We’ve been talking a bunch of home automation on the Podcast lately, and this week, in the Mailbag segment, a reader asked us about our setups. Neither Kristina nor I are poster children for the home automation movement: she has absolutely no smart anything because she didn’t want her data up in “the cloud”, and I have an entirely local system that’s really nothing more than a bunch of ad-hoc scripts that talk to an MQTT broker, everything fully DIY but held together with metaphorical duct tape. Neither of us are doing it right, but we’re doing it wrong in interestingly different ways.
Kristina thought, probably because of the range of commercial devices out there that tie you into using their remote data storage services, that giving up control of her data was necessary to use it. And it might be, if you insist that setting up the system be as easy as possible. But the tradeoff for this ease is a drastic reduction in simplicity. You shouldn’t need a remote server in some foreign country to turn your lights on and off. Adding “the cloud” into the mix brings a lot of complexity, mostly in the form of servers that have to be paid for somehow by whatever company is providing the service. It needs to be secure. You might even have to create accounts, remember passwords, and manage that whole deal. Sure, that’s easy enough, but it’s a lot of moving parts, and you can’t blame her for rejecting that complexity.
My system is hosted on a now-ancient OrangePi in the corner, and the network in question is an old WiFi router that it sits on. Nothing needs to leave my four walls, but actually some of it does – I bridge some of the MQTT topics out to an external server for my own amusement. There is no protocol, and no real “system” frankly. Each device in the network has its own topic, and I’m responsible for knowing what it means. The thermometer in the basement has an ESP8266 that transmits on the home/basement/temperature topic, and it puts out its temperature in degrees Celsius. It was the simplest system I could think of, but I have to write whatever software I want to log, display, or act on the data. Of course, that’s simple if you can write some four-liner scripts on the OrangePi broker, but it’s not easy enough that my wife wants to hack on it.
So if the full-buy-in commercial systems are easy but overly complex, and my DIY network is transparently simple but requires a level of hands-on that isn’t easy for “normies”, is there a middle ground? I know half of you are already screaming Home Assistant or Domoticz, and you’re also thinking of which client device libraries you like the most for all your DIY applications: ESPHome vs Tasmota, for instance. And you’re all right!
We are living the in the golden age of the home automation projects. Open-source software and firmware, combined with an abundance of online tutorials and worked examples, have made huge strides toward bridging the gap between simplicity and ease of use. You can set up a hub for everything on a single-board computer, upload the software of your choice, and you don’t need the complexity or loss-of-support liability of a cloud provider. At the same time, setup is easy enough if you’re willing to roll up your sleeves a little bit, and when it’s not, chances are good that someone else has already figured it out for you. These days, interoperability with popular commercial products is shockingly easy to boot.
I need to spend some time and rationalize my system: given the state of the art, it’s simply too simple, and taking a step into an open-source solution would make it easier to use for the rest of the family, without overly complexifying things, adding sketchy dependencies, or losing our data sovereignty. I haven’t finished exploring my options yet, but from what I can see, the community has converged on some goldilocks setups: not too simple or too easy, but rather just right. Thanks, y’all!
Instagram Engineering
Shopify Engineering
