Good Enough and the Tyranny of Perfection in Science
In science, perfect is attractive. We talk about errors and deviations, but we always want to know that our equipment is working ideally. We spend our younger years in classes where we study concepts in a perfect world, and then we have the shock of going to a lab where things are decidedly messy. We may have an experiment held together with cable ties and electrical tape, but we want peak power out of it all the time.
When I was in graduate school, one of the banes of my existence was the frequency doubling cavity that gave me one of the laser colors I needed. I would take all my pump laser power, put it into this cavity, and get out a tenth of that in the color that I wanted. And that was on a good day. I probably spent at least a third of my graduate life fighting with the doubler power. There was always an elusive benchmark that if I could hit it, that would be enough power. In order to actually graduate, I had to step back and look at what I had, maybe do a couple rough calculations, and realize that I didn’t need optimal power, I just needed it to be good enough.
“Good enough” has become my rallying cry since then. Good enough means that you run your experiment as soon as it’s good enough to get a result. Because the result doesn’t really care if you had optimal power. The result cares that it was good enough to see an effect. And let’s face it, we’re probably not going to use the first data we take. We’re going to use that initial data to guide our experiments, refine it. So maybe along the way, we’ll see we need a smidge more power, an ounce more stability, and few more atoms in our trap. And maybe one day that will add up to “well, we need to overhaul the experiment.” But, in general, it’s best to save the obsessive perfectionism for those times when you’re waiting for your paper to be accepted for publication and have some down time to make big repairs. Which doesn’t actually happen that often.
Because let’s be clear, it’s really easy to get bogged down in the details. And the details are sometimes not even that interesting. Sometimes they’re even a bit depressing. Sometimes you’re turning two knobs back and forth, seemingly doing the same thing, but somehow getting a slow, steady improvement. And it’s boring and doesn’t have a lot to do with science. Or maybe you’re debugging code because you forgot to capitalize that ancient subroutine you called in the 268th line and also there might be a semicolon missing somewhere. It’s not adding to our understanding of the beauty of the universe, but it has to be done because things don’t work without it.
So why not let the things that don’t prevent forward momentum go? If you’re making steady forward progress, maybe it doesn’t matter that you’re operating on the edge of usefulness, just this once. It helps you keep sight of the big picture, of why you got into science in the first place. Because that’s important. Nothing kills dreams quicker than losing sight of them. And it’s especially important in graduate school because the tunnel gets really dark before you see the light at the end. So don’t linger in the dark places any longer than you have to, and listen to the adage: “Don’t let the perfect be the enemy of the good.”