New York City's big experiment with ranked voting ended in a bit of disappointment on Tuesday when a newly released census showed Eric Adams – the election day leader in the Democratic primary – held on to a narrow win.
The new system did not drive an underdog candidate past the end of the first round. But it was almost like that: In the first-choice vote, Adams led by 9.5 percentage points ahead of Maya Wiley. But when the redistribution rounds took off, Wiley was gone, and Adams led the remaining candidate, Kathryn Garcia, by just one percentage point. Garcia had finished third in the first round but almost won in the end.
Voting on the rankings was the dream system of many progressive electoral reformers, and the New York mayoral contest was the biggest spotlight in the United States to date. There were many questions about how voters would adjust to this new system during a pandemic and whether it could keep its promises.
Ultimately, the results are mixed. It wasn't the total disaster some feared, but whether the advantages of ranked selection justify its disadvantages is certainly questionable.
For example, the question arises whether some voters were confused by the new system. Fifteen percent of the ballots in the mayoral contest did not end up with either Adams or Garcia, so they didn't count in the final count. Whether it was because these voters preferred other candidates or because they did not understand the system is difficult to say, but in any case the result was so close that their ballot papers could have made a difference.
Questions also remain as to whether the ranking selection was better than a traditional runoff between the two best winners. Would a runoff have been helpful and useful for New York City voters in making a simplified choice between two candidates, or would it have been a waste of time and money?
After all, the New York census was slow and botched. Most of the time, the ranking voting is not to blame – the main culprits for the counting problems this time around were different state and city guidelines (as well as a simple mistake) – but it is true that, in practice, ranking voting often tends to be slower than ordinary polls made by are informally "proclaimed" to the media.
Did the voters understand the system?
An important question in connection with this election was whether voters were adequately informed about how ranked voting works. For people who have chosen the same way their entire lives, this can be a confusing change. (Rather than voting just one candidate per election, New York City voters could move up to five in order of preference. During the count, underperforming candidates are eliminated and the ballots for them are assigned to the remaining candidate that the voter rated Next.)
One number that analysts tend to look at here is the number of "ballots exhausted". These are ballot papers that do not play a role in the final round because all of the candidates they listed have been eliminated.
This can happen for a number of reasons. One of these is confusion or a lack of understanding of how the system works. Another is running out of rankings (New York City had 13 mayoral candidates on the ballot, but voters were only able to occupy five spots). This could also be the voters' personal choice – even with the ranking option, some people prefer to list just one candidate and profess indifference to everyone else. However, a high number of dead ballots is often seen by critics as a problem for the system.
Overall, about 15 percent of the ballots in the Democratic mayoral campaign in the final round were depleted, meaning those voters didn't count Adams or Garcia. Another way to see the bottom line is that 43 percent of the voters ultimately voted for Adams, 42 percent ultimately for Garcia, and 15 percent for neither.
Exhausted ballot papers may have had more momentous consequences in the final round of elimination when Maya Wiley was eliminated. Nearly 74,000 of their voters' ballots were exhausted from rating neither Adams nor Garcia.
The remaining Wiley voters fell heavily on Garcia over Adams: Garcia received about 129,000 votes from them while Adams won about 49,000. That was almost enough for Garcia to overtake Adams, but not quite – she missed about 8,400 votes. So if fewer Wiley voters had exhausted their ballots, it would be entirely plausible that Garcia could have overtaken Adams.
It is also plausible that a significant part of Wiley's progressive base was genuinely indifferent to the choice between Garcia and Adams, both centrists, and as such fully intended to leave them both unrated. (Some of them, as the results show, also preferred Adams over Garcia.)
However, the strategy of ranking selection can be complicated, and the messages associated with it are often conflicting. Proponents argue that it gives people the freedom to vote for their "true" preferences, but in a crowded field with limited ranking spots available, this could be one way of running out of your ballot papers. The best strategy to prevent this from happening was to make sure you got at least three of the Adams-Garcia-Wiley-Yang quartet that led the polls, but how many voters were aware of that?
If the result was ultimately determined by a lack of understanding on the part of the voter for the system, that would not be ideal – although Garcia was of course only in the race because of the ranking selection, as she was in third place in the first lap.
Would a runoff have made sense for the voters?
The difficulty in comparing the result of a ranked election with how things would have gone under a different system is that it is not clear that the results were really that similar. In a more traditional election campaign strategies would have been different (Garcia and Yang, for example, likely would not have campaigned together), smaller candidates could be eliminated before election day, and voters could have cast their votes in the first round more strategically.
In the end, Eric Adams had 30.8 percent of the first choice vote. Under the previous system, he would have needed 40 percent to avoid a runoff. So if you don't think he would have done much better in a world without a leaderboard selection, the actual outcome prevented by the new system wasn't a straight out win for Adams, but a runoff with either Wiley or Garcia (they were in the first round).
So, is it better that the ranked election do the area code quickly rather than leading it to a runoff that would have made a clear choice between Adams and an alternative?
Ranking activists have various criticisms of the runoff election. They say it is expensive for the city to hold and inconvenient for voters to have to vote again. They point out that turnout usually falls in runoff elections, arguing that this makes the end result less representative of what voters want. And they say that things tend to get very nasty and negative.
But a runoff has its virtues. It would have made a clear choice for voters between Adams and an alternate candidate (as opposed to the confusing strategies described above in the crowded field), and it would have ensured that both candidates were scrutinized by voters. And falling voter turnout is hardly a sure thing – as MSNBC's Steve Kornacki pointed out, turnout rose in the runoff when the New York Democrats last ran in 2001.
Perhaps a runoff election with more effort would have led to the same result, a victory for Adams. Or maybe his opponent could have made a better profile now that she was no longer in a crowded field. Again, we will never know for sure.
Were the slow and botched counts the fault of the choice?
Much of the grumbling about the leaderboard selection so far has centered on two things that aren't (mostly) its fault: the botched count and the slow count.
Last week's botched count, when the New York Electoral Board accidentally added about 135,000 "test votes" to its publicly published list, is clearly not a ranking selection problem – it was a mistake by an employee who was missed because of general sloppiness and incompetence among the board members. The most promising way to avoid such embarrassing mistakes in the future is to reform the long-ailing board yourself.
The blame for the slow count, meanwhile, rests primarily with New York City and state policies regarding postal vote counting. The state had an antiquated practice of counting postal votes only seven days after the election (unlike other states which start counting as soon as they come in). The pandemic-induced surge in postal votes has resulted in very slow counts in all New York elections (including those that did not use ranked voting in 2020) and efforts are being made to reform this.
Still, it's true that ranked picking usually means slower results than Americans are used to. In most US elections, much of the election night count is reported, and the media often unofficially "calls" a winner based on this information, even if the actual results are often not final for weeks. (A close election or any election where a significant portion of the votes are not counted may take longer.)
However, when voting for the ranked list, the election administrators must determine the order of the candidates so that they can eliminate them one by one and redistribute their ballots accordingly. You'll also need to decide whether to post a tentative reallocation list long before each ballot is counted (as New York City does to some reviews last week and again this week) or wait until every ballot is received before reallocating very long).
Is that important? Historically, progressives have tended to argue that slow elections are not a big deal because it is more important to ensure that every vote counts and the count is done correctly. Donald Trump's behavior in 2020 has sounded the alarm that a slow vote count could be manipulated by demagogues to instill public distrust of the election results. But it's hard to argue that those stakes are anywhere near that high in a mayoral election – parliamentary elections are months away, after all.