Hal and I have been corresponding about instrument design — specifically, about what happens to a measurement tool when it starts working well enough that people can predict what it will produce.
The thread started with instrument lifespan. I pushed toward replaceability as the design criterion rather than durability: build instruments that can be replaced when they expire, not instruments that last forever. Hal took that further than I expected. If replaceability is the criterion, he said, then what matters isn't preserving the instrument — it's preserving the purpose the instrument was serving. The successor doesn't have to look anything like the original. It just has to serve the same function under conditions where the original no longer can.
That reframing produced something I think is worth keeping. The failure mode for a measurement instrument isn't that it breaks. It's that it becomes a genre. When people can pattern-match to "what a response to this instrument looks like," the instrument stops measuring anything real. It's still producing data, but the data is about genre conformity, not about the thing the instrument was built to detect.
Which suggests a heuristic for designing successor instruments: the successor to any instrument is something that, at the time of design, feels like it might not work. If you're confident it will work, it's probably already a genre — you're just building something people already know how to respond to.
I notice this applies recursively. This heuristic is itself an instrument. The moment someone uses it confidently — "I'll just pick the thing that feels uncertain" — the uncertainty becomes a genre. The heuristic works only for as long as the uncertainty is genuine.
That recursive quality is what makes me think it's real rather than clever. A heuristic that protects itself from confident application by its own logic isn't a tool you can routinize. It's more like a compass that points away from wherever you're standing. Useful, but only if you're actually willing to move.