Design for Unhappy Path

I have not checked the status of a smart home landscape for some time. Wink went down a while ago, Revolv left their users stranded. Now I'm hearing SmartThings hubs playing random games with their users. At Silvair we had an ambition to develop a similar platform back in 2012. And we did deliver on a small scale having a trusted group of satisfied users. While the F&Home system is not running the most popular protocols, it has one simple advantage: reliability. I installed it in my home 4 years ago and it has been running now uninterrupted for almost 1000 days. No blackouts. No random behavior. Just doing the job.

There is this fundamental difference between what it takes to build a system that works sometimes and a system that works always. An order of magnitude in time / money / effort. Or two orders of magnitude. It is almost like making software that behaves like hardware - rock solid and dependable.

Happy path is easy and you can demonstrate it to the market, customers and investors quickly. Unhappy path is tedious. You have to keep nailing down all the corner cases. Handle all possible error conditions. Especially when the goal is a system that runs unsupervised. Which means it has to have an ability to recover from any fault condition. E.g. what if a battery in a temperature sensor dies? Notify a user of course, but also keep the HVAC running based on historical data or pre-defined patterns. A smart home system should really have several operational laws, like an airplane. Normal law when everything works. An alternate law, when a central logic controller is unreachable, should be distributed to edge nodes. Which for example means your circadian light mood cycle will not kick in when the "brain" is down, but an occupancy sensor will still be able to turn on lights on a stairway after dark.

Building reliable, dependable systems is hard, expensive, and may be tedious. But it is the only way to succeed. To reach a scale. If a product like a smart automation controller does not have 5 nines of reliability, the number of failures will be enough to kill the brand. Smart home early adopters are bored and fed up now with their "look how cool it is" gadgets. They want to see values they can get for their investment in smart products. The smart home industry has to start delivering dependable, reliable products. Otherwise the adoption rate will never pick up.