This is one reason I have almost no interest in IoT / home automation. It just replaces extremely mild annoyances (turning on the lights) with something complicated and fragile.
But there are ways to build/architect home automation so that it's an "addition" and at worst will go back to "dumb switch".
Most of the light switches in my house are wired up through Z-Wave. Those z-wave boxes are in the electrical box in-between the normal, everyday, traditional light switch and the power.
100% of the time, if i flip the switch a direction, it changes the state of the light. It never fails. Power just came back on? still works. Z-Wave controller down? Still works. Internet out? still works.
Then, I can "add on" automation using that zwave network to do things like turn lights on/off from my phone.
If anything were to cause issues, have bugs, have problems, etc... I can unplug the controller and it goes back to fully functioning light switches.
I think your hitting the aspect most IoT designers never think about: systems should fail like an escalator, not an elevator. You should still be able to get the basic usage out of the product if everything else fails.
While I'm working on my own system similar to yours, and I'm designing around the same concept, most IoT manufacturers aren't. I get why people's default reaction to IoT stuff is to run away, because they are mostly very poorly designed.
I love that expression! "Fail like an escalator" describes it perfectly!
The issue I see is that there are many things that are designed like that, but they often require professional installation or are significantly more expensive, which leads to many just getting the cheaper option which works in the best case scenario.
And that's the right way to go about designing home automation. I want to have the ability for my bedside lamp to turn on whenever my alarm is set to go off and then shut off when I have left for work. But I also want to retain control over the light with manual control (aka, the pullstring). Getting a hue or any of the other similar bulbs would help the alarm situation but if you pull the switch of a lamp with a smart bulb in it, the bulb ceases to be smart. Thus one solution is to add a second manual control so I control the actual light and not the power to the wireless hardware inside but this means buying or making additional hardware that has to work flawlessly. I don't want to want my bedside light to not work in the middle of the night. The other option is to use a lamp with two sockets in there, one analog bulb and one smart bulb.
I think the easiest solution for now is to just keep my blinds open a bit more and just use the sun like people have been for thousands of years.
There's one part of home automation that's interesting I think: energy savings. With electric cars, electric batteries and solar panels and smart heating getting cheaper than the gas version, there is a need for all home devices related to high energy usage to be automated...and sadly machine learning needs lots of data that is going to be provided on the internet by the home devices (that will create lot of privacy issues of course).
I concur. Are there benefits and use cases for some of these things? Sure, but in the vast majority of cases it's simpler, easier, and less expensive to go with what we've got now.
The geek side of me really wanted to install a smart thermometer. It would record my movement and learn when to turn the heat on for me!
Then I realized I could just program my old one to turn on in the morning and evening when I am home on my set schedule. It works just fine, and I know enough to think that replacing it with a complicated machine-learning process might cost me more headache than occasionally turning the heat up.
