My family leaves it on voluntarily at all times. I check to see if my wife’s left work yet to know if I should start dinner. The other day I happen to notice that my grown kid was at a good restaurant and I texted them to be sure to try a certain dish. Another kid texted me yesterday to ask what I was doing in SF; I was running errands with a friend.
For us, it’s a fun way to see what each other is up to. The reason our grown and out-of-the-house kids leave it turned on is that we’ve never, not once, used it as a spy device.
> My family leaves it on voluntarily at all times.
Same.
> I check to see if my wife’s left work yet to know if I should start dinner
I (we) don't even do that, on purpose. If we want to know, we ask by message or call.
> The reason [...] leave it turned on is that we’ve never, not once, used it as a spy device.
I agree, it's all about trust. In my mind it's exactly like being a mail server sysadmin: by design you have access to all the emails but you never read them out of ethics. In that spirit: we have location-based automations set up in Home Assistant, and by default there's a geocoded location entity per tracked device, which I disabled, relying only on zone enter/leave events, carefully balancing privacy and convenience.
Now, the use case about leaving it on at all times is:
a) If you turn it off there's a chance you forget to turn it on. That's also why the new iOS check-in feature, while nice, doesn't cut it.
b) The most trivial reason: my wife frequently misplaces her phone (and other things, which recently got airtagged).
c) My wife works in a bar; let's not kid ourselves, the nightlife way back home carries additional risk: e.g bars close at about the same time, which generates a flux of drunk people in the streets, who are more prone to do dangerously idiotic or hostile things (from driving under influence to downright assault). While not frequent incidents are not unlikely either: over the course of two years there has been about a dozen, a few were life threatening; luckily she managed to either avoid, defuse, or escape such situations, but one day she might not and end up being unable to actively alert. Once while riding her bike to work she was hit-and-run by a car, dumb luck had it so that she was still conscious. Similarly I do sports like skateboarding, if by happenstance an accident happen and I'm knocked unconscious or otherwise unable to alert I can be found.
> we have location-based automations set up in Home Assistant, and by default there's a geocoded location entity per tracked device, which I disabled, relying only on zone enter/leave events, carefully balancing privacy and convenience.
Did you set up location based events in home assistant using iOS devices? I tried to set it up, but it always says that I’m at home, which I think is because I use WireGuard to remain connected to my home network at all times. I tried setting up a separate iCloud integration for location tracking, but it bothered me every hour about entering a 2 factor code. I ended up disabling that integration. Wondering if you have faced this issue and potentially solved it.
I tried a similar setup with Tailscale but it disconnects too frequently for such a use case due to iOS VPN limitations, and that's not even considering the fact that it's userland and eats the battery 30% faster when on. Maybe an IPSEC VPN would fare better?
Similarly the iCloud integration suffers from the 2FA issue. There's another third party one that works better in that regard (able to handle refresh tokens or whatever) but these still have a month long lifetime or something and still require 2FA occasionally. Also both are pull instead of push so not too nice on the battery if you want a prompt reaction in zone change.
So I'm simply on the good old public facing dns/https/nginx/letsencrypt, with the phone triggering an internal zone change event to the app, and then the app pushing to the server, and that just works. Maybe I could have used a Tailscale funnel but by then it was just easier for me that way.
Fun fact, I have IPv6 and HASS is self-hosted at home with the AAAA record pointing to it, but my IPv4 is behind CGNAT, so I have a small Hetzner ARM VM, pointed the A record to that, and set up a nginx to hit home over IPv6. To solve the Let's Encrypt conundrum of HTTP challenge not being able to know which machine it'd be pointing to (the name would be resolved to either A or AAAA) I migrated to DNS challenge.
An alternative would be to simply use their Nabucasa cloud thing which serves exactly that purpose, costs money but it makes it super easy, plus IIUC it funds HASS development. I tried it, it just works, but I wanted to do all of the above on Nix as a learning project.
Yep, that all makes sense to me. It really is about trust. We don’t use it to stalk each other. Well, there’s a non-zero chance the kids have used it to see how long it would be until we were home. If I were them, I’d sure I would’ve.
On a small handful of occasions, someone’s been someplace completely unexpected, and we’ve asked them about it. “Hey, you’re in a nearby city. You ok?” “Oh, sorry! I forgot to tell you I’m going to the water park with Joe and Jane.” “Ok. Sounds fun! Have a good time!”
My wife being able to know where I am hasn’t changed my behavior one whit. I assume the reverse is true.
I mentioned it; it is a very nice safety feature, we tried to use it (we were on the public beta) but it doesn't quite fit our use case, where instead we want something completely unattended, whereas Check In is more intentional, like you plan to do something ahead and set up a dead man switch that will notify people if something looks off.
Interestingly enough when one of us sent messages like "is everything alright?" the recipient got a prompt in Messages to try out the Check In feature.
For us, it’s a fun way to see what each other is up to. The reason our grown and out-of-the-house kids leave it turned on is that we’ve never, not once, used it as a spy device.