Silence Looks Like Everything Is Fine
Working on devices that sit in fields and drains changed how I think about every app I build.
- Published
- Read time
- 6 min read
Alongside software, I build for real hardware, like a soil monitor that waters crops and a sensor that watches the water level in street drains. Hardware keeps teaching me the same lesson, and it teaches it faster than any software project has.
Things will break, so decide what happens when they do
With a website, a failure usually means an error message. With a water pump, a lost connection could mean a flooded field. So the rules lived on the server, not on the phone: the pump can only run for so long, must rest between runs, has a daily limit, and every command expires. If the signal drops at the worst moment, nothing dangerous keeps running.
Design the failure first. The happy path is the easy part.
A quiet sensor is not a safe sensor
With the drain sensor, the scariest case was not a full drain. It was a broken sensor that stopped reporting, because silence looks exactly like "everything is fine". So the system watches the sensors themselves, and treats a sensor that goes quiet as a problem to report, not a reason to relax.
What I carry back to software
- Ask "what if this stops halfway?" for every important action
- Treat missing data as a state to show, not an absence to ignore
- Put safety rules where they cannot be skipped, on the server
- Tell people by text message when it matters, because nobody watches a dashboard all day
Every app has its own version of the pump and the silent sensor. Hardware just makes them impossible to ignore.
