01 October 2021
I’ve spent some time over the last 3 months trying to build a smart home app. I wanted to yoke together all the embedded technology in my home — lights, sensors, cameras, various controllers, etc — and do something entertaining and fun with them. Most smart home apps are booooring — no one really wants to be the operations manager for their house. I wanted to create something a lot more fun, using all the smart home gear, maybe something like an Escape Room kind of game. or Hide the Thimble.
I largely failed.
The smart home space is a confusing welter of devices, oses, physical networks, protocols, apis, apps, services, very few of which work together. Getting devices to talk to services, even from the same vendor, is arcane — don’t take only my word for it — Mat Velloso tried to do some simple home automation work and realized what a can of worms smart homes are. I have Google, Apple, Amazon, Lutron, Zigbee, Zwave, Philips, Samsung, and every other kind of device you can imagine. I have tried many ways to hook them all together, having the most success with Hubitat. I really wanted to create a programmable substrate that I could use for apps in the home.
I got to a limping instance of a game with some invaluable help from Sam and Vlad. But it all fell apart repeatedly. The devices are unreliable, sometimes responding, sometimes not. Networks and protocols are unreliable. The object models for devices are wildly inconsistent, even within a single brand. Heterogeneous networks are a mess of differing transaction models, discovery mechanisms, etc. The various packages and libraries to program against smart home devices are not well supported, many have been abandoned or are stale. Naming and discovery of devices is, well, not even primitive.
At this point, it is not worthwhile to keep on tinkering. I’ve dumped most my gear into a box and into storage. There is no viable targetable smart home market. The PC software market and the mobile software market exploded because a high volume of homogenous (interconnected) targetable devices existed. No such state exists in the smart home market.
I have also tried to move upstream a little and dig into the tools for IOT development. I haven’t really found what I want there either. I’d like to write an app that I can easily move across different IOT devices.
Platformio has some promise and they have a big VSCode plugin, they have certainly aggregated a lot of devices in their tools. But they don’t really solve my problems. Their tool early binds you to a specific hardware platform and libraries, this isn’t what I want. And I had hoped that a great VSCode plugin would be appealing, but the plugin feels pretty heavy – a bunch of decorations added to VSCode, a whole new hierarchy. And if you want to target RPIs, you have to run the whole dev environment on an RPI – yuck.
There are almost no RPI plugins for vscode which I find surprising. Raspberry Pi Sync is just some lipstick on rsync (tho the developer’s name is KingWampy, have to like that). There is no plugin for Balena tho Balena separately has a pretty nice onboarding experience in the browser which I will be digging into further.
Microsoft has a suite of extensions for AWS IOT but they don’t seem to be actively maintained.
Overall the tool situation is not great — which is another comment on the market; with no volume targettable hardware, there just is not much point to great tools.
I’ve previously written about the problems in auto software. Not shocked to see the chip industry is frustrated with the auto industry as well – “It just makes no economic or strategic sense,” said Gelsinger, who came to the auto show to convince carmakers they need to let go of the distant past. “Rather than spending billions on new ‘old’ fabs, let’s spend millions to help migrate designs to modern ones.”
A nice examination of housing affordability. As nutballs as the housing market is, this analysis suggests it is sustainable.
Seth Godin on change and the status quo. I really like his thought exercise around milk or solar power – imagine that the new idea is the status quo, how would that change your view of the change?
A perfume truck. If trucks work great for food, why not other retail categories?
The future of big cities as predicted in 1922 — very prescient
The best software for every need — a good list to start from.
Making the workplace work for moms (and all parents and kids, which pretty much covers everyone)
World tour of mustards – I do like a good mustard.