Last time, in anticipation of Seattle’s November 2022 ballot choice between ranked choice voting and approval voting, I summarized what we know about approval voting from empirical research (not as much as we might wish). This time, I check the claim that approval voting will be ready for showtime before ranked choice voting.
Proponents of approval voting claim it is much faster to implement than ranked choice voting. For example, Seattle Approves, which sponsored the approval voting (AV) measure on the November ballot in the city, writes: “In Seattle, Approval Voting could be adopted in November 2022 and potentially could be used in the 2023 primary election. RCV [ranked choice voting] would have a difficult and long road to adoption because it would require changes to the law and voting software.”
Is this true? Not really.
In Seattle, any new voting system adopted in November 2022, whether AV or RCV, will almost certainly not be ready for the city’s August 2023 primary election nine months later.
The real question is whether it will be ready within two years and nine months, for the August 2025 primary election, or within four years and nine months, for the August 2027 primary. The balance of evidence suggests that either system would launch in August 2025.
AV: Simpler but not faster
AV proponents are not wrong: AV is simpler to administer than RCV, which should make it faster to launch. It requires smaller changes to ballots than RCV. Julie Wise, elections director for King County, which encompasses Seattle, told Sightline, “Approval voting would be pretty familiar to our voters—the instructions would change but the overall ballot wouldn’t need to change.” Likewise, AV would require much smaller changes to ballot counting software than RCV.
But that simplicity would almost certainly not advance the launch date to August 2023 from August 2025, which is the deadline written into the ballot measure for AV. In other words, it’d be faster, but not fast enough to matter.
Under state and local law and policy, all new and upgraded voting systems (including ballot designs combined with vote-counting hardware and software) require 1) certification in one of a handful of federally accredited testing laboratories, 2) testing by the state, 3) authorization by the secretary of state in Olympia, and 4) a final set of tests and audits at the local level. Completing these steps in nine months would be next to impossible.
King County uses ballot counting software from a company called Clear Ballot, and Clear Ballot does not have software systems ready and certified for RCV or AV. Seattle Approves leader Troy Davis told Sightline that King County is already authorized to use its existing Clear Ballot systems to tabulate approval votes, but Elections Director Wise disagrees. She says it’s unclear what certification and authorization AV would need.
One key line of state law, for example, says that any change in any voting system that “extend(s) its function” requires reapproval by the secretary of state. A wholesale change to approval voting sure seems like it “extends the function” of Clear Ballot’s system. Clear Ballot would therefore probably need to run the full gauntlet of certification and authorization before tallying AV votes. That process alone might take a year, as noted below. The final word in how to interpret state law, of course, is neither Julie Wise’s, nor Troy Davis’s, nor mine. It would have to come from the secretary of state or the courts, but even getting an answer would take time, slowing implementation.
Asked for a preliminary interpretation, Les Bowen, a Certification & Training Program Specialist in the Office of the Secretary of State told Sightline, “any change in currently approved voting systems or use of new voting systems that are not currently approved for use would require certification and approval from the Secretary of State’s office.”
King County is meticulous (and unhurried) in its election administration
Even without a full certification gauntlet, though, AV is unlikely to be ready by August 2023. Julie Wise and the staff of King County Elections treat ballot design and voter education with admirable seriousness. Seattle has twice as many voters as St. Louis and six times as many as Fargo, other places that have implemented AV. King County Elections is a national leader in election administration, and it tests, tests, and tests again its innovations before launching them: user testing, focus groups, real-world simulations, and more. Asked when she thought she could launch AV or RCV, Julie Wise demurred, citing the many unknowns. Then she told Sightline:
What we won’t do is rush through implementation because of political pressures or campaign promises. With the state of our democracy and trust in elections right now, it’s critical to get it right from the start. As election administrators, we generally want to be conservative in our timelines. Our mission is to conduct accurate, secure, and accessible elections—no matter what the voting method is—and we will always strive to match the same high level of execution that we’re known for.
In addition, the extra cost for King County Elections to run the gauntlet is not currently budgeted anywhere, according to Wise. The city of Seattle would have to negotiate with her about costs, and those negotiations could slow the preparations for AV.
For all these reasons, if adopted, AV would likely arrive in August 2025. If AV has to run the full approval gauntlet, implementation in under nine months would be impossible. If good fortune in the form of an agreeable secretary of state were to spare it the full gauntlet, nine-month implementation would still be unlikely, because Julie Wise won’t rush.
Why could St. Louis and Fargo move faster with AV?
AV supporters counter that St. Louis implemented approval voting just four months after adopting it, and Fargo did it in seven months. Why would Seattle take longer?
Policies and practices vary by state, and neither Missouri nor North Dakota’s rules are as strict as Washington’s. Missouri does not require testing in a federally accredited lab. Missouri’s rules are ambiguous, moreover, about when a voting system change triggers re-certification. The St. Louis Elections Board did not respond to Sightline’s inquiries on the subject, but given its speedy implementation, it must not have recertified its systems. In North Dakota, meanwhile, state law allows the secretary of state to approve substantive changes to voting systems without recertification or testing, and DeAnn Buckhouse, election coordinator for greater Fargo’s Cass County, confirmed that she did not recertify anything before launching AV.
August 2025 is therefore the likely arrival date for approval voting in Seattle, if voters choose it in November.
RCV would share a similar timeline for implementation
The RCV ballot question in Seattle sets a deadline for implementation of August 2027, two years after the AV deadline and four years, nine months after the November 2022 vote. This 2027 date is not the result of a careful feasibility study. Rather, it was a last-minute addition in a hurried legislative process. Understandably, officials erred on the side of caution, not knowing all that might be required.
Implementing RCV in Seattle would require new ballot design, either new voting software from Clear Ballot or a change of software providers, and either a full gauntlet of certification and approvals or a friendly intervention by the secretary of state. It would take more than AV, in other words. Still, even assuming the most time-consuming scenarios, implementation by August 2025 seems extremely likely.
Give or take a few months, two years would probably be plenty of time to prepare Seattle for RCV, and two years, nine months seems ample. All of the needed steps have now been completed dozens of times elsewhere in the United States during the hundreds of RCV elections already conducted. RCV administration is now mundane.
Clear Ballot can catch up with the other major election software firms, which already have RCV modules to their systems, and write new RCV software in 6–12 months, according to election experts. (Meanwhile, free, open-source software can do the job already if Clear Ballot proves slow off the mark.) The longest testing and certification process in Clear Ballot’s history only took one year and four months. Ballot design and voter education can take place on a schedule overlapping with other to-dos. Adding up the months, it’s hard not to conclude that implementation will fit comfortably into the two years and nine months between November 2022 and August 2025.
RCV implementation elsewhere took two years or less
RCV’s implementation track record elsewhere confirms this reasoning. All but one of the 34 US jurisdictions that implemented RCV in the last decade did it in two years or less, from legal adoption to first use (see table).1 Twenty-five of them did it in one year or less. The city of Santa Fe, New Mexico, and the entire state of Maine implemented RCV in 100 days or less.
RCV jurisdictions lie in states with the full range of election laws and ballot counting systems, some with stricter certification standards than Washington’s (including all of the 21 Utah RCV cities that implemented in less than a year and those in Massachusetts and Minnesota that did so in two years or less) and some with laxer rules (including Alaska, California, Maine, and New York). No relationship stands out between the stringency of a state’s certification standards and the time it took to implement RCV. Whatever the rules, almost every one of them got the job done in two years or less, which suggests Seattle will be able to implement RCV in plenty of time for its August 2025 primary election.
All but one of the 34 jurisdictions that adopted RCV in the last decade has implemented it in 2 years or less.
Jurisdiction | Implementation time, adoption to first use |
---|---|
Santa Fe, NM | 3 months |
The state of Maine | 3 months |
St. Louis, MO (approval voting) | 4 months |
Cottonwood Heights, UT | 6 months |
Elk Ridge, UT | 6 months |
Magna Township, UT | 6 months |
Midvale, UT | 6 months |
Millcreek, UT | 6 months |
Moab, UT | 6 months |
River Heights, UT | 6 months |
Salt Lake City, UT | 6 months |
Sandy, UT | 6 months |
South Salt Lake, UT | 6 months |
Bluffdale, UT | 7 months |
Genola, UT | 7 months |
Newton, UT | 7 months |
Nibley, UT | 7 months |
Woodland Hills, UT | 7 months |
Fargo, ND (approval voting) | 7 months |
Heber, UT | 8 months |
Springville, UT | 9 months |
Draper, UT | 10 months |
Payson, UT | 11 months |
Vineyard, UT | 11 months |
Lehi, UT | 11 months |
Bloomington, MN | 12 months |
Minnetonka, MN | 12 months |
St. Louis Park, MN | 18 months |
Las Cruces, NM | 19 months |
New York, NY | 20 months |
The state of Alaska | 22 months |
Palm Desert, CA | 23 months |
Albany, CA | 24 months |
Easthampton, MA | 24 months |
Eureka, CA | 24 months |
Benton County, OR | 48 months |
2025 likely for either—and both dependent on software and the Secretary of State
Might AV or RCV come sooner, by 2023? Conceivably, I suppose, but it’s highly unlikely. One possible wildcard affecting implementation speed is, who will be Washington’s secretary of state starting in 2023? The secretary of state has great influence over the interpretation of state voting laws: can AV skip some of the certification gauntlet? Can RCV?
The secretary also has authority to issue waivers from certain testing requirements in cases of emergency. Would a successful ballot proposition in the state’s largest city constitute an emergency, though? That seems farfetched. Incumbent Secretary of State Steve Hobbs is lukewarm on RCV and therefore unlikely to rock the boat. He’s not shown his cards about AV, but his record in public life suggests he will be opposed to novel, little-tested election proposals. Challenger Julie Anderson strongly supports RCV, but she is a veteran election administrator and like her peers, she tends toward meticulousness, not hurry. Neither seems likely to speed implementation.
Might the RCV implementation timeline slip? Possibly. Clear Ballot could bungle its RCV code, and King County Elections could refuse to switch to an RCV-ready alternative, such as RCTab. (After all, elections administrators have other priorities at present, such as security for their own staff members.) Still, Clear Ballot holds the election tabulation contracts for two other large jurisdictions where voters will consider adopting RCV in November: Portland and surrounding Multnomah County, Oregon. Clear Ballot is intensely motivated not to botch RCV. The company’s business is heavily concentrated in the Northwest. King County is its top client, and Multnomah is its second or third largest, according to Carolyn Cote, the company’s marketing manager.
Might AV implementation slip? Yes. Courts could rule against AV, as I described previously, delaying its use substantially or even permanently. This prospect seems entirely possible but is a subject for another day. (9/19/2022 UPDATE: Sightline commissioned a legal analysis, which found it possible but not likely that a state court might stop AV before implementation. It found no legal risks to RCV.)
Still, the balance of evidence weighs heavily in favor of a 2025 launch date for AV or RCV, if Seattle voters opt for either.
Thanks to Erica C. Barnett and Anna Louise Campbell for research and reporting.
Clay Shentrup
St Louis adopted approval voting in November 2020 and used it in their March 2021 primary FOUR MONTHS LATER. There is no reason for it to take longer in Seattle.
There ARE reasons that the bottoms up ranked voting process would take longer.
Clay Shentrup
> Under state and local law and policy, all new and upgraded voting systems (including ballot designs combined with vote-counting hardware and software)
approval voting does not use a new ballot design.
Ben Chapman
Great article — thanks for digging into this! I’m glad you noted how broadly RCV has already been implemented. I think it’s so important that RCV has a whole infrastructure of people ready to support smooth implementation and education. I don’t think the same thing exists for AV.
Felix Sargent
The same infrastructure also exists for Approval Voting. It’s supported by The Center for Election Science, the North Dakota Secretary of State, the Missouri Secretary of State, and was endorsed by all 29 county clerks in Utah. https://utahapproves.org/ranked-choice-voting-in-retrospect-and-the-future-of-voting-in-utah/
Ranked Choice Voting is a chore for clerks, while Approval legitimately solves problems that clerks are seriously worrying about these days – election integrity and audit trails.
Eric Bidstrup
Felix, your claim “RCV is a chore for clerks…” is misleading at best.
When discussing infrastructure readiness, interesting you cite North Dakota when their election software (from Election Systems & Software) still doesn’t support reporting the count of “approvals” vs. voters, as the Center for Election Science acknowledges directly: https://electionscience.org/commentary-analysis/fargos-second-approval-voting-election-runs-smoothly/
Clear Vote election software used in King County has the same limitation currently.
The point of this article is that changes to election infrastucture are needed for either AV or RCV. Claims that RCV (used in over 50 places in the US successfully) is “a chore”, when 1 of the 2 cities using AV currently still doesn’t supporting reporting after 2 elections is misleading.
clay shentrup
This doesn’t affect the election result. It’s just a vote count versus a percentage.
Logan Bowers
Hi Alan,
Unfortunately, this article is not factually accurate. The existing voting software _as written and certified_ already supports Approval Voting, so there is no “extension of functionality” needed.
Conversely, are you seriously arguing that ClearVote can write all new software AND get it certified in just two years? Those timelines are not realistic for either software development or government bureaucracy.
It’s interesting that you had to make pessimistic and factually inaccurate assumptions about AV and then make maximally optimistic (and unrealistic) assumptions about RCV to reach the conclusion they both take the same amount of time to implement.
Eric Bidstrup
Logan, your claim on AV supported by ClearVote is not factually accurate, per Hillary Lincoln, VP from Clear Ballot (vendor of ClearVote software). Or can you show the EAC certification report that supports your claim?
Alan Durning
Hi, Logan. Thanks for reading and chiming in. (For others’ information, Logan is one of the leaders of Seattle Approves.)
I know that you and your team interpret the law and facts as indicating King County could implement AV without any testing or certification. I quoted your colleague Troy to that effect. And then we asked King County Elections and the Washington Office of the Secretary of State if they agreed with your interpretation. They did not.
With regard to the rest of your comment, the article answers for itself. Key points: All but one of 34 RCV implementations in the US have taken two years or less. Clear Ballot’s longest certification process ever was 1 year, 4 months, and RCTab is available free as a backup if Clear Ballot is slow.
I didn’t say “they both take the same amount of time to implement.” I said AV will be faster but not fast enough to be done in nine months.
Eric Bidstrup
Nice write up that offers facts & data to support the points made, including differences between Seattle vs. Fargo and St. Louis.
The quotes from Julie Wise, whose office ultimately is responsible for rolling out and operationalizing any changes to voting in King County, are telling. With all of the misinformation and suspicion on elections in our country today, rushing any changes to elections systems is both unnecessary and potentially risky.
Beyond what is included here, there is also the fact that the current federal testing standards for election systems (Voluntary Voting System Guidelines 1.1) does not offer any requirements that AV could be tested against. Upcoming changes in VVSG 2.0 do offer AV requirements, but 2.0 isn’t expected to be operational via the Election Assistance Commission until 2024.
John Whitmer
A fine article on an issue about which there appears to be some confusion – surely not by intention; no one wishes to muddy the waters on this issue.
Judging by some comments posted here there are differences of opinion (of fact even). On one side we have Alan Durning and his team at Sightline Institute, an organization, with a long well-documented history of interest in reform efforts and of carefully researched, thoroughly reported articles supporting their positions.
The backgrounds of the folks making comments questioning this article are not as well-known to me nor do they seem especially well-versed in the history of various voting systems or of local election protocols. And the arguments offered to support their positions are not – at least don’t seem to me – as transparent as might be desired. Perhaps if I knew a bit more about their backgrounds or was a bit sharper in following their arguments I’d be more open to their positions. But until then I’ll trust the accuracy of Sightline.