This afternoon, I wound up more fried from project-switching than I’d been in a while, so I sat down with Aardvark’d:12 Weeks With Geeks, which arrived in my mailbox this week. A documentary about a group of interns developing Copilot, a new software product, at New York’s Fog Creek Software — would that be like watching paint dry? or Revenge of the Nerds V? or Apollo 13 Takes Manhattan? or Deliverance II: The Golden Master? Just the sort of question I was equipped to answer today.
The film is definitely more interesting in concept than actuality — any one of the essays from Fog Creek CEO Joel Spolsky’s blog, Joel On Software, will tell you more about software development, or Joel or Fog Creek, than the movie will. (In fact, Joel’s relative distance is one of the film’s biggest faults, since he is the most engaging interviewee in it. Maybe, as the company financed the movie, this was a deliberate choice, a sort of modesty, or a desire to make a movie more about the interns than him, but I wound up wishing for more.) As a result, the film isn’t really for fans of Joel On Software and its author’s point of view on development. The opening title sequence shows views of New York reminiscent of or directly mirroring Joel’s photography, which he often uses on his blog, but you don’t hear about photography as an interest of his, or for that matter anyone else’s interests outside of work. While I’m sure Joel had the interns coding on a whiteboard during their interviews, we don’t see that, nor any of the other Joel On Software tactics that might make this software project different from other software projects. The topic isn’t discussed, but it’s interesting to see Joel’s “bionic office” in everyday use (though the stars of the film, the interns, apparently don’t rate private offices). You get some good hints and clues about what makes software development interesting, though, and a few of the sequences are either directly engaging as documentary or useful as engineering practice.
As a film, it’s a clumsy and poorly-crafted piece. The sound, lighting, editing, and music are all anywhere from uneven to terrible. Some of the interview segments are very well shot and lit, but other sequences feel more “home movie” than vérité. The worst directorial decision is including three re-creation segments (actually, two re-creations and one total fiction) that, if intended to provide light moments or synthetic drama, fall completely flat instead. (If anything they make it seem like the filmmakers are laughing at the audience, trying to see what they can get away with.) A very heavy-handed interlude, showing a Fog Creek developer growing tomatoes on the balcony in time with the growth of Copilot, made me want to throw tomatoes.
The best sequences for software developers show the product becoming a product live on screen. A usability test segment winds up making the product real; we really see it for the first time as the test subjects fumble through it. The engineers’ conclusion — “I thought we’d see the bugs we already knew about, but we wound up finding a bunch of new problems” — is the same reaction every engineer has when first going through a good usability test, but still brings the product development (however briefly) to life. Of course finding a good domain name is enormously frustrating; of course the product crashes during the first demo for Joel; of course the trade show doesn’t bear enough fruit; how many interns, though, see all of that in a summer? A betting pool for how long it will take for the first paid order to come in after launch is another great moment (though, unless I missed it, they don’t tell you who won nor how long it took!).
For the general audience, the rewards are much more thin — this probably won’t tell a non-developer much about making software. We do meet the interns and get to know about their decision to come to Fog Creek; one turns down a competing offer from Microsoft in order to join the Copilot team. Unfortunately, their character development from there is pretty homogenous. With the exception of Ben, cast as the argumentative team member, it’s hard to tell what is different about the interns, their contributions, or their experiences at Fog Creek. We do hear that one of them (I won’t spoil the surprise by saying who) gets a full-time job offer, but we don’t see why, nor why he is any different or better than the others. One great set of interviews talks about the relationship between the developers, their work, and the Joel On Software Forums, which all of the characters read and naturally react to. You almost hope that some of the more vituperative forum posters wind up watching the film and — could it be? — learning from it. Perhaps most lacking, though, we don’t have a sense of how the team grew or changed through the experience. They came, they built, they left — and that’s that.
A short trip up to Y Combinator tries to draw a contrast between Paul Graham’s and Joel’s approach to their summer programs, and fails to elucidate what it might be — but the contrast is interesting to consider nonetheless. The Y Combinator partners are explicitly trying to cultivate the next generation of entrepreneurs; Joel explicitly is not, but instead wants to catch the best recruits during their only trip to the job market. Joel, after all, already has a company to run, and isn’t out to create more founders. As Y Combinator’s Jessica Livingston (like Paul, an O’Reilly author) points out, though, many of these insecure, unformed interns are likely much the same as today’s successes were at that age. Just as Joel’s experience at Microsoft gave him fodder for essays and business strategies for years to come, these interns likely got plenty of the same good training.
There is something brilliant about the film, after all of this: like Joel On Software, and in a way like Copilot itself, the film connects you with the people behind the product. Seeing the interns going through development, getting the product to launch, and taking pride in their work, you can’t help but be more interested in what they’ve built. It’s very hard to get me to spend an hour and a quarter learning about a product — I usually consider even thirty seconds an absurd intrusion — and yet I watched the whole thing — even paid for it — and came away more interested in Copilot than before. For all of its faults, the film succeeds in making the product more real and interesting than any of the standard marketing tactics Joel could have used. If that was the purpose, the film succeeds at it.