5) 10%: By Jan 1, 2013, there will be a way to directly input your estimated probability distribution across a range of different possible quantities when making a prediction on PB. (http://predictionbook.com/predictions/5063)
Number 5 is my somewhat lame attempt at a feature request. What I mean is that, for example, on #6, I’d like to be able to say, that I assign, say, a 15% chance to there being 0 such papers, a 15% chance of 1, a 20% chance of 2, and so on. Of course, I could make multiple predictions, but this is tedious. It’d be really nice to be able to assign probabilities to a full range of quantities on one question. (And I expect it would make my predictions more accurate, too.) Each individual probability assignment would have to be judged “correct” or “incorrect” independently.
60% seems way too high to me. The number of votes on 12 Angry Men and Pulp Fiction are both very high- for one to move up and the other to move down seems like it would require a large number of votes distributed differently from past votes.
12 Angry Men has continued to rise every year and I just don’t see it ending. Pulp Fiction is very violent and I expect the zeitgeist of film watching (and rating) to move farther and farther away from that. Finally, the movies on the top 250 are shrinkage estimated, scaled to the number of votes, and I expect the ratio of votes between 12AM and PF to decrease in the next year. Anyway, we’ll see!
imdb voting always had a strong bias towards old movies, but this is getting ridiculous.
These were decent movies, but not even remotely close to top 10. I don’t see how they’d even manage to legitimately get into top 100.
This might be the source of the problem: “For this top 250, only votes from regular voters are considered”. My suspicion is that their “regular voter” filter is broken and causes this problem.
These were decent movies, but not even remotely close to top 10. I don’t see how they’d even manage to legitimately get into top 100.
If you’re so sure, then what would your top 10 be?
“For this top 250, only votes from regular voters are considered”
They famously don’t say what the filter is, to prevent gaming. If it is broken now it must have been broken for a long time, because I don’t remember any major single-day jumps of late, except for times when they altered the “m” parameter.
If you’re so sure, then what would your top 10 be?
I have no issues with the following movies from the top list (there are some big omissions but they’re all awesome movies):
The Shawshank Redemption (1994)
Pulp Fiction (1994)
The Dark Knight (2008)
The Lord of the Rings: The Return of the King (2003)
Star Wars: Episode V—The Empire Strikes Back (1980)
Inception (2010)
Fight Club (1999)
The Lord of the Rings: The Fellowship of the Ring (2001)
Star Wars: Episode IV—A New Hope (1977)
The Matrix (1999)
Schindler’s List loses at least one star for its ending alone.
City of God is the only movie there I don’t remember watching, so I cannot say much either way. Godfather I’ve watched so long ago, I won’t be saying anything about it now.
Movies really get a lot better as time goes. Old movies near the top are simply not that good.
IMDB’s treatment of single multipart movies like Lord of the Rings and especially Kill Bill as multiple separate movies annoys me a lot, but that’s an entirely different story.
They famously don’t say what the filter is, to prevent gaming. If it is broken now it must have been broken for a long time, because I don’t remember any major single-day jumps of late, except for times when they altered the “m” parameter.
Broken as in drastically unrepresentative, not as in gameable.
Added some predictions:
1) 75%: On Jan 1, 2013, there will be 3 or fewer movies from 2011 on imdb’s top 250. (down from current 6) http://www.imdb.com/chart/top (http://predictionbook.com/predictions/5059)
2) 50%: On Jan 1, 2013, there will be seven or more movies from 2012 on imdb’s top 250. http://www.imdb.com/chart/top (http://predictionbook.com/predictions/5060)
3) 85%: The Shawshank Redemption will be #1 on imdb’s top 250 on Jan 1, 2013. (it is currently #1) http://www.imdb.com/chart/top (http://predictionbook.com/predictions/5061)
4) 60%: 12 Angry Men will be #5 or higher on imdb’s top 250 on Jan 1, 2013. (it is currently #6) http://www.imdb.com/chart/top (http://predictionbook.com/predictions/5062)
5) 10%: By Jan 1, 2013, there will be a way to directly input your estimated probability distribution across a range of different possible quantities when making a prediction on PB. (http://predictionbook.com/predictions/5063)
6) 50%: At least three papers with the word “connectomics” in their title or abstract will be published in Nature in 2012. (http://predictionbook.com/predictions/5064)
Number 5 is my somewhat lame attempt at a feature request. What I mean is that, for example, on #6, I’d like to be able to say, that I assign, say, a 15% chance to there being 0 such papers, a 15% chance of 1, a 20% chance of 2, and so on. Of course, I could make multiple predictions, but this is tedious. It’d be really nice to be able to assign probabilities to a full range of quantities on one question. (And I expect it would make my predictions more accurate, too.) Each individual probability assignment would have to be judged “correct” or “incorrect” independently.
60% seems way too high to me. The number of votes on 12 Angry Men and Pulp Fiction are both very high- for one to move up and the other to move down seems like it would require a large number of votes distributed differently from past votes.
12 Angry Men has continued to rise every year and I just don’t see it ending. Pulp Fiction is very violent and I expect the zeitgeist of film watching (and rating) to move farther and farther away from that. Finally, the movies on the top 250 are shrinkage estimated, scaled to the number of votes, and I expect the ratio of votes between 12AM and PF to decrease in the next year. Anyway, we’ll see!
imdb voting always had a strong bias towards old movies, but this is getting ridiculous.
These were decent movies, but not even remotely close to top 10. I don’t see how they’d even manage to legitimately get into top 100.
This might be the source of the problem: “For this top 250, only votes from regular voters are considered”. My suspicion is that their “regular voter” filter is broken and causes this problem.
If you’re so sure, then what would your top 10 be?
They famously don’t say what the filter is, to prevent gaming. If it is broken now it must have been broken for a long time, because I don’t remember any major single-day jumps of late, except for times when they altered the “m” parameter.
I have no issues with the following movies from the top list (there are some big omissions but they’re all awesome movies):
The Shawshank Redemption (1994)
Pulp Fiction (1994)
The Dark Knight (2008)
The Lord of the Rings: The Return of the King (2003)
Star Wars: Episode V—The Empire Strikes Back (1980)
Inception (2010)
Fight Club (1999)
The Lord of the Rings: The Fellowship of the Ring (2001)
Star Wars: Episode IV—A New Hope (1977)
The Matrix (1999)
Schindler’s List loses at least one star for its ending alone.
City of God is the only movie there I don’t remember watching, so I cannot say much either way. Godfather I’ve watched so long ago, I won’t be saying anything about it now.
Movies really get a lot better as time goes. Old movies near the top are simply not that good.
IMDB’s treatment of single multipart movies like Lord of the Rings and especially Kill Bill as multiple separate movies annoys me a lot, but that’s an entirely different story.
Broken as in drastically unrepresentative, not as in gameable.
EDIT: Here’s my theory.