Cahiers du software

Yesterday I posted a dismissive tweet about John Gruber’s Pulse presentation at Macworld Expo, the video of which is currently on display at Macworld’s site. I’d stopped the video after 5 minutes because it was clear that I wouldn’t be rewarded watching further. Today, I felt guilty and watched the whole thing. It seemed even worse.

The talk is called “The Auteur theory of design,” and the gist of it is this: In collaborative work, leaders with good taste raise the quality of the work done by their staff, while leaders with poor taste lower the quality of the work done by their staff. This is hardly earth-shattering news to anyone who’s worked in an organization of more than a dozen or so people—or even anyone who’s read about working for an organization of more than a dozen or so people—but maybe the people at Macworld Expo lead sheltered lives.

The audience for the talk are the Apple faithful, whom you would expect to view this thesis as an explanation of why Apple, under the direction of the tasteful Steve Jobs, produces exquisite products, while Microsoft, led by the tasteless Gates and Ballmer, produces shit. Worried, apparently, that his audience is too dense to put two and two together, Gruber engages in a tedious and predictable insult of Microsoft near the beginning of the talk.

Much of the talk is spent discussing film and, in particular, the auteur theory of film criticism. He uses Alfred Hitchcock as his example of an auteur and tells the story of how Hitchcock was so meticulous in his planning and so economical in his shooting that the studio couldn’t recut his films; they had to be released pretty much as he wanted. If you’ve taken a film course in college or read much about filmmaking, you’ve probably heard this story, too1. In the one bit of the talk I enjoyed, Gruber refers to this as Hitchcock’s “hack,” his way of preventing the studio executives from ruining his work. Tying the subject matter—which had by this time strayed pretty far from the usual Macworld topics—back to the audience with the word “hack” is pretty clever.

But the hack story completely undercuts Gruber’s thesis. Hitchcock—and all the usual suspects in auteur theory, like John Ford, Howard Hawks, and Orson Welles—were not really in charge. They were supremely talented underlings who got their vision out onto film and into the world despite the execrable taste of the studio executives. They worked for Steve Ballmer, not Steve Jobs. And yet we have North by Northwest and Stagecoach and Bringing up Baby and Citizen Kane.

So the talk was a bust. Poorly delivered, which is excusable2, and self-contradictory, which isn’t. But I do think relating auteur theory to software design is an idea worth pursuing. Auteur theory is really about great directors leaving their mark on the films they make. That mark may be a theme the director explores, or it may be a visual vocabulary he uses3, but it’s something that identifies his films as his. Are there things in software—things visible to the user, not libraries or the choice of variable names—that identify the program’s designer?

I think there are. To cite a couple of old Mac examples, MacPaint and HyperCard bear a resemblance in both the way they look and the way they work that must be due to Bill Atkinson’s sensibilities. I’m sure there are things that distinguish the works of Wil Shipley, Gus Mueller, and Daniel Jalkut. Oddly enough, John Gruber, with his eye for user interface design, may be the best person to develop a real auteur theory for software.


  1. That the story may not be true need not concern us here. What’s important is that Gruber believes it. 

  2. Not everyone is a great public speaker. The professor whose lectures taught me the most and had the deepest influence on me had a terrible stutter and was constantly interrupting himself. 

  3. Hitchcock left both thematic and stylistic marks, which is why he’s the canonical auteur.