Mr Smith goes to . . . an ID parade

Looking at the line of usual suspects

Every aspiring screenwriter has woken up with a hangover and an impression of his keyboard embossed on his forehead after trying to come up with something as good as The Usual Suspects. Perversely, the next best thing would be finding a tiny flaw in Christopher McQuarrie's Oscar-winning story, which raised the bar for prentice narratives. "I guess it's about a bunch of criminals who meet in a police line-up," says a nonchalant McQuarrie in the introduction to my tear-stained copy of the script (Faber and Faber, £4.99).

In case you're struggling to reproduce the faultless underworld patois of Guy Ritchie, or are blocked on your period drama at the point where the central character enters through the French windows practising his backhand, here is the shabby consolation of a howler from The Usual Suspects. In the United States, where the film is set, police line-ups are extremely rare, and photographs are the preferred means of identification in fully 90 per cent of cases. So the crooks who are rounded up in McQuarrie's screenplay are actually the unusual suspects.

I discovered this while talking to Tim Valentine, a psychologist who is studying ID parades. He has been evaluating a pilot scheme run by West Yorkshire Police, in which the familiar crocodile of possibles to be inspected by a witness is replaced by a gallery of video images. Rather than combing the streets for volunteers every time identification is an issue, the police only need to see the ringers once. In the manner of stars who don't show up to collect their awards in person, these lookalikes make their contributions on tape.

The Home Office is consulting before backing this method as the new orthodoxy. Provided the video library is large enough, it should become easier to put on show faces that are broadly similar to the suspect's. At present, 50 per cent of parades are cancelled because officers fail to produce the legal quorum of nine (including the suspect himself).

Professor Valentine says that making a selection from a bank of video portraits is "fairer" than the conventional process, or "the swimwear round", as it is known. Statistically speaking, according to the prof, there's a one-in-nine, or 11 per cent, chance of the suspect being picked out in a "perfectly fair" procedure (although the law-abiding citizen might hope for rather better odds than that). In his tests, the figure was 15 per cent for video ID, compared to 25 per cent for the classic line-up. One drawback is that a head-and-shoulders videotape is less effective in suggesting height than a flesh-and-blood subject.

Law enforcement still has a place for the apparent anachronism of the artist's impression. It continues to furnish the manhunt and the court report, as it seems to have done ever since Cain was encouraged to come forward and be eliminated from inquiries. "We now have e-fits, using digitally stored images, but artists are frequently called in at the end, to make them look more human," says Valentine. Although the human factor is still part of ID parades as well, a troupe of regular panellists earn £10-£15 an appearance. Surely there's a film to be made about these men and women who habitually appear as suspects. Pity the obvious title is already taken.