I thought some more about A.S. Byatt's Still Life and realized the things I don't like about it are the same things I don't like about Ingmar Bergman's film Persona. That is, in the middle of a perfectly absorbing story rooted in human psychology, the author/filmmaker inserts herself/himself to point up the artifice of it all. It feels jarring--and yes, that's the point, to "shock us out of our complacency," but sometimes I like being complacent in the face of a good narrative, OK?
What annoys me even more is that these metafictional elements often win these works of art their greatest praise. One critic wrote of Still Life: "Among the best new novels I've ever read. (...) Byatt, at last, shows that a novel can think about itself without resorting to nervous Borgesian paper-puzzles, tricks, tics" (citation). Woody Allen called "the opening montage of Persona" and "the chutzpah to stop the engrossing story at intervals and let the actors explain to the audience what they are trying to do with their portrayals" "moments of showmanship at its best" (citation). These statements irk me because they imply that a work of art is automatically brilliant if it makes reference to the fact that it is a work of art. Does that mean these critics wish every work of art had a metafictional twist?
A great theme, by itself, is not enough to make a great work of art. For example, film scholars adore Rear Window because it's a metaphor for how movie-watching appeases our secret voyeuristic desires. That's true, but misses the larger point. If Hitchcock had made a movie with the theme "character's voyeurism implicates the audience's voyeurism," but it was poorly paced or badly acted or just plain dull, it would not be a brilliant film in the same way that Rear Window is a brilliant film. This hypothetical movie might be "smart," its theme might provoke discussion, but if its aesthetic qualities weren't good, it would be forgotten two weeks later. Meanwhile, the craft and narrative of the real Rear Window strengthen its point about the seductive lure of watching. It comments on itself, but much more organically than Persona does.
I should perhaps give Bergman the benefit of the doubt: maybe his meta-games were truly shocking, innovative and ingenious at the time (though don't they descend from Brecht's alienation effect?). But it's 80 years after Pirandello and 50 years after Borges: the time has come to either go a lot further in our metafictional experiments, or to give them up. Nowadays, meta gestures, no matter how stale, are just something that artists know will get their work called "smart" and thus "worthy"--cheap shortcuts to critical success. Or, they're an easy way of inserting ideas into a narrative without properly establishing and dramatizing them. For instance, it's hard to write exposition, but these days, you don't even need to try: just baldly state the necessary information, interspersed with jokes about how hard it is to write exposition. Cf. Urinetown, The Little Dog Laughed, Orson's Shadow--and more.
Oh, but there's always a flipside. How can I rail against meta-art when I've already admitted to loving Rear Window--and Adaptation and Borges and Arrested Development besides? Well, it's all subjective, but I feel like the self-referential qualities in these works don't disrupt their rhythms the way that Persona's or Still Life's do. For instance, you're always aware that Adaptation is just a movie and the screenwriter Charlie Kaufman is not the same as the character Charlie Kaufman, but the movie never needs to explicitly state this. These works of art expect you to be smart enough to identify the meta games that underlie their entire structure, rather than alternating passages of straight narrative with passages of the author's metafictional commentary.
In short, a great work of narrative art must, in addition to its themes, have great craftsmanship that will dramatize these themes and make them feel organic to the composition. Only in this way will we create art instead of essays, polemics, or propaganda.