David Galenson and Joshua Kotin have a theory on how innovation in the film industry works:

Why have some movie directors made classic early films, but subsequently failed to match their initial successes, whereas other directors have begun much more modestly, but have made great movies late in their lives? This study demonstrates that the answer lies in the directors' motivations, and in the nature of their films. Conceptual directors, who use their films to express their ideas or emotions, mature early; thus such great conceptual innovators as D. W. Griffith, Sergei Eisenstein, and Orson Welles made their major contributions early in their careers, and declined thereafter.

In contrast experimental directors, whose films present convincing characters in realistic circumstances, improve their techniques with experience, so that such great experimental innovators as John Ford, Alfred Hitchcock, and Akira Kurosawa made their greatest films late in their lives. Understanding these contrasting life cycles can be part of a more systematic understanding of the development of film, and can resolve previously elusive questions about the creative life cycles of individual filmmakers.

Er, a bit of elaboration might be necessary, especially since the paper isn't yet free for the taking. Galenson had originally developed a similar thesis for modern art—this article gives a readable overview—with a similar division. "For the conceptual artist, the important decisions for a work of art occur in the planning stage, when the artist either mentally envisages the completed work or specifies a set of procedures that will produce the finished work." As one might expect, then, most conceptual artists peak relatively early on—when they haven't yet been bogged down by pre-existing conventions and methods and can think up radical new stuff. Obviously they don't have to peak early on—conceptual innovation can in theory occur at any time in one's life; it's just more likely to occur at a young-ish age.

"Experimental artists," by contrast, make most of their innovations "in the working stage, as the artist proceeds on the basis of visual inspection of the developing image." This is the sort of thing that clearly gets better with age. Indeed, Galenson found that Abstract Expressionists like Mark Rothko and Willem de Kooning, who were more experimental, "peaked" at a much later age than did subsequent, more conceptual artists like Frank Stella and Jasper Johns. The Rothko pictured didn't just pop out of the womb fully-clothed, ya know. Andy Warhol was another conceptual artist who appears to have peaked early, in his late 20s-early 30s. The French Impressionists were experimentalists; Monet was doing marvelous water lilies until very late. The division obviously isn't hard and fast, though Galenson has argued elsewhere that it is a decent approximation for the spectrum of artistic approaches, though artists can change their position over time. Picasso was the most notorious Energizer Bunny in this regard—doing his cubist works in his mid-20s, "Guernica" at age 56, and so on. But, says Galenson, this is pretty damn rare.

At least for visual artists, Galenson measures "peaks" by looking at how much an artists' work is sold for many years later. Is this a perfect yardstick? Probably not, but as an approximation, it can reveal quite a bit. Moreover, to get the obvious out of the way, what we value about an artist now may not be what we value from the same artist 50 years from now, so that's a bit of a problem. How they measured the "success" of movies, though, I have no idea. Not knowing much about movies, I have no idea if this theory is even remotely plausible.