Does this mean more movies during math class?

There are no rules to writing, despite what grammarians and high school English teachers might tell you. But if there is one important maxim to live by, it’s that effective writing is structure. Learning to write is learning to structure and organize information. Undoubtedly, there is art, there is craft, there is research, verbal flourish and all of those things that might writing wonderful. But effective writing is, at its most basic, well-structured writing, and no matter how much we might want more than that, strong structure is the most necessary element.

Hold on to that thought for a moment, and let me say that my first response to Malcolm Gladwell’s New Yorker piece on a company that uses raters and computer algorithms to analyze and predict the box office success of film scripts was something like that of the skeptical movie executive reported in the article:

The executive seemed to think of the Epagogix team as a small band of Martians who had somehow slipped their U.F.O. past security. “In reality, there are so many circumstances that can affect a movie’s success,” the executive went on. “Maybe the actor or actress has an external problem. Or this great actor, for whatever reason, just fails. You have to fire a director. Or September 11th or some other thing happens. There are many people who have come forward saying they have a way of predicting box-office success, but so far nobody has been able to do it. I think we know something. We just don’t know enough. I still believe in something called that magical thing—talent, the unexpected. The movie god has to shine on you….My first reaction to those guys? Bullshit.”

The story brings up all sorts of questions about the nature of art, commerce, and human experience, far beyond the sort of "neat technology meets a little philosophy" angle that Gladwell takes. Gladwell starts by talking about Platinum Blue, a company with a computer program that analyzes the mathematical signatures of songs and predicts—with supposed 80% accuracy—whether or not the songs will be hits. He then gets into Epagogix’s similar system for the movies. That system uses two raters who’ve analyzed a wide array of narrative elements in a large sampling of films to read scripts and assign values to these elements. These values are fed into a computer, and out pops a predicted box-office result. The article reports only successes, but doesn’t indicate on way or the other whether the program has had any significant failures. Nor does anyone claim that it predicts “good” movies—only movies that will make money.

Attempts at this sort of thing have been made before, and, predictably, they’ve failed. And at first glance, this method seems just as improbable. After all, it suggests that factors external to the narrative don’t matter: stars, directors, marketing, current events, etc. To believe that the cumulative effect of these elements is practically nil seems rather far fetched. After all, does anyone really think that George Lucas’ ludicrous, mediocre Star Wars script would’ve been as effective under a different director? That Blair Witch would’ve made more than a hundred million dollars without its marketing scheme?

Platinum Blue’s musical program seems, initially, like evidence that this sort of thing might work. How can you judge music, after all? It’s totally subjective, or seems to be. But unlike narrative and visual arts, music is just math. You can break down every sound, note, timbre, and rhythm into a mathematical signature pretty easily. Anybody who’s ever taken a music theory course knows that notes and rhythms are connected by numerical relationships. Computers may never be able to produce completed pop songs, but it doesn’t really surprise me that they can analyze those numerical connections.

But it occurs to me that maybe there’s something to this, and in part it’s because it’s a text-only analysis. If, as I believe, writing is structure, then maybe a well-trained analyst really can accurately assess its value on a numerical scale. After all, this is what AP exam test readers do with thousands of high-school essays. If those graders can be trained to give roughly accurate numerical grades to essays, what’s to say it isn’t possible—if obviously somewhat more complicated—to similarly assess the characteristics of a screenplay? Unlike music, text doesn’t have any intrinsic numeric values, but a skilled, trained reader can, I think, assign approximate values with reasonable accuracy.

The problem, still, is that this means that the screenplay is the only thing that matters when tallying up the box office, and that still seems fishy to me. What do we do with movies that have really terrible scripts but are saved by strong direction and performances? Batman Begins and Face/Off, for example, are both pretty putrid, script-wise (the stories are weird and underdeveloped, and the dialog in both is really wretched stuff saved by performances even more brilliant because of the awful source material). But both still ended up strong films—and, importantly, financially successful films—because of great direction, design, and performances.

Somewhat counter intuitively, however, I don’t think that if something like this were to take over it would prove deadly to making interesting, unique, oddball films. Obviously, a system like this would only reinforce Hollywood’s tendency to repeat what’s worked before, as you can’t make statistical predictions about untested material. But my guess is that, even still, a small to mid-sized market would develop for interesting, niche films that would cost a little and make a little—just as has happened with the indie music scene. And maybe with all the money studios would be saving on not making any more failed hundred million dollar behemoths, they’d be able and willing to finance a larger number of smaller, experimental films for the small but passionate group of us who enjoy them.