I've been thinking about Che. Not the man, nor exactly Steven Soderbergh's film of the man's life, but the experience of watching a film that may or not be the finished product.

All that was known about the status of Soderbergh's long-awaited film on the eve of its Cannes debut was that it would be divided into two sections (originally to be called The Argentine and Guerrilla, later simply Part One and Part Two); that they would be screened continuously, with a combined running time of just under four-and-a-half hours; and that Soderbergh was rumoured to be still editing the film (at the subsequent press conference, he merely said that "the process of editing was intense").

However, neither the festival nor the film-makers were presenting this as a work in progress - unlike Francis Ford Coppola's original three-hour version of Apocalypse Now, which picked up an ex aequo Palme d'Or at Cannes in 1979.

Still, the suspicion that we would never again see this film in exactly this form was very much in the air - and though it's not up there with Apocalypse Now, experiences like this always have a certain tell-the-grandchildren cachet.

But here's where it gets interesting. I came away from a gruelling but strangely invigorating 268 minutes feeling that Che was left deliberately unpolished; that its unpackaged package - which extended to a complete lack of opening or closing credits - was about more than close-to-the-line delivery deadlines. Soderbergh's comment that he envisaged the release as "a sort of roadshow engagement, no credits... (just) a printed programme that comes with the movie" suggests that a whiff of the rough-cut was part of his vision for the film.

The director has explained the doubling up of what was originally supposed to be a single movie by saying that "if you're going to have context, then it's just going to have to be a certain size". But one of the fascinating (or, for several critics, frustrating) things about Che is precisely its lack of context.

It would have been relatively easy to write a two-hour biopic that filled in Che's back story and made more of the arc of the revolutionary leader's triumphant guerrilla beginnings, comfortable but restless second-act life as a minister in Castro's government in Havana and final, doomed return to the guerrilla frontline in Bolivia (I'm guessing that Peter Buchman's original script looked a lot like this).

Instead, Soderbergh dwells on the day-to-day challenges of a guerrilla commander's life. Endless details about food supply, training recruits and gun care make the film an excellent manual for anyone planning to take to the hills for some armed resistance.

A deliberate paradox

As in Gillo Pontecorvo's The Battle Of Algiers, guerrilla warfare is presented via a succession of near-documentary episodes with the morality and most of the obvious emotion leached out of them. And yet, in Che, the skeleton of the Hollywood rise-and-fall parable can still be made out. The film thus becomes a sort of deliberate paradox, a heroic failure - like its subject.

In the process, and in defiance of the pressures his reported $60m-plus budget must have exerted, it seems not entirely fanciful to suggest the director was fumbling towards a Zeitgeisty proposition about how the unfinished can be a kind of finish. (It may also be a sign of the times that digital technology - specifically the brand-new Red One 4K high-definition camera used in the shoot - facilitated the gamble).

The cult of the "non finito" has long been recognised in art: think of Michelangelo's Slaves, sculptures in which the emergence of the finished form from the marble block is left in queasy suspension. Though a few art historians have attempted to assert the contrary, I don't believe that Michelangelo intended to leave his Slaves incomplete any more than Soderbergh really intended for Che not quite to work as a commercial feature film. (Neither am I suggesting that Soderbergh is a Michelangelo; to my knowledge, the Florentine genius never had an Ocean's 11/12/13 phase).

But in both cases, it's the way the work 'fails' by the dominant standards of the artform that makes it interesting - and turns it into something that feels a lot like success.