BIOGRAPHIES are generally regarded these days as sexual exposes, and so most of the excitement surrounding Michael Shelden's life of Graham Greene has centred on revelations of the writer as adulterer and homosexual. But the boldest claim in the book, it seems to me, is the biographer's contention that Greene produced no novels of real merit after the 1961 publication of A Burnt-Out Case, its title presented as emblematic and confessional. As Greene kept publishing fiction until 1988, it is effectively suggested that the last three decades of his writing life were a waste of time.

My aim here, though, is not to defend Greene but to consider the question of the shape of literary careers. Shelden has stated with unusual frankness, helped by his subject being dead, the conventional view of a creative oeuvre. This is that, even in the case of a great writer, juvenilia are succeeded by masterpieces and then by fag-ends. According to general critical prejudice, if you take the work of any author you have read exhaustively, and draw a graph - with the dates of publication on the horizontal and your marks out of 10 for enjoyment on the vertical - you will more or less draw a pyramid, rising towards and then falling from the work of the middle years.

As an apparent corrective to this belief, our culture currently offers us the example of Arthur Miller. After 50 years as a stage playwright, he has written in recent years a series of pieces which, particularly in Britain, have received dizzy acclaim. General critical opinion since the Sixties had been that Miller's major work was produced on an amazing eight-year roll between 1947 and 1955, which resulted in All My Sons, Death Of A Salesman, The Crucible and A View From The Bridge. Yet now, at least according to English theatre critics, Miller presents an exception to the pyramid-shaped graph of literary productivity.

We are asked to draw the American playwright's career as a V: early high achievement and a mid-life decline compensated for by a set of brilliant pensioner miniatures. In this example, we see the alternative to Shelden's bleak view of Greene. Miller is held out as a representative of the pleasures of Great Late Work: the theory that, in the final phase of creativity, an artist distils an accumulated weight of philosophy and technique in a last pitch for immortality.

As readers and playgoers, we should perhaps hope that there is some truth in this view, for longevity and medical science ensure that we will suffer or enjoy a glut of late work. Penicillin's effect on literature may have been under-estimated. Jane Austen, dead at 42, never had to face the publisher's polite pressure for a new novella to give a push to the 80th birthday collected edition reprints.

Chekhov provides another instructive example. His literary achievement graph is virtually unique, neither a pyramid nor a V-shape, but something that might be called the Hill Start. His progress is all upwards, culminating in four masterpieces - The Seagull, Uncle Vanya, Three Sisters and The Cherry Orchard - in the years just before his death. But, with modern treatment for TB, he could perhaps have survived beyond 44 for the 'not as good as his early stuff' and 'cold, regurgitated Seagull' reviews which might have given him the classic pyramid output.

Drawing graphs for today's writers, it is possible to find exceptions to this depressing stereotype. After indifferent middle years, the recent novels of Kingsley Amis seem to suggest that, to use an indelicate metaphor, old farts can develop second wind, and that his final graph may show a V - and V-sign to those who wrote him off in the Seventies.

Most long literary careers do, though, conform to the pyramid shape suggested by Shelden for Greene. Despite being initially chilled by Shelden's dismissiveness, I find on reflection that all of the Greene I would want to re-read or recommend to others does come from the middle period. Similarly, despatched to the recent Arthur Miller premieres by the enthusiasm of reviewers, I have been quite unable to comprehend the veneration shown to these thin revisitings of earlier themes.

Even so, I am certain that I will read or see anything produced by my own artistic heroes - Tom Stoppard, John Updike, Stephen Sondheim, Martin Amis - until they or I die. These choices are subjective but anyone reading this could substitute their own. So why do we remain faithful to writers whose work we must know in our hearts is uneven and degenerating as they age?

The first reason, I think, is that critics are football writers, while readers are football fans. It is the job of the soccer correspondent to say that Team A is unlikely to win the championship this season, or even at all, but this will not deter the keen supporter from turning out in the rain on Saturday.

Secondly, complete exposure to the work of any single artist allows for cumulative pleasures, however disappointing the individual product. There is a buff's comfort in spotting that 'he did that 20 years ago in that other one' which far surpasses regret at repetition. The reviewer who objects that Anita Brookner has written the same novel 15 times gets letters from indignant aficionados who delight in this very reliability. Following the entire development of a writer is like watching a child growing up, and you're not going to complain if they're sick on your shoulder occasionally.

Last week, on Broadway, I saw the new Stephen Sondheim musical Passion. Frankly, it wouldn't rank more than eighth among my favourites. Indeed, it lacked the very qualities - lyrical invention, labyrinthine rhyming - that attract me to his work. And yet I greatly enjoyed the evening, for my awareness of its departures and inventions, for the memories it prompted of the earlier works. My pleasure was contextual rather than textual.

The risk of this phenomenon, however, is that it can lead to Queen Mother Syndrome: the artistic equivalent of 'isn't she amazing for 94?' Each new work produced in defiance of time becomes an occasion for thanksgiving, a memorial to all that has gone before, rather than an examination of the evidence of the present. Critics hymning the Late Great Work of Arthur Miller are, in effect, paying further deferred praise to Death Of A Salesman and The Crucible.

This is, brutally, probably also what happened to the later fiction of Graham Greene, and his biographer is right to point it out. But should the 80-year-old Updike or Amis publish their Shopping Lists - and the octogenarian Stoppard or Sondheim musicalise or dramatise them - I will still be first in the queue.