Automation is arbitration plumbing
A sensor reports, a hub decides, an actuator acts. From inside the house, each of those three roles fails silently in its own way.
A house full of automation looks like magic from the doorway. Lights adjust, a valve shuts, a notification lands before you've noticed anything wrong. From inside, it's arbitration plumbing: sensors that report, a hub that decides, actuators that act — and every one of those three roles fails differently.
The chain has three roles and I try to keep them separate in my head, because they fail in different places. A sensor's only job is to report a condition — moisture, motion, a door left open. A hub's job is to evaluate a rule against that report. An actuator's job is to act on the result: a valve, a lock, a light. My leak sensor sits under the kitchen sink on Z-Wave. When it detects moisture, the hub checks the rule and fires downstream — a notification, sometimes a shutoff. Three roles, three distinct ways for the chain to break, and only one of them looks anything like a bug.
The actual engineering problem isn't any single sensor or actuator. It's that Z-Wave, Zigbee, Hue, and plain Wi-Fi don't talk to each other natively. The hub's real job is translation — getting a Zigbee motion sensor and a Z-Wave water sensor and a Wi-Fi plug to act on the same rule without knowing the others exist. My hub is a Home Assistant Yellow, run local, no cloud dependency, tracking around 1,280 entities across the house. It's worth more for what it understands across those protocols than for anything it directly controls.
The failure I keep coming back to involves that same leak sensor. Diagnostics show its battery at 20 percent. It hasn't missed a report yet, which is exactly the problem — a dying sensor doesn't announce a threshold crossing. It just goes quiet one day, with no error and no flag, and the hub has no way to distinguish "no leak" from "no sensor." The reliability of the whole chain is bounded by whichever part is cheapest and least watched, not by the hub that cost real money and sits in a closet running fine.
same wiring, same rule — the chain breaks at the first link, silently, with nothing downstream able to tell the difference.
That's the part worth sitting with. Every entity in the chain is a potential quiet failure, and the expensive, visible parts are almost never the ones that go first. It's the cheap sensor under the sink, the one you forget exists until diagnostics happen to surface its battery level. Monitoring the hub tells you the hub is fine. It tells you nothing about the sensor that stopped whispering.
I made a deliberate trade the other direction. I pulled local voice control — speech recognition and speech output — off the hub. It was a real convenience and it was also a resource drain on the same box running the rule evaluation I actually care about. Convenience lost, stability of the core pipeline gained. I'd make that trade again.
The next layer isn't more sensors. It's software that can take a stated intent — going to bed — and turn it into the right sequence across actuators that already exist, in the right order, checking the right conditions first. The plumbing is mostly built. What's missing is the layer that speaks in intent instead of rules.