I am submitting these comments on behalf of RepresentUs. RepresentUs is a nonpartisan, nonprofit organization committed to protecting and strengthening American democracy. We have extensive experience with ranked choice voting (RCV) and have assisted efforts to bring RCV across the United States, from Alaska to Maine. We appreciate the opportunity to submit comments on proposed regulations 1VAC20-100-10 through 1VAC20-100-80 to implement § 24.2-673.1 of the Code of Virginia.

The promulgation of implementing rules is a necessary and welcome development, however the proposed regulations contain significant omissions that would seriously impair or even prevent the successful operation of RCV elections in Virginia. As a general matter, the proposed regulations don’t have any explicit tabulation procedures that instruct election administrators how to run an RCV tabulation. Rules implementing RCV should provide a clear, transparent, step-by-step process that offers administrators (as well as candidates, voters, the media, courts, and any other interested parties) specifics about things like how votes are transferred or when and how a candidate is declared elected, neither of which are directly addressed in the proposed regulations. Specificity about the basic, mechanical parts of the tabulation process is essential for both the practical administration of elections (from the vote count to any subsequent audits and recounts, and even post-election litigation if it comes to it) and the public’s understanding of, and trust in, the electoral process. This is why implementing rules and guidelines for RCV in states like Colorado (Rule 26 of 8 CCR 1505-1), New Mexico (1.10.14 NMAC), and California (Secretary of State’s Guidelines for Instant Runoff Voting in Charter Counties and Charter Cities, available at https://votingsystems.cdn.sos.ca.gov/oversight/directives/irv-guidelines.pdf) contain clear, explicit, step-by-step instructions for conducting an RCV tabulation. We therefore strongly recommend that the Board of Elections adopt or incorporate the proposed language submitted by Chris Hughes of the Ranked Choice Voting Resource Center in comments.

There are several crucial issues that the implementing regulations should address.

First, the proposed regulations don’t establish an explicit and comprehensive process by which candidates are eliminated and votes are transferred (or “reallocated”). Proposed regulation 100-50 establishes that the inclusive Gregory method should be used, however “the inclusive Gregory method” as commonly understood and as defined in the proposed regulations, applies specifically to the reallocation of surplus votes in multi-winner or “single transferable vote” (STV) elections. Reallocation also occurs in STV elections when there are no elected candidates with a surplus, in which case the last-place candidate is eliminated and votes for that candidate must be reallocated to the voter’s next choice. In single-winner or “instant runoff” (IRV) elections, there are no surplus votes and reallocation only occurs after candidate elimination, so the inclusive Gregory method wouldn’t apply at all.

Second, the proposed regulations don’t address what is necessary for a candidate to be elected. For IRV elections, this would be when a candidate receives the majority of votes on active ballots (some jurisdictions wait until only two candidates remain and declare the candidate with more votes the winner, but the outcome would be the same in either case). For STV elections, candidates are elected when they reach a predetermined election threshold. The proposed regulations mention the election threshold in definitions but do not establish how a specific threshold will be determined in an election. We recommend using the Droop method, which is common in places that use STV. Under Droop, the threshold is determined by taking the total votes counted for active candidates in the first round of tabulation, divided by the sum of one plus the number of offices to be filled, then adding one, disregarding any fractions, or ((Total votes cast)/(Seats to be elected+1))+1. For example, in a four seat election where 1,000 votes were cast, the election threshold would be 201 votes (since ((1000/(4+1))+1 = 201). Without providing a specific threshold formula in the regulations, election administrators won’t have any way to determine what the threshold is for a particular election or recognize when a candidate has crossed it.

We are also concerned by the specific method of breaking ties among last-place candidates proposed by the regulations. The standard practice is to eliminate tied candidates one at a time, until the tie is broken and tabulation can continue (typically through a random draw). Eliminating all tied candidates simultaneously, as the proposed regulations do, is an unusual approach that could lead to counter-intuitive outcomes and undermine the purpose of RCV. For example, consider a scenario where three candidates, A, B, and C, are running for a single seat. Candidate A has 100 votes, candidates B and C are tied for last place with 80 votes each. 60 of candidate B’s voters have ranked candidate C next and vice versa. If only B is eliminated then C will have more votes than A and will be elected (or vice versa, if C is eliminated). But if they are both eliminated simultaneously, A will win even though most voters preferred B and C over A. We strongly recommend that the Board follow the standard practice among other jurisdictions that use RCV and adopt a method to eliminate candidates tied for last place one at a time rather than all of them simultaneously.

It’s unclear why 100-30(3)’s prohibition on the use of RCV in elections with fewer than three candidates is limited to primary elections. The same logic (that there is no point in ranking when there are only one or two candidates) would apply in general elections as well.

Finally, the adoption language in 100-30 appears to require local governments to continuously readopt RCV ahead of each election in which they wish to continue to use it (“The adoption for ranked choice voting shall only be valid for the election specified in the ordinance signed by the governing body.”). Requiring local governments to continuously decide how to run elections with as little as 90-days notice can make it unnecessarily difficult for administrators, voters, candidates, and potential candidates to prepare for upcoming elections. Administrators may need to know in advance what voting method to prepare for when devising an annual budget or developing training and educational materials for their staff and the voters. Those considering a run for office may need to know what voting method will be used when making their decision. Requiring a local government to determine the rules of an election ahead of each new election removes the consistency and predictability necessary to make those kinds of decisions. We recommend that a city or county’s adoption of RCV under these regulations be effective for all subsequent elections unless the city or county’s legislature affirmatively votes to stop using RCV. 

The proposed regulations are a positive step for RCV implementation, but they need revision, and we again recommend the Board of Elections adopt or incorporate the proposed language submitted by Chris Hughes of the Ranked Choice Voting Resource Center in his comments. This language has been developed in conjunction with some of the country's foremost experts on RCV and has been developed from the successful use of RCV in other jurisdictions. Thank you very much for your consideration of our comments. We are excited to see Virginia join the growing list of jurisdictions that give voters the freedom to rank candidates. We are available to answer questions and provide the Board of Elections with any additional information necessary.

Sincerely,

David O’Brien

Policy Counsel