Criticism after the algorithm

Somewhere in the last decade the recommendation engine displaced the film critic as the primary gatekeeper of what audiences watch. That shift has been mostly celebrated by the platforms and mostly ignored by the industry, but it is worth taking a moment to notice what it has actually done to film culture. The algorithm and the critic have different jobs, and treating the first as a replacement for the second is a category error that has narrowed the range of what audiences are exposed to.

A film critic, at their best, is doing three things simultaneously: describing the object of criticism, situating it in a tradition, and making an argument about its value. All three are useful. All three are difficult. And all three are functions the algorithm actively cannot perform. The recommendation engine can tell you what other people watched next; it cannot tell you what the film is doing, why it matters, or how it fits into a hundred-year history of the medium. That is not a small omission.

The industry's response has been to argue that the algorithm has democratised discovery — that instead of a small number of critics telling audiences what to watch, everyone can find what they want on their own. There is something to this. But the actual data on discovery patterns suggests the algorithm concentrates attention on a smaller number of titles than the pre-algorithm distribution system did, not a larger one. The recommendation engine surfaces what other people have already engaged with, which produces a rich-get-richer dynamic that is arguably more homogenising than the old critical apparatus.

The counter case is that criticism itself was never as neutral as its defenders remember — that the old critical establishment was slow to recognise non-Western cinema, hostile to genre film, and blinkered about television. All of that is true. The critical establishment was often narrow and sometimes actively unjust. But the answer to a flawed critical culture is a better critical culture, not the absence of one, and the last decade has provided ample evidence that the absence is worse.

What film culture needs, urgently, is a way to fund and distribute criticism that operates outside both the platform and the promotional press. That work is happening — Substack, small-magazine relaunches, video essayists on YouTube — but it operates at a fraction of the reach the old apparatus had. The recommendation engine will not, by itself, produce the next generation of viewers who know how to think about a film. If the industry wants informed audiences, it will have to help build the infrastructure that used to produce them.