When one thinks of experiments in voting systems, you rarely think of motion pictures. About a decade ago (starting with the 2009 awards), the Academy of Motion Picture Arts and Sciences (AMPAS) changed the rules for picking the Academy Award (Oscar) for best picture. They made an additional change several years ago to the nomination process for Best Picture.
For most of the awards, the rules are that academy members belong to a given branch (e.g. actors, writers, etc.) and each member gets to submit a “nomination” ballot for the awards given by their branch (i.e. actors for the acting awards, directors for best director). Potential nominees who meet a certain threshold (up to 5 total) become the nominees. After nominations are announced, a second ballot goes out to all academy members who vote for one nominee. The nominee who gets the most votes (the traditional first-past-the-post system) is the winner.
In choosing the nominees, however, AMPAS has used a “single transferable vote” system for picking nominees. This system is similar to the one used in Ireland and in Australia (for Senate elections). In this system, voters list multiple candidates with a rank next to each candidate. There is a certain number of votes (quota) needed for a nomination (the total number of votes plus one divided by the available slots plus one). If a potential nominee received more than the require number of votes, their excess votes are distributed to the second choice of the voters. There are variations on how the distribution works. Some randomly pick excess ballots. Others assign a fraction of all ballot (i.e. if 10% of your ballots are excess, each ballot counts as 0.1 votes in the next round). For most of the awards, AMPAS uses a fractional system with one catch — surpluses are distributed only if the nominee exceeds the required number of votes by more than 20%. After the excess ballots are distributed, the process moves to the bottom ranked candidates. Those candidates are eliminated and those votes are redistributed to the second choice of the voters. (Or third or fourth choice depending upon who is already nominated or eliminated.) This process continues until all the slots are filled.
Over the past decade, AMPAS made two changes to nomination process. First, in 2009, AMPAS changed the maximum number of nominees for Best Picture to 10. Second, in 2011, AMPAS adopted a modification to its single transferable vote system. Under the modified system, instead, of multiple rounds of reassigning ballots, there is only one round of reassigning ballots. The 20% rule for excess votes over the quota remains. Additionally, a 1% threshold on the first ballot is used to determine which ballots are reassigned. If a movie got less than 1% on the first ballot, it is eliminated and all who voted for that movie as their first choice have their ballot reassigned to their top-ranked remaining movie. After the ballots are reassigned, it takes 5% of the total vote to be nominated. If more than 10 movies have more than 5%, the top ten are the nominees. Since the 2011 changes, there has not been a year in which 10 movies made the threshold with — depending on the year — either 8 or 9 nominees.
When AMPAS increased the total nominees from 5 to 10 for best picture in 2009, they also changed the rules for the final ballot. For the best picture ballot, AMPAS now uses alternative/preferential voting (similar to the system that Australia used for its House of Representatives). In this system, voters rank all of the nominees. After the first round, the last-placed candidate is eliminated. For voters who voted for the last-placed candidate, their ballot is reassigned to their second choice. This process continues until one candidate has a majority of the vote.
Part of what makes this process functional is the small number of voters and small number of awards. With approximately 6,000 members, fifteen branches (meaning an average of 400 members per branch, assuming every member votes) and 24 awards, the number of nomination ballots for a given award are not that large (other than potentially best picture). And for the final awards, only best picture is not first-past-the post. Finally, each ballot is a separate ballot making it possible to do a hand count with a limited team of counters. These features make it possible to have a large number of potential nominee candidates (every movie released over a calendar year and every actor/actress who had a role in a movie) with the final field still having a significant number of candidates (5-10 nominees). To replicate this in American elections would require a change from the current system in which we have a limited number of election dates and multiple levels use the same election and everybody votes for every office (meaning that in a typical election, the voter faces five or more county offices, state representative, state senator, three or more state-wide offices, U.S. representative, U.S. senator, and president for thirteen or more races) with voters for many of the offices exceeding 100,000. In the countries that use alternative voting or single transferrable voting, voters only face two or three offices up for election on the same day — allowing separate ballots for each office.
The other significant feature of award shows is that the counts remain secret. It would be interesting to know how the final nominees differed from the candidates who led after the first round. For best picture, it would be interesting to know how the films ranked after the initial round.
We know from regular elections that larger field tend to favor the front runners. For decades (prior to their being a producer’s guild award for best picture — first awarded in 1989), the winner of the director’s guild award tended to be a good predictor for both best picture and best director. (From 1950 to 1988, the winner of the director’s guild award won the Best Director’s Oscar 36 times out of 39 awards, and the film that received the Director’s Guild award won Best Picture 28 times out of 39 awards.) In the 28 years since the Producer Guild began giving a separate award, things have gotten a little less predictable. ( In those years, the Best Directors Oscar has gone to the Director’s Guild winner 25 times, and the Best Picture Oscar has gone has gone to either the winner of the Director’s Guild award or the Producer’s Guild award 24 times out of 28 years. In only one of those years did a movie win both Best Picture and Best Director without winning either guild award (Braveheart beating out Apollo 13 which won the two guild awards for 1995). We also know that alternative voting only makes a difference in a small number of elections. Particularly, if one candidate has a significant lead after the first round of voting, it is unlikely for another candidate to come from behind to win. On the other hand, in close elections, the trailing candidate can come from behind to win. There have even been close races when the third placed candidate can win under an alternative vote system.
Of course, with the actual results (other than the winner) being secret, this process is mostly hidden from the public. Given the attention that awards shows get, having the details of how the process worked made public might make the public more comfortable with different ways of conducting elections. One response to gerrymandering that I have seen being put forward by some interest groups is to move to multi-member districts with single transferrable vote. (Such districts are not completely immune from gerrymandering. As the Democratic presidential primaries show, even multi-member districts have key vote thresholds and it would be possible to draw lines that would switch a district from barely 3-2 to barely 4-1 or vice versa. Even with all districts being 3-2, you can pack the other party into 65% districts while the districts that favor your party are 56% districts.) However, convincing people to change from a familiar voting system to a new system is hard. In 2011, amid a lot of campaign rhetoric that misrepresented how alternative voting worked, the United Kingdom rejected alternative voting by approximately a two-thirds majority.
