Streaming's algorithm blues

The most powerful person in television commissioning is not a person. It is an algorithm, or rather a set of them, tuned to reward the shows that produce the highest completion rates in the first two episodes. That optimization has slowly rewritten what gets greenlit, and the pattern that has emerged — thrillers with strong first-episode hooks, family dramas with high-frequency plot beats, procedurals with a clear crime-of-the-week structure — reads back at the audience as a shrinking range of possibility.

Netflix has been the most public about the strategy. Their internal metric of choice is completion rate, not audience appreciation, and their retention modeling has produced a slate weighted heavily toward shows that hold viewers for the first hundred and twenty minutes. That is a rational strategy for a subscription business. It is also, in aggregate, a strategy that penalises any show whose pleasures take more than one episode to reveal.

Compare this to how HBO commissioned during the John Landgraf era at FX: the network was willing to accept a slow-building first season if the show promised depth downstream. Some of those bets paid off spectacularly — Fargo, The Americans, Atlanta — and some of them did not, but the range of what was commissioned was wider. Streaming has narrowed the range while widening the volume, and it is not obvious that is the right trade.

The counter case is that audiences do have shorter attention spans, that first-two-episodes triage is a rational response to actual behavior, and that the shows that get made under the current regime are, by and large, competent and enjoyable. Fair enough. But the algorithm cannot commission a Twin Peaks, or a Better Call Saul, or a Deadwood, because none of those shows would clear the completion-rate hurdle in their first two hours. That is not a flattering statement about where we are.

What television needs, more than another prestige drama, is a distribution mechanism that can afford to be patient. The BBC has one, occasionally. Apple has, on a limited slate, been willing to be one. HBO still is. The rest of the industry has effectively decided that the algorithm's taste is the taste that matters, and the shows we get reflect exactly that.