning thresholds in state law for the final round -- not any prior rounds. The rules allow for campaigns to petition for a recount at their own expense at any round, but not at the taxpayers’ expense. We recommend Bloomington follow suit.

Static vs. Dynamic Threshold

Different RCV cities show the final results differently depending on the denominator of votes used. In some of the cities, this is described explicitly in their rules; in others, it is how they report the results as an interpretation of the rules. 

Most jurisdictions in the U.S. that use RCV report results using a dynamic threshold, meaning the winning candidate has a majority of continuing or valid ballots (total votes cast less those that are exhausted for lack of further rankings). This is helpful in reporting final results. In the final round, the winning candidate will always have a majority of votes continuing in that round, but not necessarily of the total votes cast. 

It is clearer for the public to understand the final outcome if continuing ballots instead of total ballots cast are used as the denominator. Therefore, we recommend that Bloomington adjust the denominator.