Saturday, October 14, 2006

Malcolm Gladwell has an interesting article (does he have any other kind?) in this week's New Yorker about Epagogix, a company that uses a neural network AI to predict box office receipts with better-than-studio-executive accuracy. They analyze what -- the star package? The director? The release date? Competition? No, they analyze only the script elements.

Naturally the analysis is most accurate when you're looking at major studio releases that get enough publicity. With indie pictures it's not so accurate because those films need to catch the breaks.

It's surprising, because much as I think screenwriters are undervalued in showbiz, a bad director can screw up a film pretty badly, as can miscasting. And while a bad director will usually insist on screwing up the script before he shoots, he doesn't always. Toys would be a good example -- a very funny script misdirected as a poignant drama by Barry Levinson.

If Epagogix is as good as Gladwell says, it doesn't mean studios will stop chasing star right away, because that is what they know how to do. It will take one studio adopting a script-based system for greenlighting projects, and creaming the competition by hiring cheaper stars and writing better scripts, before it catches on. Sort of like what happened with the Arizona Diamondbacks Oakland A's, an underfunded baseball team whose managers adopted a new metric for evaluating what players they wanted to hire, and kicked butt. [Correction thanks to Webs].

Studio execs can use their service to decide which direction the script needs to go. As Gladwell points out, the analysis doesn't fix your problem for you -- you still need a screenwriter to fix it -- but it can show where you're at, and what might work fixing it. It's like a similar service that analyzes your song and tells you whether it is in hit territory or not. No computer can generate a hit. But a computer can now tell you if something is not a hit, and give you hints about what aspects of the song are keeping it out of hit territory. It can tell you your bass line isn't working; it can't tell you how to fix it.

What the article doesn't really address, though, is that Epagogix's service only works as well as the people who are coding the script into the computer. In this case, it's two guys from Britain who have watched thousands and thousands of films and thought about them. And their judgments are subjective. So you don't really have a stand-alone computer program. You have intelligent readers enhanced by a neural network. Neither can function without the other. The computer can't do its thing without the readers' subjective judgment. The readers can't duplicate the delicate weighting of dozens of factors that the neural network can do in a blink.

It will be interesting to see how much of our job AI's start to take over. Will we be assimilated?

9:24 AM

3 Comments:

I'm guessing resistance is futile.

By  Kelly J. Crawford, at 1:36 PM  

Not the Diamondbacks. The Diamondbacks won the World Series in 2001 on the backs of high-priced talent like Curt Schilling, Randy Johnson, and Matt Williams.

You're probably thinking of the low-budget Oakland A's, who have bought into alternate theories of player evaluation, as covered in "Moneyball". The current A's have had phenomenal success in terms of wins and making the playoffs, but haven't made it into the World Series yet.

Here are the latest World Series winners with their rank in payroll:

2005 Chicago 13th
2004 Boston 2nd
2003 Florida 25th
2002 Anaheim 15th
2001 Arizona 8th
2000 New York 1st
1999 New York 1st
1998 New York 1st

The only one that stands out is Florida, but that franchise has adopted any novel approach. That was dumb luck and good, durable pitching.

By  Webs, at 5:21 PM  

It will take one studio adopting a script-based system for greenlighting projects, and creaming the competition by hiring cheaper stars and writing better scripts, before it catches on...team whose managers adopted a new metric for evaluating...

This is similar to what I have experienced in the DVD Premiere realm:

Concept is first.
Artwork that sells the concept is second.
Feedback from retailers is third. The script is fourth.
Star power is fifth.

Why?

Primarily because the first three steps are steps that can be controlled and tested before the script and pre-production phases. The last two steps depend on the first three. In the low budget DVD world - no one makes a movie they don't think they can make their money back on relatively quickly. The stars are there to deliver name value and credibility.

In terms of writing, its the "pulp screenwriter's" job to write a script that lives up to the concept, the artwork and the retailers expectations as well as staying within the strict budget parameters.

DVD Premieres in this baseball analogy are "base hits" - individually not spectacular, but taken as a team effort, produce great financial results.

Studios need to quit this "all or nothing" blockbuster mentality that not only limits the number of movies made, but strains the financial underpinnings of a company. Thankfully they are starting to realize this as more and more films are being made in the "$25M or under" category. (Example: Michael Bay's deal with Rogue Pictures). We need more movies with smaller, fiscally-responsible budgets.

By  Cunningham, at 6:24 PM  

Post a Comment

Back to Complications Ensue main blog page.