Oscar's Complicated Nomination Math -- Stats Genius, Help!
#1
DVD Talk Legend
Thread Starter
Oscar's Complicated Nomination Math -- Stats Genius, Help!
Entertainment Weekly has set out to explain the complicated process by which the 10 Best Picture nominees are decided.
Here's a link to that: http://insidemovies.ew.com/2011/01/2...-poll-results/
The process is so complicated that I couldn't summarize it. And the EW article is too long to post in its entirety. Here's a bit of the process:
[Up to] 5755 Academy members cast ballots listing ten films in their order of preference. The ballots are sorted according to the voter's choice for #1 film.
Any movie that is not at least one voter's #1 choice is completely eliminated from the voting process, no matter how often it is listed on ballots as choice 2-10.
[# of ballots cast] divided by 11 equals a "magic number". If a film has more #1 votes than the magic number, it gets a slot as a Best Picture Nominee.
Then things start getting more complicated.
Suppose a voter chose The Social Network as the #1 film and many other voters agreed and TSN far surpassed the magic number.
Then that voter's #2 pick (perhaps Winter's Bone) is given weight and added to all #1 votes for WB.
Why include all this extra math? It assures Academy members that should they pick an extremely popular film, their vote won’t be wasted.
“You are essentially casting one vote,” explains Academy executive director Bruce Davis, “and the system gives your vote to the picture that most needs it.”
The process continues with the #2 picks being elevated on ballots where the #1 pick lacks support to get to the final ten nominees.
I was hoping someone who really understands statistics could explain why this is a better system than assigning values to each pick [10 points for #1, 9 points for #2, etc] and just totaling up the points.
Here's a link to that: http://insidemovies.ew.com/2011/01/2...-poll-results/
The process is so complicated that I couldn't summarize it. And the EW article is too long to post in its entirety. Here's a bit of the process:
[Up to] 5755 Academy members cast ballots listing ten films in their order of preference. The ballots are sorted according to the voter's choice for #1 film.
Any movie that is not at least one voter's #1 choice is completely eliminated from the voting process, no matter how often it is listed on ballots as choice 2-10.
[# of ballots cast] divided by 11 equals a "magic number". If a film has more #1 votes than the magic number, it gets a slot as a Best Picture Nominee.
Then things start getting more complicated.
Suppose a voter chose The Social Network as the #1 film and many other voters agreed and TSN far surpassed the magic number.
Then that voter's #2 pick (perhaps Winter's Bone) is given weight and added to all #1 votes for WB.
Why include all this extra math? It assures Academy members that should they pick an extremely popular film, their vote won’t be wasted.
“You are essentially casting one vote,” explains Academy executive director Bruce Davis, “and the system gives your vote to the picture that most needs it.”
The process continues with the #2 picks being elevated on ballots where the #1 pick lacks support to get to the final ten nominees.
I was hoping someone who really understands statistics could explain why this is a better system than assigning values to each pick [10 points for #1, 9 points for #2, etc] and just totaling up the points.
#2
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
It's amazing that they get 10 films nominated that way if every film nominated has to be someone's #1. Which means that somebody out there thinks The Kids Are Alright is the best film of the year. Weird.
#3
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
#4
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Okay I think I get it. First, if you hit the magic number, you're in, end of story for that (more on my assumptions on this magic number below).
Next, the important thing to remember is that everybody gets 1 vote for a movie to be nominated for BP. But, that "1" can be split up depending on how popular a choice can be.
Now if a movie hits the magic number and exceeds it by 20%, then they have to split that vote. My assumption on why it's 20%
Now it's time to move down to determine what else will get in. Everything that has 1% of the vote is eliminated. The vote is divided up
and redistributed between the person's #1 pick and that same person's #2 pick. Since the #1 pick is already guaranteed, it does not lose any "votes" and the person's #2 pick still gets some recognition. Now if the #2 pick is nominated/eliminated, you go down the list until you hit a movie that is still eligible.
Reason for the distribution:
Okay say that out of a pool of 2000 votes, 3 movies hit the magic number and two movies hit it and exceeded it by 20% exactly. That gives us 52.72% of the votes accounted for.
Now what about the remaining 5 movies, and the remaining 47.27% of the vote (945 votes left)?
We could just go down the rest of the #1s and add them up, but that is only 945 amount of votes, and they did not hit the magic number. Why should we allow 47% percentage of the votes to determine the bottom 50% of the movies? That is unfair to the movies and voters that got the top 5 picks, and it would defeat the purpose of choosing your top ten. That means that people who voted for the top few movies had their vote count for 50/52, around 0.96 votes, and 50/47 for the people who voted for the bottom half. That's 1.06 votes.
But in reality...
This is a really simple example where movies get exactly 20%. But a movie can also get WAY over 20%, and thus reduce the voting pool even further, giving the "lower half" even more unbalanced voting power.
So each person's vote needs to have equal power. Having the top half and their respective voters penalizes them to get less than 1 vote (collectively) and the bottom half gets more than one vote for movies that didn't even get as many votes in the first place. Not a fair distribution.
So then we go to the #2s of the top movies, since otherwise what was the point of ranking them? Only for those voters whose movies hit 20% or above (2 movies in our example), do their votes get redistributed. This helps to make the distribution of votes fair, as well as give a chance to the movies that are most likely to get the most votes.
CONCLUSION
This is a vote from the list of all eligible movies and so there is going to be a huge distribution of movies as well as top ten lists. If they just chose everybody's #1 choices, there is no guarantee that each of these will hit that magic number (though it is possible, it is extremely unlikely, but the numbers allow for it to occur). That magic number is set such that a significant amount of votes are cast and it is a representative amount of the Academy that selected the flick (9.09%), and then after that go down the list (and backwards from movies that are too low to get in). In addition the redistribution is necessary to ensure that each vote has relatively equal weight.
As for not doing it by 10/9/8 points for 1/2/3, that would give uneven distribution, so for example, a movie gets more #1 votes than a #2 movie, but that #2 movie also has more #3 points, and beats out #1.
Next, the important thing to remember is that everybody gets 1 vote for a movie to be nominated for BP. But, that "1" can be split up depending on how popular a choice can be.
Now if a movie hits the magic number and exceeds it by 20%, then they have to split that vote. My assumption on why it's 20%
Spoiler:
Now it's time to move down to determine what else will get in. Everything that has 1% of the vote is eliminated. The vote is divided up
Spoiler:
Reason for the distribution:
Okay say that out of a pool of 2000 votes, 3 movies hit the magic number and two movies hit it and exceeded it by 20% exactly. That gives us 52.72% of the votes accounted for.
Spoiler:
Now what about the remaining 5 movies, and the remaining 47.27% of the vote (945 votes left)?
We could just go down the rest of the #1s and add them up, but that is only 945 amount of votes, and they did not hit the magic number. Why should we allow 47% percentage of the votes to determine the bottom 50% of the movies? That is unfair to the movies and voters that got the top 5 picks, and it would defeat the purpose of choosing your top ten. That means that people who voted for the top few movies had their vote count for 50/52, around 0.96 votes, and 50/47 for the people who voted for the bottom half. That's 1.06 votes.
But in reality...
This is a really simple example where movies get exactly 20%. But a movie can also get WAY over 20%, and thus reduce the voting pool even further, giving the "lower half" even more unbalanced voting power.
So each person's vote needs to have equal power. Having the top half and their respective voters penalizes them to get less than 1 vote (collectively) and the bottom half gets more than one vote for movies that didn't even get as many votes in the first place. Not a fair distribution.
So then we go to the #2s of the top movies, since otherwise what was the point of ranking them? Only for those voters whose movies hit 20% or above (2 movies in our example), do their votes get redistributed. This helps to make the distribution of votes fair, as well as give a chance to the movies that are most likely to get the most votes.
CONCLUSION
This is a vote from the list of all eligible movies and so there is going to be a huge distribution of movies as well as top ten lists. If they just chose everybody's #1 choices, there is no guarantee that each of these will hit that magic number (though it is possible, it is extremely unlikely, but the numbers allow for it to occur). That magic number is set such that a significant amount of votes are cast and it is a representative amount of the Academy that selected the flick (9.09%), and then after that go down the list (and backwards from movies that are too low to get in). In addition the redistribution is necessary to ensure that each vote has relatively equal weight.
As for not doing it by 10/9/8 points for 1/2/3, that would give uneven distribution, so for example, a movie gets more #1 votes than a #2 movie, but that #2 movie also has more #3 points, and beats out #1.
Last edited by bluetoast; 02-01-11 at 12:48 AM.
#6
DVD Talk Hall of Fame
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
I realize there's a lot to cover here, but it's section (6) that really explains the nuts and bolts of this process. Reading that really cleared it up for me:
In short, that clause applies to "over-achievers." In those cases, rather than have an unnecessarily high number of ballots dedicated to a movie that was already qualified for a slot, their votes are redistributed to help out their second choice which may not have otherwise had as many #1 votes. It makes sense to me knowing that's what's actually going on.
What follows is a tangent in which I try to relate to the system using a voting project of my own. It may be of no used to you and you're welcome to skip it entirely.
(6) For any movie that initially exceeds the magic number by at least 20 percent, there’s an extra (and extra-complicated) step: Each ballot is still worth only one point, but that point is split between the voter’s already-selected No. 1 and his or her next eligible choice. For example, Inception received more than twice the No. 1 votes required to get nominated, so its ballots were re-sorted and now worth half a point, with a half point going to another film. Why half a point? More math! Again, this only applies to films that exceed the magic number by 20 percent [or 182 X 1.2 = 218.4], and this step only occurs after the first round of ballot sorting — it’s never repeated in a later round.
Inception received 425 No. 1 votes, clearly passing the 218.4 votes need to activate this so-called “surplus” rule. The formula: [# of ballots the film received – magic number] / [# of ballots film received] = [425 – 182] / [425] = [243] / [425] = 0.57. This result is always rounded down to the nearest 10th, so for Inception, it became 0.5. Thus, every voter who selected Inception as No. 1 has their ballot redistributed to his or her next eligible choice, with each ballot now being worth only 0.5 points, or half a vote. Why include all this extra math? It assures Academy members that should they pick an extremely popular film, their vote won’t be wasted. In our simulation, the ballots for Inception, The Social Network, and Black Swan were redistributed, and as a result, The King’s Speech got enough new points to pass the magic number — it’s nominated.
Inception received 425 No. 1 votes, clearly passing the 218.4 votes need to activate this so-called “surplus” rule. The formula: [# of ballots the film received – magic number] / [# of ballots film received] = [425 – 182] / [425] = [243] / [425] = 0.57. This result is always rounded down to the nearest 10th, so for Inception, it became 0.5. Thus, every voter who selected Inception as No. 1 has their ballot redistributed to his or her next eligible choice, with each ballot now being worth only 0.5 points, or half a vote. Why include all this extra math? It assures Academy members that should they pick an extremely popular film, their vote won’t be wasted. In our simulation, the ballots for Inception, The Social Network, and Black Swan were redistributed, and as a result, The King’s Speech got enough new points to pass the magic number — it’s nominated.
What follows is a tangent in which I try to relate to the system using a voting project of my own. It may be of no used to you and you're welcome to skip it entirely.
Spoiler:
#7
DVD Talk Hero
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
That said, it's 5755 people. I can't imagine that the "not even one vote" happens to any worthy contenders. But then I guess you always hear about the 'snubs' (like Zodiac). Though I'd gamble and say that at least one person wrote in Zodiac as their number one
The ten point system leaves a strong possibility that a movie, who the Academy thinks "hey, that was alright, it's a number three" as the Best Picture. It's either you win, or you don't. They don't hand out silver and bronze awards at the Oscars.
#8
DVD Talk Limited Edition
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
I blame Coopers from PricewaterhouseCoopers. Ever since they joined the firm, the math has been out of control!
#9
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
This is just the system for how they determine the nominees, not the ultimate winner.
#10
DVD Talk Limited Edition
Join Date: Oct 2002
Location: Muskegon, MI
Posts: 5,199
Likes: 0
Received 0 Likes
on
0 Posts
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Steve Pond of The Wrap explains it here. Maybe that will help.
How Oscar Comes Up[ with it's Nominees: A Demonstration
The winner is simply the movie with the most votes.
How Oscar Comes Up[ with it's Nominees: A Demonstration
The winner is simply the movie with the most votes.
#11
DVD Talk Platinum Edition
Join Date: Oct 2008
Posts: 3,046
Likes: 0
Received 0 Likes
on
0 Posts
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
It's a somewhat ridiculous system... it will get the majority of the list right, but the system allows for oddball choices to sneak in to the bottom 2-3.
Also while the system is supposed to not punish people for making a popular choice (as the 2nd choice can get some of the vote), it also doesn't do anything to prevent people from making oddball choices as if it that movie doesn't get enough votes, the movie gets dropped and the 2nd choice on that ballot kicks in. The problem is you only need a small minority to put a film first (or 2nd) to get it nominated and it could be left off over 85% of the ballots and it wouldn't matter.
With only 5 movies on the nomination list this was a lot more difficult (you would need just under 20%)... now that bar has been lowered to just under 10%. I'd expect some oddball choices to sneak in from time to time.
The other issue is that people tend to put more effort in getting the top 3 or 5 movies right and the rest of the list can be a throwaway for most folks. So you get enough people sticking a random movie in at # 7-10 on their list and as the rounds go by and the first 5 or 6 movies get crossed off, you now have the portion of the list that people probably did not put a lot of time and effort into.
Also while the system is supposed to not punish people for making a popular choice (as the 2nd choice can get some of the vote), it also doesn't do anything to prevent people from making oddball choices as if it that movie doesn't get enough votes, the movie gets dropped and the 2nd choice on that ballot kicks in. The problem is you only need a small minority to put a film first (or 2nd) to get it nominated and it could be left off over 85% of the ballots and it wouldn't matter.
With only 5 movies on the nomination list this was a lot more difficult (you would need just under 20%)... now that bar has been lowered to just under 10%. I'd expect some oddball choices to sneak in from time to time.
The other issue is that people tend to put more effort in getting the top 3 or 5 movies right and the rest of the list can be a throwaway for most folks. So you get enough people sticking a random movie in at # 7-10 on their list and as the rounds go by and the first 5 or 6 movies get crossed off, you now have the portion of the list that people probably did not put a lot of time and effort into.
#12
DVD Talk Hall of Fame
Join Date: Jun 2002
Posts: 8,487
Likes: 0
Received 0 Likes
on
0 Posts
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Not anymore. As of last year, when the 10-movie system was introduced into the Best Picture category, the final voting in that category is now weighted as well. The winners in the other categories, however, are still chosen the traditional way -- by the most votes.
#13
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
You only get to vote for one film so the voting can't be weighted.
#14
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Steve Pond of The Wrap explains it here. Maybe that will help.
How Oscar Comes Up[ with it's Nominees: A Demonstration
How Oscar Comes Up[ with it's Nominees: A Demonstration
#16
DVD Talk Hall of Fame
Join Date: Jun 2002
Posts: 8,487
Likes: 0
Received 0 Likes
on
0 Posts
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Again, as of last year, this isn't the case. The voters rank the films from #1 to #10 on the final ballot for the Best Picture category. This was a major part of the discussion in the Hurt Locker vs. Avatar vs. Inglourious Basterds race last year.
More info:
More info:
Hollywood has become embroiled in a row ahead of the Oscars over a new 'transferable vote' system which critics claim could mean the award for best picture could go to a film with just 11 per cent of the vote.
Film industry insiders argue that it is possible for a film to win the Oscar even if it doesn't have the most first choice votes, increasing the chance of a compromise choice winning.
It means that although Avatar, the science fiction epic, and The Hurt Locker, the Iraq war film, are clear joint favourites Quentin Tarantino's "Inglourious Basterds" or "Up in the Air" starring George Clooney could sneak through to win.
For the first time in six decades there will be 10 nominations in the best picture category, instead of the usual five.
This new "preferential voting" system, which is widely used in Australian political elections, requires judges to rank all 10 films in order of preference.
If no movie wins 51 per cent in the first round the one with the least support is eliminated and its votes are allocated to second choices. The process continues until one emerges as the clear winner.
Detractors say that the system has led to "ugly" campaigning and it has unleashed twice the amount of behind-the-scenes lobbying in the run-up to Sunday's awards ceremony.
Nicholas Chartier, one of the producers of The Hurt Locker has been barred from the event after sending an email asking judges to rank his film first and Avatar at number 10.
Anne Thompson, the former Variety deputy editor and influential Hollywood blogger said: "The campaigning has got ugly, the voting is confusing.
"One thing you've got to give the Academy is that all the old tea leaves we used to predict a winner are out the window. Really, anyone could win."
Steven Zeitchik at the LA Times warned that new Oscar rules could mean a contentious win, reminiscent of the Bush – Gore presidential election.
He said: "A movie could pull a Bush vs. Gore – win best picture despite not getting the most first-place votes. But, because the academy guarantees a secret ballot, few people would ever know."
He said understanding the new system was like "attempting to understand the system can sometimes feel a little like trying to divine the secrets of cold fusion".
The new voting system was introduced in an attempt to boost television ratings, attracting viewers by showcasing 10 films.
Television viewing figures for the Oscars have evaporated since 1998 when 55 million watched Titanic win best picture. A decade later, in 2008, they reached a nadir of 32 million.
That has had a crushing effect on advertising revenue and this year a 30-second slot costs just $1.4 million (£930,000) – down from 1.7 million (£1.13 million) two years ago.
A confident Tarantino said: "I think we're definitely in the race. I think it's anybody's game."
However, Oscars expert Gabriel Rossman, a professor of sociology at UCLA, said: "Inglourious Basterds might have a few fans but also a lot of people who hate it. This is the kind of film that the new system is designed to exclude."
The first people to know, whether Tarantino managed to win over the two favourites, will be PricewaterhouseCoopers auditors Rick Rosas and Brad Oltmanns who will conduct the complicated count in a windowless room at a secret location.
They will separate the 5,777 paper ballots into 10 piles and then begin the process of reallocating second choices until one film reaches the magic figure of 2,889.
Mr Rosas said he believed the new system should ensure the most popular film wins.
He said: "The key is to have the highest number of first-place votes. That will always be the case because that gives you the highest probability of winning."
Film industry insiders argue that it is possible for a film to win the Oscar even if it doesn't have the most first choice votes, increasing the chance of a compromise choice winning.
It means that although Avatar, the science fiction epic, and The Hurt Locker, the Iraq war film, are clear joint favourites Quentin Tarantino's "Inglourious Basterds" or "Up in the Air" starring George Clooney could sneak through to win.
For the first time in six decades there will be 10 nominations in the best picture category, instead of the usual five.
This new "preferential voting" system, which is widely used in Australian political elections, requires judges to rank all 10 films in order of preference.
If no movie wins 51 per cent in the first round the one with the least support is eliminated and its votes are allocated to second choices. The process continues until one emerges as the clear winner.
Detractors say that the system has led to "ugly" campaigning and it has unleashed twice the amount of behind-the-scenes lobbying in the run-up to Sunday's awards ceremony.
Nicholas Chartier, one of the producers of The Hurt Locker has been barred from the event after sending an email asking judges to rank his film first and Avatar at number 10.
Anne Thompson, the former Variety deputy editor and influential Hollywood blogger said: "The campaigning has got ugly, the voting is confusing.
"One thing you've got to give the Academy is that all the old tea leaves we used to predict a winner are out the window. Really, anyone could win."
Steven Zeitchik at the LA Times warned that new Oscar rules could mean a contentious win, reminiscent of the Bush – Gore presidential election.
He said: "A movie could pull a Bush vs. Gore – win best picture despite not getting the most first-place votes. But, because the academy guarantees a secret ballot, few people would ever know."
He said understanding the new system was like "attempting to understand the system can sometimes feel a little like trying to divine the secrets of cold fusion".
The new voting system was introduced in an attempt to boost television ratings, attracting viewers by showcasing 10 films.
Television viewing figures for the Oscars have evaporated since 1998 when 55 million watched Titanic win best picture. A decade later, in 2008, they reached a nadir of 32 million.
That has had a crushing effect on advertising revenue and this year a 30-second slot costs just $1.4 million (£930,000) – down from 1.7 million (£1.13 million) two years ago.
A confident Tarantino said: "I think we're definitely in the race. I think it's anybody's game."
However, Oscars expert Gabriel Rossman, a professor of sociology at UCLA, said: "Inglourious Basterds might have a few fans but also a lot of people who hate it. This is the kind of film that the new system is designed to exclude."
The first people to know, whether Tarantino managed to win over the two favourites, will be PricewaterhouseCoopers auditors Rick Rosas and Brad Oltmanns who will conduct the complicated count in a windowless room at a secret location.
They will separate the 5,777 paper ballots into 10 piles and then begin the process of reallocating second choices until one film reaches the magic figure of 2,889.
Mr Rosas said he believed the new system should ensure the most popular film wins.
He said: "The key is to have the highest number of first-place votes. That will always be the case because that gives you the highest probability of winning."
#17
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Sorry, but I didn't follow some of what you were explaining.
Looking over the information that zekeburger1979 linked to, I found a concrete example of the difference between the two methods.
Using the critics' lists and the 1/2/3 --> 10/9/8 scoring method, True Grit is the #9 movie and Carlos is #10.
Using the Oscar's method, Carlos is the #4 movie and True Grit doesn't make the list at all.
Carlos got ten #1 votes and True Grit got six #1 votes. But True Grit appeared on more lists overall (66) than did Carlos (41).
So, the upshot is that the Oscar math will result in a nomination for a film with a smaller but more passionate pool of voters, and a film with a broader less enthusiastic base of support will be left out.
It seems like this is the exact opposite of the Academy's agenda for expanding the nomination field from five to ten.
Using the critics' lists and the 1/2/3 --> 10/9/8 scoring method, True Grit is the #9 movie and Carlos is #10.
Using the Oscar's method, Carlos is the #4 movie and True Grit doesn't make the list at all.
Carlos got ten #1 votes and True Grit got six #1 votes. But True Grit appeared on more lists overall (66) than did Carlos (41).
So, the upshot is that the Oscar math will result in a nomination for a film with a smaller but more passionate pool of voters, and a film with a broader less enthusiastic base of support will be left out.
It seems like this is the exact opposite of the Academy's agenda for expanding the nomination field from five to ten.
#18
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
I was wrong. I apologize.
#19
DVD Talk Ultimate Edition
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
Why was it necessary to go to 10 films? Doesn't make sense to have that many films nominated. Five, maybe 6 films is enough. Was this simply politically correct nonsense taking over?
#20
DVD Talk Hall of Fame
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!
For instance, I don't think Up would have made a ballot of five nominees, but I certainly agreed it was one of the ten strongest films of 2009. Knowing that The Hurt Locker won Best Picture helps establish a continuum for me of those ten movies. For the purpose of deciding which film to award the statuette, it's entirely irrelevant. But for the purpose of generating thought and renewed criticism of film, I think it's healthy to have ten movies in contention.
#21
DVD Talk Legend
Thread Starter
Re: Oscar's Complicated Nomination Math -- Stats Genius, Help!