Hacker News Reader

Comment by smackfu | Hacker News Reader

This is essentially just Google using their search data to replace the tracking polls that movie studios do already. It's not hard to predict a movie's box office right before it comes out, but what good does that do you, since all the money has been spent already?

View on HN

Exactly, this would be a awesome predictions if it started earlier in the production, like at the kernel of the idea or even at the first draft of a script before it's even green lit. Studios have been trying to predict what will be hits since the early part of the 20th century and they still get it wrong. This just seems to draw a correlation to marketing of a movie or buzz to dollars spent at box office, which is the magic behind the studio industry.

They are measuring buzz/some sort of interest. If buzz is not there, you could pump more money into marketing and see if it raises the buzz. That's probably the only thing you can do that late into the game, marketing spending.

It's not hard to predict a movie's box office right before it comes out

Well, there's the Netflix prize[0] which tells otherwise

Predicting movie appreciation is actually very hard to solve. One of the infamous examples is the "Napoleon dynamite problem"[1]

[0] http://en.wikipedia.org/wiki/Netflix_Prize

[1] http://www.techdirt.com/articles/20081123/1212542927.shtml

The Netflix prize results show that recommending movies that an individual viewer is likely to enjoy is, in certain cases, a hard problem. That doesn't say much about the difficulty of predicting a movie's aggregate popularity over a large population.

IIRC the netflix prize is about making predictions for individual users based on their history & interests, which is orders of magnitude harder than making a prediction for the average over all users, which is all that you need for box office predictions.

Reply on news.ycombinator.com