After a talk with my ever-insightful PhD advisor, I need to amend my model of robust design. I had conceived of robustness as a spectrum with hard-to-break on the end close to fragility and easy-to-fix on the end close to antifragility.
I think the picture is actually two-dimensional and more complicated than ‘hard-to-break’ bad! ‘easy-to-fix’ good!
Instead of a line between ‘hard-to-break’ and ‘easy-to-fix,’ now I’m thinking of robustness like a triangle, with ‘cheap-to-replace’ on the third corner.
Before, I had kind of ignored things that are ‘cheap-to-replace.’ Think of dollar store items – they don’t last very long, but you don’t expect them to and the price reflects that. But ‘cheap-to-replace’ design definitely deserves to be treated separately. In space exploration, chip satellites and objects that are 3-D printed or otherwise built in orbit exemplify ‘cheap-to-replace.’
I also had an overly derisive view of ‘hard-to-break.’ Like so much else, quality robustness has some balance between the different extremes. Something that is easy-to-fix is still terrible if it breaks so often that you’re spending all your time fixing it. Spacecraft don’t currently have any of that balance though. The pendulum is still shoved way over towards ‘hard-to-break,’ but it might begin to swing back.