Deconstructive criticism (or: The Pixar model)

Getting a Ph. D. is hard, you guys.

As an undergraduate, I could be a guy who knew stuff. Classes gave me material to learn, and as long as I mastered that material I could confidently expect success. It wasn't easy, necessarily—learning subtle and unfamiliar concepts demands effort—but it was clearly defined; there was never any mystery in how to succeed. Classes had well-encapsulated curricula, insulating me from my ignorance, distracting me with newfound knowledge while keeping me from knowing what I didn't know.

Graduate school is exactly the opposite. Your first task is to become familiar with the literature in your discipline, which is no small feat. Even within a single discipline there are more ideas than you can internalize, more papers than you can ever possibly read. And since graduate research is highly individualized, there can be no master syllabus of necessary and sufficient papers. You must therefore decide for yourself what to learn. Rather than being led through a carefully-crafted curriculum, you get a shocking, unstructured look at your ignorance—an ignorance you can only selectively remedy. And, for good measure, there's no one to administer a quiz at the end to make sure you understand it correctly. All of this means that even after reading a bunch of papers it's hard to have confidence that you know enough to carry out successful research.

It's even harder when doing your own research. This isn't just because the problems to be solved are difficult. In fact, I bet most researchers would agree that actually doing the work is the comparatively easy part of graduate school. Instead, the hard part is knowing what to work on. The great struggle of a scientific Ph. D. is finding a problem that is simultaneously important, unsolved, and solvable. And there's no formula—at least, none that works—for finding such a problem. It's relatively straightforward (again, not easy, but usually straightforward) to sit down and start doing a little math. It's hard to know whether or not that math is going to lead to anything important. The path to Ph. D. is a ragged serpentine, full of blind alleys and unintended excursions.

And you have to do it alone. One of the most terrifying realizations of graduate school is that your advisor does not have the answers. He doesn't fully understand what you're working on. He doesn't know whether or not that work will yield publishable results. Hell, he usually doesn't even know whether or not there are mistakes in your work. He doesn't have time to hover over your shoulder and micromanage your progress. He can give you valuable advice from his experienced (but information-poor) perspective, but no one knows your research as well as you.

This shift the burden of criticism back onto you. You have to look through your own work, scour it for flaws and weaknesses, and decide whether or not you're doing fruitful research. In part this is a healthy exercise: anyone who hopes eventually to head his own research program needs to learn to discern good work from bad. And, in any case, healthy self-criticism is an important part of being a well-adjusted member of society.

But let's be honest: self-criticism is hard. Every talented person wants to believe he is talented, secretly fears he is not, and furtively goes about proving to himself and others that his fears are unjustified. Graduate students routinely suffer crises of confidence, and even honest self-criticism can aggravate the symptoms. Yet academic research is sufficiently demanding that confidence—perhaps even unreasoning, absurd overconfidence—is an essential component of success. It's nearly impossible to fruitfully chase down an idea while tending nagging fears of making mistakes or wasting time down blind alleys.

So, as a researcher you face a dilemma similar to that of an artist. Pay too much heed to your internal critic, and you end up paralyzed by self-reflection. Ignore it entirely, and you produce flawed, unimportant, or otherwise self-indulgent output. Every artist, composer, scientist, and writer faces this dilemma, each walking the knife-edge in his own way.

But some walk it really, really, well. I recently came across an article describing the creative process at Pixar. (If you don't like Pixar films, then you're either too cool, have a heart made of stone, or otherwise suffer from crippling personality flaws. Seek professional help.) Pixar has consistently output innovative, high-quality films, and they achieve that success through interesting means. Every morning, animators and directors gather to examine the work completed the previous day and pick it apart in excruciating detail—deciding what works, what doesn't, and how it should be improved. The animators then take the criticism back to the drawing room and implement these changes. Rinse and repeat until you have a masterpiece on your hands.

In one sense it's not terribly surprising that Pixar's formula is successful. Take a bunch of creative and talented people, create a highly collaborative work environment, and you get phenomenal success! How... non-obvious? But their story struck me in a non-obvious way. How much easier is it to silence the self-critic when you know that a group of your smartest, most talented friends is going to go over your work tomorrow, looking for flaws you overlooked? How much easier is it to be freer, more daring, and more innovative when you know that other talented people are going to keep you from going too far off the deep end?

As someone who suffers from an overactive internal critic, it sounds liberating to me. It's not that I want to avoid the responsibility of evaluating my own work but that I could do it much more effectively with constant, systematic oversight from respected, talented colleagues. As I look forward to the (hopeful) future of running my own research group, the Pixar model appeals to me.

Of course, implementing it presents serious challenges. In describing the model I've tacitly assumed that you have a group of talented people, each of whom is secure enough in his talents and secure enough in his colleagues' that he will not only accept criticism without feeling attacked but also give criticism without attacking. That's a work environment that must be difficult to create and even more difficult to maintain—especially among smart, successful people whose overdeveloped superegos have gotten them far in life. It probably requires a strong sense of shared objective that's hard to foster among creative, independent thinkers.

I don't know what the answer is, but Pixar's success gives me optimism. They've proven that such an environment is possible to create and to maintain for well over a decade. Can't I hope to achieve a creative environment that crusty ol' Steve Jobs has pulled off?

[PS: Made on a Mac!]