Slashdot Mirror
Nate Silver, Microsoft Research Predict the Oscars
Nerval's Lobster writes "Nate Silver, famous for applying rigorous statistical methods to U.S. political elections, has focused his predictive powers on a somewhat more lighthearted topic: this weekend's Academy Awards. As part of his predictive analysis, Silver rounded up the various awards that precede the Academy Awards, including those from the Directors Guild of America and the Screen Actors Guild; in his calculations, he gave additional weight to those awards with a higher historical success rate, and doubled the score 'for awards whose voting memberships overlap significantly with the academy.' But he isn't the only statistician predicting this year's Oscar winners: David Rothschild, a member of Microsoft's massive research division, has also developed a data-driven model. What does their number-crunching predict? That Argo will win Best Picture, and a bunch of people will win other things."
"That Argo will win Best Picture, and a bunch of people will win other things."
No shit!
Because I could predict "a bunch of people will win other things."
"National Security is the chief cause of national insecurity." - Celine's First Law
Argo is a steaming pile of Ben Afleck's opportunitic, small minded piddle.
He could use his powers of predicition to devine when the Azure SSL cert needs to be renewed.
Microsoft research predicted with better than 98 percent accuracy of the results of the U.S. presidential election in 50 of 51 jurisdictions. And still, they released Windows 8? I suppose it might have been the 'best' of many even worse ideas, but you'd think they'd be better at predicting winning ideas with this kind of research.
hey! they admit it. newspapers and movies.
When life gives you lemons, release Win8.
Anonymous Coward, famous for anonymity and cowardice, doesn't care.
So having a guess from our Microsoft Research expert's blog:
1. Grabs the odds from Intrade, Betfair and HSX
2. Sources data from 'user generated data' ie social networks
3. Does a little a maths
4. Claims to be a forecasting guru
And with no real detail on #3 beyond being heavily weighted towards the betting/prediction markets and effectively just picking the favourites in every market, this is kind of useless.
The only interesting aspect is that the certainty for high likelihood winners is higher than any individual predictor ...
For example, Spielberg for Best Director with Lincoln:
Intrade: 75%
HSX: 51%
Betfair: 76%
User-data: 81%
Forecast: 88%
That suggests either:
- historically these prediction markets have under-estimated the numbers for popular favourites, which is consistent with inefficient betting markets where people will back long-shots more than they should due to the perception of good odds
- his model concludes that if 4 data points to a win, then the likelihood is even stronger
And still, they released Windows 8?
Ironically Windows 8 could very well be the result of Predictions, on what is successful, when you have Android overtaking Windows as being the dominant OS, taking *the selling features* without looking at *consistency* is just the kind of mistake this kind of analysis can create.
Its not though, its simply a method of forcing its entrenched developers, to develop mobile Applications in Windows...and force monopoly on a Desktop into a sizeable mobile marketshare if you add them together with an ill thought out hybrid device that maybe will blossom into a half useful Android tablet with a keyboard [they call the whole misguided thing an ecosystem] after they lost early entry to market [laughably by not being able to respond to market trends...things forecasting and predictions are all about]
He could use his powers of predicition to devine when the Azure SSL cert needs to be renewed.
sicky note. I would say outlook, but its not very good.
if you can't vote for CowboyNeal
Stay sentient. Don't drink bad milk.
[quote]...Microsoft have had there behinds kicked around so badly, that when they describe new technology they describe the pack of 4 "Apple, Google, Microsoft and Amazon", you would have thought with all this analysis, they could done better with the internet[all of it], social media, smartphones, mp3 players, tablets...all their money still comes from Office/OS.[/quote]
How is any of this even remotely relevant to a story about statistics?
Excuse the Unicode crap in my posts. That's an apostrophe, and slashdot is busted.
that's because their product success prediction mechanism has always been based on threats, lots of money and exaggerated advertising. It worked when those had value such as with Microsoft's monopoly position in desktop PC sectors. They have little correlation value outside of that area. Everyone at Microsoft probably knows that they guy behind the curtain throwing....the levers.... doesn't want anyone around who thinks differently from him so this is probably why these people are looking at the Oscars. ie they like Microsoft throwing money to them to keep them from doing anything of value for anyone else so they don't try to help Microsoft.
or not.
Just because I got laid last night doesn't necessarily mean I'm getting laid tomorrow night.
It's OK, but the acting is so-so, the bit at the end on the runway is laughable, the token whiny guy is just annoying... I'm not quite sure why it gets such good press.
that his performance on this is closer to his political prognostication and not his NFL predicitions.(You know, where he said the Patriots would go on to win the superbowl.)
Did you know 80 to 90% of the moderators on slashdot wouldn't recognize a troll even if one dragged them under a bridge.
Predictions (movie, director, best actor, best actress, supporting actor, supporting actress - the ones he got right in bold):
Argo, Spielberg , Day-Lewis, Lawrence, Jones, Hathaway
He only missed Spielberg (Oscar went to Ang Lee). Theirs were the closests predictors (0.58 vs 0.56 - compare with the next closest, best actress, 1.49 and 1.22 for the first two places).
I read because he has that rare geek skill to actually explain how he gets there with his result - and he does it masterfully, entertaining me. Right or wrong, his prediction for Oscars made sense and almost all fell in place, except he did not see Christoph Waltz coming. Also he agrees that this is not similar to predicting politics or sports, because lot of unknown data involved.
On unrelated note still lot of good cinema comes out every year, even in Hollywood - and not giving any movie clear victory this year is evidence of that. While I hate it for what I see as control obsession over their produce, I still loud artistic victory where I (subjectivelly of course) see one. And I celebrate all kind of good cinema.
user@ubuntubox:~$ stfu This server is going down for shutdown NOW!
Argo is offensive (to Canadians) because almost every plot point in the movie is utter fiction and the true story is a Canadian one, not CIA. Just for example, CIA Agent Mendez was in Iran for less than 36 hours. "Argo" wasn't his idea, it was a Canadian's. Research this all on the Internet for yourself. You'll see.
Argo should be offensive to all filmgoers because it's fiction being sold as truth. Affleck said the story was tweaked just a bit to maintain audience interest--and that's another lie on top of the lies.
It's not an awards show. It's an advertising campaign. Nothing but a big event various studios fund to slap a 'Go watch this!' stamp on their own products. The big awards have little if anything to do with the actual quality of the movie - it's all business.
As evidence of this claim, I just point out that Transformers won three oscars. Two of them for the sound.
Why the hell bother?
So Nate Silver does politics, and slashdot is all over it, and the /. article about it gets many many responses [currently shows that it had 576 responses in all, with 144 scoring at or over 2, 79 at or over 3, 56 at or over 4, and 32 scoring 5 /. poll asking about tonights Oscar Awards ceremony asking who was watching, who would be attending Oscar-watching parties, who even cared about who might win awards. The winning poll entry would probably be "crickets chirping", because since there are only 7 postings thus far that score over 2 points (and only 49 postings at all [!!!] at 2 a.m. PST the morning after), it's obvious that this is not the topic for /.
.
Now, Nate Silver predict-o'matics the Ocsar Awards and it become obvious that the slashdot crowd is definitely not the Oscar-watching, red-carpet fawning, entertainment-industry-drooling set of consumer-bots that Hollywood really wants to advertise to.
:>)
There ought to have been a
Only real miss: Best Supporting Actor (Swartz vice Tommy-Lee ... prediction was off by a good margin)
Other miss but too close to call: Best Director (Lee vs Afleck but they margin between the two in the prediction was so close as to be noise)
Good enough.
"Consensus" in science is _always_ a political construct.
How awards shows work http://www.youtube.com/watch?v=pmMCbsaiuNQ
Here's my own advanced statistic model about who will win Best Picture:
1. Any movie that says Hollywood is very important wins
2013: Argo - Hollywood director saves hostages
2012: The Artist - Dramatic portrait of Hollywood in 1930
Wow, I'm shocked. You mean the movie about how the world was saved by movie makers was the favorite among people who make movies for a living? How did that happen?
Chelloveck
I give up on debugging. From now on, SIGSEGV is a feature.