Take it from the computer nerds - it's better not to be cute

Predicting what people want - what might delight them or surprise them or attract them - is what everyone in entertainment is supposed to be good at. And we're not.

The movie The Social Network, which tells the tortured, complicated story of the early days of Facebook, is either completely accurate or a pack of lies, depending on whom you ask.

If you're Mark Zuckerberg, the majority shareholder of the behemoth website and the youngest billionaire alive, you might quibble with your depiction in the movie. The film-version Mr Zuckerberg cheats and swindles his film-version friends out of sizable stakes in the business. The real-life Mr Zuckerberg maintains he did no such thing.

On one thing, though, they all basically agree: Facebook, the fastest-growing site on earth and a worldwide cultural phenomenon, was initially created to help computer nerds meet girls.

Matchmaking and romance are, in fact, two big things that drive a lot of tech geeks to impressive feats of computer engineering. Figuring out and predicting the rules of matchmaking and personal attraction - something tech types call "predictive algorithms" - combines two areas that geeks worldwide obsess over: complex mathematics and the mystery of women.

The matchmaking site OKCupid.com, for instance, is run by the nerdiest of nerds. So nerdy, in fact, that they collect the huge amount of data their site generates - clicks, likes, messages between members, attractiveness ratings, etc - and that swirl of data gets crunched and processed and discussed in their company's blog. The results are never less than fascinating.

Recently, for instance, they took a deep dive into the different patterns of "messaging" their female members receive from interested males. They wanted to know, with the desperation typical of guys who are great at maths, which ladies get noticed more often and receive the most e-mail messages from guys, and why.

Let me stipulate that this is all barbaric and disgusting. But also interesting, in a creepy way.

The results were counterintuitive. What they discovered is that, statistically, the most popular women on the site aren't the ones with the highest attractiveness ratings. They're the ones with the most disparate.

For instance: a lady whose attractiveness ratings gets lots of ones and 10s gets more attention than a lady with an average rating of 7. Better, apparently, to have some guys think you're hot and some guys think you're totally not, than have most guys think you're cute.

Predictive algorithms are being crunched and tested all over the web, by merchants and news sites and banks and, yes, matchmaking sites, because if you can figure out what your customer wants, you can make a lot of money. Even if your customer doesn't really know what he wants, and doesn't know that he doesn't know. The algorithm, if it's a good one, will tell him.

Predicting what people want - what might delight them or surprise them or attract them - is what everyone in entertainment is supposed to be good at. And we're not.

Developing a hit TV show or blockbuster movie is a lot like successful matchmaking. But we still do it, here in Hollywood, the old fashioned way. The pre-nerd way.

We spend lots of money on focus groups and test screenings. We analyse ratings and survey answers, and adjust our products accordingly. But our mistake is that we're aiming to get a movie or a TV show into the broadly appealing range, the "cute" range.

That made sense back when the TV audience was huge and monolithic and measured in the tens of millions rather than the ones of millions, and when the movie audience had to drive to the theatre and wait in line to see a picture. But now, the TV audience has shrunk and split into hundreds of channels, and movie audiences can see the most recent releases sitting comfortably at home. Going for the broad middle of the range means that a lot of Hollywood product just gets ignored, like a cute girl on a matchmaking site.

According to the nerd number crunchers, television networks and movie studios would be better off going for more disparate audience reactions. A lot of people who hate a show or a film and a lot of people who love a show or a film are more valuable - and, according to OKCupid.com, probably more financially lucrative - than a whole lot of people who just think a show or film is cute.

There's no money in cute, according to the dating site nerds, and they have the data to prove it. Maybe the they should move to Hollywood. They couldn't do any worse.

Rob Long is a writer and producer in Hollywood