Tim Wu writes for the New Yorker about how Netflix uses a ~70/30 combination of data and human judgment to determine their recommendations and what shows/movies to make.
Over the years, however, I’ve started to wonder whether Netflix’s big decisions are truly as data driven as they are purported to be. The company does have more audience data than nearly anyone else (with the possible exception of YouTube), so it has a reason to emphasize its comparative advantage. But, when I was reporting a story, a couple of years ago, about Netflix’s embrace of fandom over mass culture, I began to sense that their biggest bets always seemed ultimately driven by faith in a particular cult creator, like David Fincher (“House of Cards”), Jenji Leslie Kohan (“Orange is the New Black”), Ricky Gervais (“Derek”), John Fusco (“Marco Polo”), or Mitchell Hurwitz (“Arrested Development”). And, while Netflix does not release its viewership numbers, some of the company’s programming, like “Marco Polo,” hasn’t seemed to generate the same audience excitement as, say, “House of Cards.” In short, I do think that there is a sophisticated algorithm at work here — but I think his name is Ted Sarandos.
I presented Sarandos with this theory at a Sundance panel called “How I Learned to Stop Worrying and Trust the Algorithm,” moderated by Jason Hirschhorn, formerly of MySpace. Sarandos, very agreeably, wobbled a bit. “It is important to know which data to ignore,” he conceded, before saying, at the end, “In practice, its probably a seventy-thirty mix.” But which is the seventy and which is the thirty? “Seventy is the data, and thirty is judgment,” he told me later. Then he paused, and said, “But the thirty needs to be on top, if that makes sense.”
This reminds me of the situation in chess, where cyborg human/computer teams can beat computer- or human-only players in chess, although perhaps for not much longer.
Some of you will know that Average is Over contains an extensive discussion of “freestyle chess,” where humans can use any and all tools available — most of all computers and computer programs — to play the best chess game possible. The book also notes that “man plus computer” is a stronger player than “computer alone,” at least provided the human knows what he is doing. You will find a similar claim from Brynjolfsson and McAfee.
Computer chess expert Kenneth W. Regan has compiled extensive data on this question, and you will see that a striking percentage of the best or most accurate chess games of all time have been played by man-machine pairs. Ken’s explanations are a bit dense for those who don’t already know chess, computer chess, Freestyle and its lingo, but yes that is what he finds, click on the links in his link for confirmation. In this list for instance the Freestyle teams do very very well.
I wonder what the human/cyborg split is at Buzzfeed or Facebook? Or at food companies like McDonald’s or Kraft? Or at Goldman Sachs?