My light switches integrate with Home Assistant, but I can just press the button or turn the knob to dim. Home Assistant Zigbee is there as a bonus, not the sole way of doing things and certainly not a single point of failure.

You can set it up in a way that it does work without it but can be enhanced with it. Some zigbee switch work like that (Like the Aqara one).

Also, Home Assistant is the most rock-solid piece of software I ran in a long time, it just never breaks for me

If HA goes down I can just walk over and use the switch. A downgrade over yelling into the air for sure but my downgrade is your daily experience.

If someone comes over that doesn’t know how or doesn’t want to use the smart speakers, they can just walk over and flip a switch.

After moving HA off a Pi I found it to be rock solid and I don’t think about it at all unless I want to make or tweak some automation or I get a new device. Adding Claude into the mix for creating/modifying automations is awesome.

Smart devices work as expected for a year, fail in new and exciting ways, and then you replace them with dumb devices.

Smart light bulbs are another minefield of low-CRI, bad connectivity, worse firmwares, nerfed local featureset, etc.

* Hue bulbs, which get active updates and are still maintained by Philips, and honestly are pretty good, if insanely overpriced, especially the HomeKit bridge

* "iHome" smart outlets which I belieeeeeeeve were once upon a time made by Belkin? They've ended support but the outlets themselves are very reliable, and pairable/managable entirely within the Home app so I barely noticed when they were discontinued. Picked up a 4-pack of them on fleabay recently for $20 and they work great still.

Mind you I'm not saying I'm great at this, if anything I'm lucky in that I really haven't had any smart home tech go poorly for me, apart from some other outlets that were some no-name brand. They're still kicking around my house but they need power cycles far more often than the iHome ones.

I think there was also a novelty around a lot of smart devices that just didn't translate to much real world benefit. And in my case here, there was also the matter of having more pressing issues than fussing with smart lights and the like--like rewiring a substantial portion of the house when various walls were open..

I haven't made a vlan yet for IoT stuff because my ISP's router do not let me, I can't wait to have my own openwrt / opnsense router and put the ISP in bridge mode.

If your device requires, in any way, an internet or cloud connection, even just for setup, it doesn't exist for me. If you have some FULLY OPTIONAL services I might like as a SAAS free or subscription option, ok; but the device MUST 100% work without ever knowing your cloud exists. It must also be designed to fail-safe and have manual fall-back operability (e.g., when the power goes out, I can still manually switch its state to OFF).

I recall an article or comment on HN talking about the advantages of a "MidWit Home", devices that weren't really smart, and just had remote control. I've got a bit of that with some remote switches to use around the house with enough range we can turn on from the car when arriving home. Also timers. Seems like a good way to go.

Question is - are these things good enough that restaurants actually use them? I would be surprised if they were.

And the answer still keeps coming back, "not very". Not zero. But the problem is that there isn't a lot of value there to capture, so I'm not willing to pay all that much, especially when you consider the full cost-of-ownership of failure modes that they create. The daily annoyance of "flipping some light switches" just isn't that great.

That this industry keeps getting run by MBAs who keep calculating

    1. Hey we can lock them in with these centralized services.
    2. (A couple of years later) Hey we can make more money if we 
       shut down our old centralized services.

At least my smart phone, as much as we may all like to complain, brings a lot of value to me that can be leveraged for that sweet, sweet MBA-approved lockin. The smart home proposition never had anywhere near as much to leverage. And the smart phone world still is full of MBAs delusional enough to think that "storing the user's heartbeat information in the cloud" is somehow enough leverage to be worth a $9.95/month subscription to me.