Berkeley had an issue. A constituent saw a problem at Gilman Street and Peralta Avenue and petitioned the City Council to create a four-way stop at the location. This was referred by the council to the city staff and then languished. The Berkeley city manager’s office was overwhelmed with requests from the council and constituents and requested that council members prioritize their concerns.
The Berkeley City Council is made of eight districts and the mayor. Each, except the mayor, has concerns within their district, along with the city-wide concerns. Berkeley was faced with the question of how to prioritize the work for the city government while being fair to each district.
A council member reached out to the Center for Election Science requesting their recommendations. The method needed to be easy to use for the council members, to put out a ranked priority for each referral and to be fair so that a majority of districts did not dominate the ranking system over a minority of districts. To make a voting system fair, you need to make it proportional. Proportional representation gets deep into the math quickly, but the easiest way of looking at it is when a district gets its concerns met, it should take a step back and give favor to another district so it can get its concerns met.
With these considerations in mind, Berkeley implemented Re-weighted Range Voting (RRV). Council members vote 0–5 stars on each item. The item with the highest score gets first place. But something interesting happens for the remaining items — this is where weighting comes in. District voting power is boosted in proportion to how little support they gave to previous referrals. A City Council member who gave the first referral zero stars has twice the voting power as a council member who gave it five stars. For intermediate scores, the effect lies between those limits. The table is re-calculated without the first place item, then the item with the highest score is selected for second place — and so on. But rather than getting bogged down in the math, let’s look at the results.
Above is the outcome of the Berkeley 2019 referral prioritization process using Re-Weighted Range Voting. As I live in Oakland, I can’t contribute any comments from the peanut gallery on the correctness of the rankings personally. Let’s see if Re-weighted Range Voting did what we said it would do.
How did weighting change the rankings? We can take the weighted rankings, and then see the difference from the unweighted rankings to see what changed.
Berkeley RRV 2019 — Movement due to Weighting
Wow! The item to install a four-way stop at Gilman Street and Peralta Avenue moved up 14 ranks due to weighting. Why? It had widespread but minor support from the council members, but the weighting helped increase the value of District 1 (Rashi Keserwani) which only gave 27 stars to the whole ballot, but a whopping 11% of their vote to that issue. Re-weighted Range Voting enables and encourages council members to support items specific to their district without fear of being lost in the shuffle.
Let’s look at a different scenario. During a recent council meeting, allegations were raised that Re-weighted Range Voting was vulnerable to a “bullet voting” strategy. District 8 (Lori Droste) chose to spend 100% of their vote share on the “Missing Middle Housing Report.” The referral moved from 6th place to 2nd place due to re-weighting. The referral had widespread support from the council, or it would not be ranked 6th when unweighted, but District 8’s demonstration of concern allowed it to move incrementally higher. The referral’s high popularity meant that District 8 could have given 58 other referrals five-star ratings without changing the ranking of the housing report. Re-weighted Range Voting encourages voters to contribute their honest feelings on issues. There is little advantage to bullet voting on an item unless that represents your true interests — and sometimes that happens!
One of the other interesting things I found in the analysis of this data set was that there’s a large difference over how much each district will vote. Both the Mayor and District 3 had opinions on most items. District 3 (Ben Bartlett) gave ratings on 94.12% of items and gave only three items zeros stars!
It’s important to note that voting for every item does not mean that the district was voting unwisely. Voting for all items but one gives the same relative impact to the weighting as voting for no items except one. Council members benefit from voting with nuance.
This system also allows us to dig deeper into the concerns of individual council members. Let’s pick on the Mayor this time. Here are the Mayor’s prioritized concerns.
Berkeley 2019 Referrals to City Manager, sorted by Mayor’s preferences
Pretty neat, right?
There was a discussion at the most recent council meeting on whether rating should be done by department or not. It’s important for the city to weigh all issues against each other or a coalition could get first place for each individual department before the weighting effect can take hold. Let’s see the Berkeley RRV process from the City Staff’s perspective.
Berkeley 2019 Referrals to City Manager by Department
Re-weighted range voting succeeds at its goal of creating a fair and inclusive way for council members to prioritize referrals to the city staff. While the reweighting formula may feel complicated, it’s the results that matter, and RRV delivers.
This article first appeared on Medium.
Felix Sargent is the chair of the board of the Center for Election Science, a nonpartisan, nonprofit dedicated to creating a better world through studying and advancing smarter voting methods.