Craftsmanship and Error Correction
Craftsmanship and Error Correction
Res Extensa #48 :: What Karl Popper teaches us about learning a craft
A couple weeks back I decided to tackle a new project, instead of finishing the 3 or 4 home projects currently in progress. My 6 year old loves to build stuff in the workshop, so some time ago we got him a set of tools he uses all the time to "build" (beat things with hammers, drill holes, and nail things together). As a sideline craftsman myself, I'd love for him to keep this going.
What he needs is a workbench. So I started with a few loose requirements jotted in my sketchbook and opened SketchUp to draw up a rough plan:
Without any hard requirements other than a maximum size and to finish it in 3 weeks, I got to work. To keep it as simple as possible I planned to start with all equal width boards, ripped up from rough construction lumber to keep it cheap. So to Lowe's I went to get material, sorting through the batch of twisted and gnarly 2x8s to find stuff that'd work.
Lots of sawdust and glue later, and we've got rough-cut parts of a bench top, legs, and braces to assemble.
Most things we build don't have an absolutist requirements list, with zero flexibility in implementation. Whether you're making furniture, doing client work, designing a website, or building a software product, most times you have a "job to be done" and a degree of freedom to derive solutions. In this case, I had rough dimensions, an array of joinery methods to put the thing together, and a batch of material with a bit of wiggle room (for those times we don't measure twice before we cut).
Craftsmanship involves knocking down a succession of steps until it feels "done". Cutting, milling, gluing, trimming, assembling, finishing, painting. Each step takes its input from the previous, and might have to correct course from an earlier error. Case in point: when I was ripping up material, some of the workbench pieces ended up bowing out of shape, so I had to mill them all down thinner than my original plan called for. No big deal — the error gets corrected later when cutting the base pieces down for the narrower-than-expected bench top.
I've heard it said before that what separates amateurs from professionals in any craft isn't necessarily the expensive tools or perfect execution or avoiding mistakes. Sure all those things make each element a little bit easier. But it's the ability to recover from mistakes, and even incorporate the mistakes into design changes along the way. As long as a screw-up isn't catastrophic, many times you can adapt and make it look intentional. Turn a bug into a feature.
Error Correction
While this half-improvised project is going on, during a YouTube jaunt I happened back across this conversation with physicist David Deutsch talking about the concept of "error correction", originally popularized by Karl Popper. The crux of it is that all scientific progress — all knowledge — is developed through a process of conjecture and refutation. According to Popper, scientists formulate hypotheses or theories to explain observed phenomena. These hypotheses are not definitively proven but are instead subjected to rigorous testing and critical scrutiny. The process of scientific inquiry, in Popper's view, is not about verifying or confirming hypotheses but about attempting to falsify them. When a hypothesis withstands attempts at falsification, it is tentatively accepted, but it remains open to future challenge and revision. Nothing is off-limits to future improvement. If you go into the archives, I wrote a long post on scientific rationalism and this process of what Jonathan Rauch calls "liberal science".
As I muddle through the process of learning a craft, I notice similarities to how innovation and scientific discovery, rightly understood, happen. I start with a rough idea of the end game, a mental model of the order of operations to step through, with the flexibility along the way to learn as I go the best way to accomplish the next step in the process. It's like I've got an idea in mind of the path of stepping stones to take, but I need to be open to adapting that as minor (or major!) mistakes close off certain options.
Craft involves proposing a design (akin to a scientific hypothesis), executing it (testing), and then observing the outcome. If a mistake occurs, the we have to assess it critically (similar to falsification in science), decide whether to correct it or integrate it into the design in a way that enhances the overall work. There's an analogy here to the scientific method of adapting theories in response to new evidence or refutations.
Like Popper's error correction, crafts involve a dynamic process of learning and adaptation. Each mistake or unexpected outcome becomes a learning opportunity, contributing to the craftsman's skill and knowledge. Screwups can be incredibly frustrating — I’ve already made several on this simple project. But the only way to know how to refine your technique, or to properly sequence your steps, or to avoid a mistake in step 2 messing you up in step 7 is to screw it up.
Openness to change, and especially an openness to making mistakes are essential to forward progress. Making mistakes means you're doing things. We learn by doing. And undoing. And fixing.
All improvement comes about through persistent feedback. Just adapt and retry.