Voting for show business awards such as Oscars, Emmys, and Tonys

(Return to main page)

The "OSCARs" (Academy Awards for Cinema) almost certainly have been using suboptimal voting systems. The purpose of this page is to collect suspicious historical OSCAR examples and consider how well different voting systems would (and did) handle them.

News flash (January 2013): after this page was written, and perhaps influenced by it, the OSCARs switched to reweighted range voting to select winners in the Visual Effects category.

Warning: I am not a film expert, and have not done enough research about OSCARs. Therefore, it is quite possible that some of the examples on this page contain stupid errors or misunderstandings by me. I am presenting them only as fuel for future research by people who understand more than I do about films, and not as the "final word." Also, the actual OSCAR vote totals have generally (always?) been kept secret, which makes analysis more difficult. Caveat Emptor.

Update (Feb. 2014): A collection of alleged OSCAR mistakes which was made by genuine film experts, was compiled by The Atlantic magazine. It includes several overlaps with my collection below (as well as several mistakes made by their "experts").

The current (2012) OSCAR voting system is basically:

However, it is not that simple; e.g. in some award-categories other systems are used. Unfortunately: Plurality voting suffers from well-known pathologies such as "vote splitting," "cloning," and "reversal failure." STV-PR sometimes can fail to nominate clear "beats-all" winners while RRV would usually avoid those failures.

A plausibly better system: would be RRV to select the nominees and then range voting to choose the winner.

News flash (August-Sept. 2009): The academy just changed its voting system for "best picture" to now have 10 (not 5) nominees, and the final vote among the 10 will now be conducted using instant runoff voting. However, as far as I understand, all other award categories will stay with the old voting system(s) and 5 nominees. IRV suffers numerous pathologies. I personally am optimistic that this rule change will yield a substantial performance improvement, but doubt the new rules are optimal. It's plausible that embarrassing pathologies like "thwarted majority paradox" will occur (just as they did in the Burlington VT mayor election of 2009) and it is a pity to sacrifice RRV's and range voting's boons of granting voters variable strengths of preference and the ability to withhold judgment on movies that voter feels too ignorant about.

(Presumed) Goals of the voting & nomination systems

Being "nominated" is itself a (lesser) "award." Their foremost goal probably is to generate "buzz" i.e. publicity to benefit the movie industry. If the nominees are more diverse, then there presumably would be greater buzz and greater overall summed benefit for the industry. That is why a PR voting system, which to some extent sacrifices quality to get a more-representative/diverse sample, is probably a good idea for nomination process.

To see that, compare

  1. Suppose they nominated 5 war movies (non-diverse). In that case, the Public would probably figure "I'm not interested in seeing 5 war movies, I'll only go to one." Result: not a lot of summed benefit for the industry.
  2. They nominate 5 diverse movies. Public might actually be interested in seeing all 5. Result: much larger summed-benefit for industry.

However, it also is desirable for the nominees to include "the best," otherwise the system embarrassingly fails to nominate the best film (which has happened).

Here are some books, academic papers, and newspaper pieces about the OSCAR voting system and its foibles.

  1. French Twist: A Fair Way to Pick Oscars? this was a 17 March 2002 New York Times article by Rick Lyman about NYU politics professor Steven J. Brams claiming the Oscars should be done with "approval voting" rather than plurality voting. Brams says he can't believe Rocky would really have beat Taxi Driver in 1972 in a head-to-head race; this had to be some kind of vote-split pathology, e.g. between Taxi and the also-excellent All the President's Men and Network. (Personally, I must agree with Brams: Rocky stunk.) We can see how range voting would have done by checking the ratings on Yahoo Movies, internet movie database, and rotten tomatoes. The results support Brams's claim Taxi Driver was best:
    MovieYahoo usersIMDb raters (0-10)
    Bound for GloryB (based on 55 raters)7.3 (from 1218 raters)7.3 (from 17 reviews)
    RockyB+ (from 40187 raters)8.0 (from 64828 raters)8.1 (from 41 reviews)
    All President's MenB (from 14366 raters)8.0 (from 25842 raters)8.9 (from 39 reviews)
    NetworkB+ (from 1276 raters)8.1 (from 22899 raters)8.0 (from 42 reviews)
    Taxi DriverB+ (from 20830 raters)8.6 (from 116049 raters)8.8 (from 47 reviews)

  2. The NY Times piece was based on this unpublished paper (pdf) by S.J.Brams and Paul Hager. They say:
    The most famous, or infamous, incident [was] in 1950. That year, both Bette Davis and Anne Baxter received Best Actress nominations for All About Eve. When Judy Holliday became the surprise winner in Born Yesterday, it was widely believed that Davis and Baxter had split the vote.
    Brams & Hager also question the Academy's choice of Kramer vs. Kramer as best picture of 1979. Again their view is supported by the internet ratings, and again a vote-split seems the plausible cause:
    MovieYahoo usersIMDb raters (0-10)
    Norma RaeB (from 336 raters)7.2 (from 2860 raters)7.6 (from 19 reviews)
    All that JazzB+ (from 346 raters)7.5 (from 7555 raters)7.4 (from 31 reviews)
    Kramer v. KramerB (based on 1831 raters)7.7 (from 17625 raters)7.9 (from 32 reviews)
    Breaking AwayB (from 7840 raters)7.6 (from 7523 raters)8.3 (from 33 reviews)
    Apocalypse NowB+ (from 4308 raters)8.6 (from 136199 raters)7.7 (from 77 reviews)

  3. Decision Rules for the Academy Awards Versus Those for Elections, in Interfaces 34,3 (2004) 226-234 by professors W.V.Gehrlein & H.V.Kler. They cited a USA Today article which had trouble seeing how Marisa Tomei won best supporting actress for My Cousin Vinny in 1992, and had speculated it must have been another vote-split pathology caused by using plurality voting instead of scoring.

  4. Statistical model predicts OSCAR winners, short piece in New Scientist 3/29/2008.

  5. E.Levy: OSCAR Fever: The History and Politics of the Academy Awards. Continuum International Publishing Group, New York 2001.

  6. Here's an interesting example where range voting (unusually for OSCARs!) was used as a nominating procedure [but note: we recommend range voting for single-winner elections and it is not clear to us what is best for multiwinner elections such as OSCAR nominations, plausibly RRV is better] and complaints followed: Rolling Stone magazine (early 2009) wrote:
    The biggest Bruce Springsteen controversy of the season... Springsteen fans are shocked that his stunning song "The Wrestler" (from the film of the same name) didn't get an Oscar nod – despite the fact that he won a Golden Globe for the tune. Instead of selecting five nominees (as in most other categories), the Acedym chose only three: Peter Gabriel's "Down to Earth" from WALL-E, and two from Slumdog Millionaire.

    The Wrestler may have been a victim of the Oscars' complex nominating rules. The Academy music branch watched all 49 eligible songs in a clip reel provided for them. Then they rated the tunes on a scale of 6.25 to 10 – and only the movies with averages of 8.25 or better made the cut. That system seems to give an edge to certain films. "A song that has a performance aspect is going to do better than one that doesn't," says Oscar nominee Danny Elfman's agent, Richard Kraft. "It's hard to compete with the music video at the end of Slumdog, while in 'The Wrestler', you've got the names of key grips scrolling by."

Some more examples of suspicious "final round" voting for "best picture" Oscar

1941: Citizen Kane, which has often been cited as best film ever, got little Academy recognition. It won "best screenplay" but nothing else in spite of being nominated for many other academy awards including best picture. That year the Best Picture was How green was my valley while the other nominees were Blossoms in the Dust, Here Comes Mr. Jordan, Hold Back the Dawn, The Little Foxes, The Maltese Falcon, One Foot in Heaven, Sergeant York, and Suspicion.

[I personally found Valley to be sickeningly saccharine and utterly unrealistic, but admit I quit watching it after about 20 minutes.]

How green was my valley holds a 7.9 rating from IMDb (5568 raters as of 2008) while Kane has 8.6 (113062 raters), Maltese Falcon 8.4 (40259 raters), and Sgt. York 8.0 (4822 raters).

With this many nominees including two strong films in Kane and Maltese, plurality voting was highly likely to malfunction. I find it very hard to believe Kane would have lost a head-to-head race versus Valley.

1944: The too-sweet musical Going My Way (8.2) won best picture (also best actor, Bing Crosby, and best supporting actor Barry Fitzgerald, best story for Leo McCarey, and best screenplay for Frank Butler and Frank Cavett). Meanwhile Billy Wilder's film noir masterpiece Double Indemnity got nothing. (It was nominated for best actress, music, director, picture, cinematography.) But over time, the raters on IMDb thought differently. As of June 2009, they gave Going My Way a rating of 7.5 whereas Double Indemnity got 8.6. Similarly on, Double Indemnity got 8.7 while Going my way got 7.1.

1950: All about Eve won best picture. But it got an 8.4 rating on IMDb. Meanwhile Sunset Boulevard got 8.7. I admit, these IMDb ratings are close enough that you cannot call this a robbery.

1957: 12 angry men was rated 8.8 by 78K IMDbers. Although nominated, it lost to Bridge on the River Kwai which 45K IMDbers rated only 8.4. The other nominees were Witness for the Prosecution (8.3), Peyton Place (7.2), and Sayonara (7.2). This decision, since it is only an 0.4 rating-gap on IMDb, might be defensible. However, one cannot help suspecting that the real reason for Bridge's victory was a vote split between Men and Witness, which were both good legal dramas and thus perhaps (in vote-theory terminology) "clones." Bridge, as the only war movie, presumably did not suffer any vote-split.

2011: See this analysis of Nate Silver's forecasting analysis of the 2011 Oscars, versus what actually happened. It appears that the IRV voting system used in the final round, robbed Toy Story 3 of what would have been a historic best-picture award (first ever victory for an animated film). The official winner was The King's Speech.

Suspicious examples where great films failed even to be nominated

Alfred Hitchkock's famous Psycho (1960) – often said to be best of its kind of all time and a classic – failed to win any OSCAR and failed even to get nominated as best picture.

Monty Python's Life of Brian (1979) won numerous votes as best comedy of all time – no OSCAR, not even a nomination.

The Searchers (1956) was named Greatest Western of all time by the American Film Institute in 2008. No OSCARs, not even a nomination.

The Good, the Bad and the Ugly (1966), often also said to be best western ever, and ranked as 4th best film ever and as best western ever by IMDb raters (as of January 2009) – no OSCAR, not even a nomination.

In 1959, Ben-Hur (8.2) won best picture, defeating Anatomy of a murder (8.1), Nun's story (7.5), Room at the top (7.9), and Diary of Anne Frank (7.6). [Year 2009 IMDb ratings shown in parentheses.] But three excellent rivals failed even to be nominated: Some Like It Hot (8.4), North by Northwest (8.6), and Wild Strawberries (Smultronstället) (8.3), all of which would have defeated all of the nominees in terms of year-2009 IMDb ratings! Indeed in 2000, the American Film Institute listed Some Like It Hot as the greatest American comedy film of all time.

When Bette Davis failed to land a nomination as best actress for Of Human Bondage (1934), there was an outcry, which led to a write-in vote for her being permitted despite her non-nomination. (In the final tally, at a time when the Academy made the votes public, she finished third.) However, write-in voting has been banned since 1936.

In January 2013, Ben Affleck won both the Hollywood foreign press association's "Golden Globe" and the Broadcast Film Critics' Association "best director" cinema awards for his movie Argo (rated 8.1 by 64590 internet raters on IMDb and 96 on the "tomatometer"). However, oddly, he failed even to achieve nomination for the more-prestigious OSCAR best-director award announced later in that same month. [Also not nominated: Kathryn Bigelow (Zero dark thirty 7.7, 93) and Quentin Tarantino (Django unchained 8.7, 88).] The five nominees were (in no particular order):

  1. Steven Spielberg (Lincoln 8.0, 91)
  2. Ang Lee (Life of Pi 8.3, 89; ultimate winner)
  3. David O. Russell (Silver linings playbook 8.2, 91)
  4. Benh Zeitlin (Beasts of the Southern Wild 7.5, 86)
  5. Michael Haneke (Amour 8.1, 93).
This surprising discrepancy caused various conspiracy theories, as well as blame-it-on-the-STV-voting-system theories, to appear in the press. For example in the Guardian,

It all rather depends on the reason Affleck was snubbed in the first place. If Argo has simply been overrated by the Oscar commentariat, that's one thing. But some are suggesting that Affleck came a cropper of the preferential voting system of the nomination process which rewards small but passionate groups of supporters – those for Michael Haneke or Behn Zeitlin, for example. Furthermore, voters may have felt they had to choose between Affleck's CIA thriller and Kathryn Bigelow's, which may have had the effect of cancelling each other out. The result was an accidental snub rather than an intentional diss.

Meanwhile the New York Times (17 January 2013 page C1) went with a headline A Snub by Oscars? Affleck Has an Answer, the NY Post and Huffington Post and both Eonline and all also called this a "snub" for Affleck – and said many were demanding "write-in votes" (currently illegal under OSCAR voting rules) to allow Affleck a chance to win despite not being nominated. E.g. the latter commented

Either it's time to dismiss the Globes as any kind of legitimate prognosticator or time for the Academy to recognize it has made a mistake and do the right thing: reintroduce a write-in vote.

One way in which even a clearly best candidate could fail to receive an Oscar nomination is demonstrated by this pathology example: /PRcond.html. It also is possible with their PR-STV system for a "vote split" to trigger that same kind of pathology, e.g. eliminating two candidates who otherwise would have been top, because neither has enough top-rank votes to survive round #1. RRV, however, does not suffer from (or anyhow far less) either of those two pathologies.

Did some such pathology happen in the 2013 OSCARs?

Return to main page