In the early 2000s, after building next-generation user interface systems at the MIT Media Lab and before founding my UI business, I had a brief stint as a science and technology ‘designer’ for a handful of big-budget films. Iron Man, Aeon Flux, and Ang Lee’s Hulk were all steep learning inclines, but the one that stands out as pure vertical (and ecstatic) ascent is also the earliest: Steven Spielberg’s Minority Report.
Making a movie and running a commercial business are clearly distinct undertakings, but the latter can learn more from the former than is supposed. I’d like to let three interrelated ideas radiate from the experience of making (and watching) Minority Report.
There’s a moment with Spielberg filming the main character hurling himself off a fifteen-story roof to grab a jetpack-borne antagonist hovering halfway down. You can still find it online in behind-the-scenes footage: Tom Cruise, doing his own wire stunts, takes the ‘high fall’ and grapples furiously with another aerial actor. The director watches through the camera’s eye in real time and then says “Cut. That’s all I need. Thank you.”
That day of extremely technical filmmaking cost at least $1m to prep and execute. How can the director know, given all that complexity, that after just one take of this shot he’s got exactly what he needs? I remember dozens of moments like that: no hesitation, no anxiety that something’s been missed or left undone. It’s galvanizing to witness such efficient, precise, and accurate decisiveness.
Decision-making involves two levels of mostly subconscious cognitive work. The lower level models the situation itself: the larger context into which the decision and its consequences fits, some analogy with earlier experiences, and simulated outcomes. Here was Spielberg probably evaluating against a kind of mental storyboard: what the shot needs to show and how it needs to look, and how it’ll fit into the editing rhythm of the scene.
The higher level models, well, the modeling process itself: how accurate does the lower-level model capture aspects of reality? Are they salient pieces of reality? And are there just enough of them to support high-quality decisions?
I painfully recall trying to make certain crucial decisions early on as a leader through purely conscious, brute-force analysis (i.e. lower-level modeling alone). The panic and paralysis of feeling — incorrectly! — that there…