Remote censoring

I don’t want to watch much of the stuff the BBFC classifies, and nor, I suspect, do the people who work for the BBFC. Surely, I thought, we could get a machine to do this?

There won’t be much chance of getting a machine to look at the pictures and work out what’s going on for some time. But what if there were transcripts? Train up a classifier on previously-classified transcripts of films, run the new transcript through the classifier and James Ferman’s your uncle.

Except where do we get the transcripts from? We can’t necessarily rely on the filmmakers, though the scripts might be a good starting point. Having to watch the film and painstakingly note down everything that happens would negate the point of the whole exercise, so we’re left with either crowdsourcing or offshoring.

I don’t want to see eyewatering stills from a potentially R18 film as a CAPTCHA, so the only remaining option is to send the films overseas.

This isn’t going to work, is it?

This entry was posted in simulations. Bookmark the permalink.

2 Responses to Remote censoring

  1. HTFB says:

    It’s an easier task than you’d think. Umberto Eco pointed out forty years ago that since a pornographic film is about the journey to a climax, rather than the climax itself, you can classify them merely by measuring the excessive time the film spends showing the characters in their cars, cleaning, or doing other mundane tasks. A Bayesian filter on entirely machine-readable metrics would surely have no trouble at all.

    Not that I’d want to encourage a world in which cinematographers got used to trying to spoof the Bayesian spam-filters. That no website in the world these days is written without SEO in mind is painful enough.

    • colin says:

      Except there are routes to climax and routes to climax, as we learnt at Southwark Crown Court this week.

Leave a Reply

Your email address will not be published. Required fields are marked *