I built an offline smart home
androidauthority.com
androidauthority.com
Before I installed the first smart device in my home, however, I had a single requirement: EVERYTHING MUST FAIL NORMAL. By this I mean that if the internet went out or a vendor went out of business or I rage quit a vendor, then the device in question had to act like it was a normal non-smart device.
I avoided things where you had to configure power-on default behaviors, and where you had to cycle power or disconnect the device to reset it. For this reasons I have no smart light bulbs in my house. I replaced every light switch with a smart light switch, and if anything goes wrong, walking over and using the switch will turn the light on or off.
Most of my automation is "zero touch" flavored - I want the house to do what I want without me triggering anything. For instance, I have luminance sensors in several rooms and during the day if it gets gloomy or overcast, or near dusk, it turns lights on. I also use occupancy sensors to turn lights on and off, and have multiple rules to set lights to the appropriate brightness based on time of day (after bedtime, the lights still come on when someone goes in the bathroom, but very dim, etc.). All this happens locally via HomeKit or HomeAssistant. And, if automation fails, then the switches still work.
I'm often surprised how little discussion I see of failure analysis and especially failure design is in software engineering. It is one of the most important concepts in engineering, but I guess we still operate like software can't kill you.
For those unfamiliar with this type of analysis I'll give you 3 examples. 1) Imagine you are building skyscrapers in city with a bunch of other skyscrapers. You don't expect them to fall but you question what happens if they do. You will explicitly design them to fall in on themselves (say after... idk... a plane hits them?) so that they do not fall on their side and take out other buildings around them. An important aspect of security design is explicitly around this whole idea. When your systems fail do you want them to "fail open" or "fail closed"? 2) Fail open would be like TSA, where if you start firing off rounds in the queue the system fails and you get "access" to the rest of the airport. 3) Failing closed is typical of safes, where you accept the fact that you will need to take a blowtorch and drill into the safe. (the TSA thing did happen, just not in America)
While these examples are a bit extreme the ideas are incredibly important for many products. I'll give a good example of this by looking at Android Auto. If a user gets frustrated with the system and is not able to accomplish their task in the way __they want to__, they will disconnect their phone, look down at it, do the thing, and reconnect. No matter how much you want to push voice operations, if the person doesn't like that, they fail over to a more dangerous set of actions than if you had just let them use the console a little longer or let them do a specific task. At least their eyes would be closer towards the road. I'm often amazed that there are lack of consideration for these type of things in home and auto platforms where there are often high stakes. How things fail is one of the most important things when designing systems (I'd argue it even matters for things like law and medicine).
This is a major reason why many "traditional" engineers do not consider programmers to be true engineers.
[0] https://en.wikipedia.org/wiki/Technology_readiness_level
Most of the time physical devices failing will have much more severe consequences, than a website going down for an hour. In some cases there are APIs which many other critical systems use and frequently those APIs do have much more rigorous process in making sure that they are reliable.
All of that seems regulated by governments and markets to me.
They have promises of reliability and big financial penalties if they were to fail on that reliability.
Outside of engineering, few people have ever even heard the phrase “risk management”, let alone understand the theory behind it. It’s really hard to argue that a developer can consider themself an engineer, when they don’t bother understanding what happens when the thing they “engineered” fails.
Yes this applies to web applications too. If someone can’t access an application that gives them directions, it’s possible that they will be stranded. If a healthcare website fails to load a list of available physicians, it’s possible the user gives up and doesn’t seek medical care.
The current prevailing attitude in web dev of “who cares” is really a testament to how little most web developers and real engineers have in common.
Software engineers working on safety-critical systems deserve that title because they actually have to understand software risks, design appropriate mitigations, and prove their effectiveness.
And not to say there aren't plenty of software engineers that do failure analysis. I mean we do have exceptions, exit codes, traces, and even QA (though many hate them). But I'm not under the impression that the idea is as ingrained in the average software engineer as it is in the average engineer. Having taught CS I've been surprised at how much less time is spent on this topic and how less explicit it is usually discussed when compared to other engineering disciples. The same goes for ethics which I think go hand in hand here. My impression is that this is because when software harms it is often less tangible and less direct than in other engineering disciplines where it's obvious how a bridge failing causes substantial harm and economic impact but where a failure to protect against injection in your not mainstream platform also causes large damage because people reuse passwords and that passing the buck to user error is unacceptable given that it's expected behavior from the user (work on expected behavior and account for unexpected behavior but never rely on domain knowledge unless under special circumstances). We all know everyone reuses their Gmail and bank password on websites like Netflix or fucking neopets.
We wouldn't have as frequent front page posts about users getting accounts unlocked only due to the fact that they were able to go viral if software engineers better planned for failures in the system. Do Gmail devs really deeply think about what happens when someone loses access to their account? If there's a means of recovery if they're devices are also taken or destroyed? That this solution works in practice, not just theory? Do they audit and dogfood this process to ensure it's working properly and understand how it fails and the consequences of that? I'm assuming no because we see people locked out of these systems ("these" because I'm generalizing beyond Gmail)
You're certainly on to something here. It feels easier to disassociate oneself from an ephemeral codebase whose existence vanishes entirely from your life as soon as you leave a company. Compare that with someone working on an airplane that a family member may rely on to safely transit.
That's what makes it even more sad that computer science majors are shielded from all those concerns in engineering. There really should be a "software engineering" minor that delves into what the rest of the engineering world contends with.
Hope this isn't too blunt, but I find your writing style to be a bit "chatty". I think you must have got the same advice I did, which is "write like you speak".
For most people, the vast majority of our verbal communications are highly informal, filled with lots of slang. Verbal phrases and euphemisms don't translate very well to writing sometimes.
If I were to offer advice, I'd say write shorter sentences and use simpler words. It should feel a bit uncomfortable to write like this. I feel slightly uncomfortable writing this even.
Because the traditional engineers do this in a lot of stuff and not just high risk domains. You can probably find somewhere an FEA analysis on your TV's legs and where the high stress points are and where it'll break.
> They have promises of reliability and big financial penalties if they were to fail on that reliability.
Ehhhhh really? Or if everyone does a shit job do we just take it? Most people aren't tech literate. Hell, most apps aren't built by tech literate people. I have a laundry app I have to use and the rooms aren't in any logical order, just randomly placed. Are you tech literate if you don't know a sort function exists? Or if you don't know people like to read things in a sorted order? If you can't google the answer? There's lots of systems that have really shit designs in them and as discussed elsewhere in the thread I don't think companies are having big financial consequences for their lightbulbs not working when the wifi goes down because the people purchasing them aren't aware that such a insane thing would happen. If you're not very tech literate you wouldn't even think that's a possible outcome. So don't just rely on markets because econ 101 isn't enough.
Unless it's a very large safe where people can walk inside, like a bank vault. The fire department gets very grumpy with you if they fail closed and can trap people inside.
So if you want to force open a bank vault, stage a purported terrorist attack, and the FBI will cut power to the building as standard operating procedure. Voila, vault opens.
Software is killing people since decades: https://en.wikipedia.org/wiki/Therac-25
Another rule I include with this is that the "Smart Home" should still be able to operate and be controlled like a non "Smart Home" by a regular person. (e.g. light switches, thermostat, door locks)
I've got dumb bulbs and smart dimmers/switches so everything can be smart, and also can be used by normal people.
He just wanted to save $28. I have these in my house. No app required.
There's actually no particular reason you couldn't do it the other way round I suspect but I've not heard of anyone implementing it.
This is the Mitch Hedberg Rule.
"An escalator can never break: it can only become stairs. You should never see an Escalator Temporarily Out Of Order sign, just Escalator Temporarily Stairs. Sorry for the convenience."
One example would be if the gearing breaks so that the segments move freely:
Congratulations, you and the crowd--given false confidence by a comedian's joke--just piled onto the top of a downhill treadmill... It will now accelerate until you crash into a smothering crunchy heap at the bottom.
Yes, this, you get it.
I can't believe how many people I talk to are happy to install permanent things in their home that have a strong dependency on (a) the internet being up, (b) the company still existing, (c) the app still being supported.
Things installed in a house have a lifetime measured in decades. IoT companies and apps have lifetimes measured in months. This is not compatible.
I do not need smart home, but if you do, then I think that you are right. This is a good idea and is the most important idea.
IMO that would be the first smart home product strictly superior to its analog counterpart.
I haven't experienced many outages recently, but I'm fine with lights off until I turn them back on as a default. Its also possible to just have the lights revert back to their previous state as an automation in HA, although it seems like a waste of time to program instead of turning lights back on as you need them.
I’d be curious to hear more about this. Most of the “smart” light solutions I’ve tried struggle to go directly to a specific lighting level without some bright flash as they activate.
What brand switches/lights are you using to do this with?
They've added a popup to the app recently saying that soon Hue will require an account to work though, so I don't recommend them anymore. I've been meaning to move them to a separate offline network before they force update them to require an account. I originally purchased them because of how well they functioned locally without needing any Phillips cloud stuff. But sometimes I still use the official app instead of my homebrew setup.
If you own your own home, smart switches with dumb dimming capable lights are the way to go IMO. I'm still renting though.
I’d stay away from consumer grade stuff, it’s pretty much all garbage. Commercial controls will come with a 5 year warranty.
Most commercial lighting controls have 0-10v dimming, but you can get forward or reverse phase line voltage dimmers to use with consumer grade LED lamps (bulbs).
No flashes from any of them.
I went Leviton as they looked the most like low-key, plain-old On/Off lightswitches. But the reliability has been terrible - constantly dropping off the network or requiring a reset (at least the early models I have). It's disappointing considering I thought it was a well-established brand; after dealing with some of their tech support / engineers I'm of the solid opinion they should stick to non-smart appliances.
It'd be even cooler in two-ways, where using the switch on the one side of the room causes the other switch to match. Not sure why this isn't available either.
It's kind of like a joy stick, so in a neutral center position and you flick it momentarily up or down from center to ask for on or off. You have to learn not to lift and hold for too long, because that is how to get it into a very primitive reconfiguration mode. As I recall, you can reconfigure the LED and maybe its behavior on power restoration.
I configured mine so the LED is on when the light circuit is on, to show me the status of an exterior light on the other side of an opaque door. But, I think the default is the inverse so it acts a bit like a night light to locate the switch in the dark. Of course, it's a hideous blue as is the modern fashion.
I linked it to an existing alarm system to set some rules like turning on and off around dusk and dawn periods and turning on in response to camera-based motion/people.
I've ordered some Inovelli switches which do this and have an RGB LED if I need a state status indicator.
Sadly it was very expensive and also locked down to their private cloud. The company went bust and took everyone's 'smart' homes with them.
As a side note Lutron did a pretty good job with led brightness which is a big deal at night with lots of devices.
- FAIL NORMAL - WORK OFFLINE - YOU CAN STILL BUY
I really want my smart home devices to meet these criteria, but at the same time I have a bunch of other stuff I do.... so it would be awesome to have a starting point from people who have done it already.
Mozilla used to have such a database, but seems to have forgotten it.
"Fail normal" would be trickier, I don't think there's one size fits all. For lights, prefer switch relays over smart bulbs. Look out for customizable "default" state when power comes back on.
Most of the stuff either looks to be subsidized to keep online access which will earn data or offline capability is charged more because they dont get data
There's also a home assistant thing for them.
Is the switch always "on" and the sensor then turns on the light and you just have to turn the switch "off"?
And then if the switch is turned "off" will you need to first turn it "on" again in order to make the automation work properly again?
By using occupancy and time of day conditions I can have multiple rules to turn on lights to different brightnesses at different times of day. I turn on sets of lights vs individual lights so it’s not the absolute most efficient in terms of power but on the other hand having lights turn off automatically is a good saver when you have kids.
I use Lutron Caseta wall dimmers and dimmable LEDs everywhere so they already have a nice fade-off effect. I once tried to code a fade-on effect but it turns out that instant on is preferable so I gave up on that (it was unreliable anyway).
For low traffic areas like closets I use Lutron Caseta motion sensors; they work great and have very little configuration.
One note of advice - there is a lot of grey market Lutron stuff on Amazon; you can get burned by clones or up-labeled stuff that doesn’t function exactly as advertised. I would stick to buying at a home store or buying from the Lutron store on Amazon.
So it doesn't matter if you manually turn them on/off - they remain network accessible.
This is why they're strictly nicer to use than smart bulbs. Even if you want smart bulbs, you'll probably want smart switches too, but if you have smart switches you only really get a benefit from smart bulbs if you want to control the colour of the light.
You can control brightness of a smart bulb. Not all dumb bulbs are dimmer-friendly. That said, you can always replace those with dimmable dumb ones.
But also, smart bulbs with a dumb dimmer is the worst combination, so if upgrading somewhere, if it has dimmer switches that's an extra reason to replace the switch too, irrespective of whether you also replace the bulbs...
hmm... wonder why bulbs can't fail normal.
maybe they should make bulbs with a deadman switch. If they don't get a "ping" from a controller in a certain amount of time, they revert to normal light bulbs?
or yeah... smart switches.
The smart bulbs may have appeal for people that are unable or not allowed to modify their light switches.
By keeping the "smarts" at the switch level, you have a lot more choice in what bulbs you use (e.g. along color temperature, color rendition, light output, power consumption/efficiency, etc) versus the smart bulb.
The Lutron Caseta series of lighting products are basically the gold standard for being "fail normal" functional without network connectivity to the system control bridge, and the system also uses different RF frequencies than 2.4GHz Wifi based "smart" switches, so it doesn't rely on or potentially create congestion issues on your home Wifi network(s).
I totally agree with your principle about fail normal, but how is this a violation to that?
Some switches allow for direct control of zigbee or zwave smart bulbs so even if the hub is non functional, a press of the switch will still toggle or even dim them.
The nice thing is you can:
1) Buy eight random Tuya Zigbee outlets ($40)
2) Three PIR presence sensors ($10)
3) Fancy Sonoff Zigbee USB device ($20)
4) Three RGB-CCCW lightbulbs ($15)
5) Three or four temperature/humidity monitors ($15)
And you're at the $100 mark with a basic smart home without breaking the bank, with a good path to upgrade if you like it.
The main annoyance are shipping times from China.
Also be very wary about devices that are not (verifiably) UL listed / CE rated, especially if the device is to be plugged into mains and tucked behind some furniture (or worse yet, installed into the wall). [According to internet folk lore] some of these can be grounds for being denied insurance coverage in the event of a fire or other damage [citation needed]. Personally, I don't mess around with random-china-devices and mains outside of the workbench.
Is there a reason you think insurance wouldn’t pay out because you used a non-listed device? This is something I frequently see people asserting without any basis in reality, and I’m curious where it comes from. People do stupid things that insurance covers all the time. I do agree that you should only use listed devices whenever possible.
Honestly, I don't know and I haven't personally heard of anyone getting denied claims because of stupid things related to DIY repairs/modifications. I think this repeated wisdom comes from things that we read in HOA and rental agreements about DIY modifications.
I'll edit my comment with a [citation needed]. Thank you for the question :)
It does make me wonder: what if a fire caused by a DIY installation of a ebay gadget resulted in loss of life? If the aggrieved parties went to court, who would be held liable?
Here is another discussion: https://old.reddit.com/r/HomeImprovement/comments/7vwe23/doe...
So if you can chip swap them, and ESP can drop right in with ESPHome. If they'd just let us flash custom firmware without needing an exploit then you'd be able to run ESPhome without any fuss on the raw chips (the Bekan firmware is included in ESPhome now - it's great).
They're sitting on a pile of easy "family and friends" recommendations if they'd just drop the ecosystem aspiration.
In an ideal case, each of these devices would:
1) Be open-source
2) Be programmable in a stack of kid-friendly languages, with a stack including Blockly, CircuitPython/JavaScript, and full-fledged kernel hacking
3) Clean, clear documentation and web interfaces for doing this
4) Integration with a secure non-wifi protocol like Zigbee
5) Have a standard, maintained back-end, so things like security fixes can hit all devices, no matter the sensors
Kids can (and some would) learn to do things much as they do with Scratch or Hour-of-Code projects, and it'd just be a different dynamic. There is a mint to be made there if a platform like Tuya could pivot here.
It's a decent chunk of work, but far from astronomical.
Step 2 would be to do something similar for HomeAssistant, or for a similar platform, and there ought to be devices which ship with that. The Zigbee coordinator should be programmable, configurable, and kid-friendly to do so. It should also automatically install security updates (ideally, via Ubuntu, rather than the vendor; Ubuntu is looking to make money; a $5-per-device plan for long-term remote maintenance would do it a lot better than giving me warnings about my machine being insecure, and if I won't pretty-please opt into having them track me with their premium platform)
Just get a Zigbee dongle for a Pi and pair them if you want a hub.
Admittedly I do not use Tuya, I'm only going off the parent comment about it being Zigbee but Zigbees ad-hoc mesh hasn't failed me.
* Wifi spy-on-you-compromise-your-network version: $2
* Zigbee version: $4.
I can spare the extra two bucks, though.
Yeah, right.
Nobody in their right mind should be making their home "smart" for the convenience or efficiency. It's a hobby. A frustrating, inconvenient, burdensome, potentially expensive, challenging, neverending hobby.
It taught me to love light switches.
I now never touch a light switch.
"Computer dim lights to 40%." I never bother touching a light switch. This guys set up is too complicated. The easy way to do it nowadays is to buy smart devices that integrate with the google or alexa echo system and you just talk to alexa or google to get it to do what you want.
If you trust that your phone isn't listening when it's not supposed to, why don't you trust the home assistant (even if it's made by the same manufacturer)?
Do home assistants claim they are not listening when they are not supposed to?
Even if you create tasks that can function through a button press or something they still rely on the cloud.
Voice control locally is really hard though. Or at least it used to be. I thick OpenAI Whisper is available for local/offline usage if you build your own wrapper around it? I tried adding (local open-source) voice to my home system about 8 years ago, and again 3 or 4 years ago and it was very rough. It might finally be feasible with the current state of local AI.
For an example, my home robot was named "Marvin" (from H2G2) and the local detection was so poor, this is my list of "matching" wake words.
['Marvin', 'arvin', 'artin', 'marlin', 'marten', 'margaret', 'why are']
I didn't really value voice control until I had young kids.
Agreed, but Alexa fails hard if you have no internet, and my understanding is that Google Home also becomes mostly inoperable.
I've tried Home Assistant but could never get it to work reliably. HomeKit is great and works offline if you have a home hub (an Apple TV, a HomePod, or an iPad).
It also fails hard with Internet. I’m sick of being on my living room asking Alexa to turn on the light just to get the bedroom light turned on (each room has its own Alexa/Light pair, so there should be no confusion).
Walls reflect sound, and things like curtains muffle it, and you might be surprised which device picks up your voice best in different positions.
I had to adjust placement etc. to prevent the Alexa in my son's room from regularly picking up my commands from my room better than the Alexa in my room when the doors are open.
I forgot to say that the problem mostly occurs when I ask Alexa to set the (unspecified, but hopefully the one belonging to the room the corresponding Echo is) light's brightness to X. Sometimes, in that case, it turns on the light from the other room.
I didn't check the audio transcript when this happens though, that's a good one, thanks.
E.g. "this request on this device refers to X", or more generally "when you hear X it probably means Y".
As an alternative, I've spent more time than I should passive-aggressively repeating the same instruction and then flagging every single failure in the app. I have no idea if it makes any difference, or if the things I've done that with that have improved were due to general improvements to their voice recognition, but it gives me some limited satisfaction (I now wonder what the financial tradeoff is of fixing the reported issues vs. any benefit potentially gained by just having the reporting function but sending it straight to the bin to make people feel like they're being heard...)
One of my goals for this year is to experiment with HomeAssistant or similar to at least start to replace or supplement my Alexas.
I’ve installed home assistant recently too for the flexibility, but until I get a zigbee stick I still depend on Alexa as the interface with the lights.
For Amazon Basics WiFi smart plugs, the device type can be configured as "Light" instead of "Plug".
Echo 4 works as a Zigbee hub, avoiding dependency on a third-party cloud service.
I’ve got a mix of Lutron and Phillips Hue devices with a Honeywell Thermostat tied together with Home Assistant on a raspberry pi. (Lutron needs a hub. The hue devices connect directly to the pi.)
It’s been absolutely rock solid. I’m using Lutron Pico remotes as switches. I haven’t had to replace a battery in over a year after installing everything. Hue motion sensors just work.
Lutron switches are absolutely critical to my setup. They work (and paired remotes work!) when the network is down, which makes automation a progressive enhancement. Same with the thermostat, it’s offline first. Wifi connectivity is optional.
I tried living with smart bulbs but they get really expensive if you’re doing a whole house. And rebooting light switches is absolute insanity. I do not want automation bad enough to make my light switches less reliable to get it.
The cloud is always someone else's computer, and any free service will likely end before your hardware needs changing.
Lutron switches are great too, I have them. I am not a super fan that you have to seem to have to buy their pro hub to get telnet access to the lutron switches and also not use the cloud.
But every time I read an article like this, I can't help but think how cool I would feel if my house could tell I was coming and turned my lights on and my heat up, or dimmed and then turned everything off once I got in bed.
- ability to turn on house heat when we're headed home from a trip, or turn on the AC for the pets if there's a heat wave when we are traveling.
- ability to turn on the tea kettle when driving back home from an errand. Needs prior setup of course, and a switched outlet that can handle high current.
- ability to align porch light schedules with sunset/sunrise. I never have to think about it.
The "don't want to get up, i'll just do it from my phone" thing is nice, though, especially if you are stuck under a cat.
I also had only a few switches and the trade off was I could now dim them, at different times and different ways.
Tied it in with a Logitech Harmony Hub at the time for instant movie mode. I got a lot of use out of that little hub to not have to go full home automation.
Home Assistant is great though, it's much easier now, it will know you're coming via app and turn off everything. You can do the good night thing too.
Now my blinds automatically go up at 8.30, and close at sundown and I don't have to think about it - but my rooms are pleasantly exactly how o want them at all times.
What I'm most happy about is how I got them hardwired - I notched my cornice and slid an archetrave (narrow) light switch up into the cornice on a mounting block, caulked around it, and then ran a thin bit of surface conduit down to another slim junction box next to the blind. You can "see" that the power is there, but it blends nicely and means the blinds are powered off my lighting circuit - no batteries to worry about, and a shutoff in the room if I ever change out the motors.
Motor-wise I'm happy so far - the limits work, and it's been set and forget with Home Assistant driving them through ESPHome. Unfortunately the motor is a Realtak-based chip, not Beken, so no Tuya-cloudcutter exploit exists to get ESPHome on their wirelessly. If I was doing it again from scratch I'd try and find something easier to de-cloud. As it is since I know the motors fit and know how to open them up, we've got a bunch more we're planning to do since we have some really big windows we'd need multiple blinds to cover, and we just never considered it practical till we did this.
EDIT: Basically it's a home automation upgrade I've been incredibly happy with the outcome for, but man if Tuya would let go of the lock-in aspirations then it would become dead-easy to recommend to people.
Did you write/record anything online about the chip swapping, or can you recommend any resources?
Did you end up deciding on something easier to decloud for your future ones?
Thanks!
Every setting you do manually throughout a day or season can be automated.
Tie it in with movie mode to close everything, etc.
Luvolite and their rebranded partners is a decent bang for the buck: https://www.louvolite.com/
I hear Ikea is worth looking at too.
I do have two accordion blinds I raise and lower daily, but only two. One wall of my house is a giant sliding glass door with floor-length curtains, though, which are as manual as it gets. They don't even have a stick to guide them. I just grab an edge and pull. I don't think my landlord would appreciate it if I mount a whole different track, but I guess it might be different if he doesn't find out.
My heater is my biggest concern, though. It's a propane-fed floor unit (house is very small, as I mentioned). It's controlled by walking over and pushing buttons on the top of it. It has no remote. The thermostat is part of the unit itself. Turning power to the unit itself off and on would kill the pilot light, never mind that in the winter I can't have it go off. I just want it to go from 65 while I'm home down to 55 while I'm not. It only takes a couple minutes to warm the house (again, house is quite small), but the pipes would freeze if I let it get any colder.
And that's where I come back to, "How inconvenient is it, really, to open my curtains in the morning and turn the temp down when I walk out the door?" The heater is even adjacent to the door.
In other cases a room can't keep it's temperature stable and might need the blinds open or closed to help regulate the temperature.
I still use switches in some places in my apartment, but haven't had a need to touch my bathroom and kitchen light switches in a year. Once you start getting used to your automations, it really does feel like a level up to your home.
Though of course the fun/cool factor is a separate thing (:
The cool factor is maybe 25% of it for me. It's mainly an obsessive hobby.
And you're entirely right it's an unstoppable gateway to learning how to code arduino's and then esp32's, and then eventually hardware projects that aren't even smart home related.
I think that consumers should have a choice, and they should only need to connect to additional services or a subscription model if there are useful reasons to do so. Most consumers simply do not need their "smart lock" to phone home. If you wanted to do more interesting things, like issue temporary keys to run an Airbnb, then sure, it makes financial sense to pay a small subscription fee. But, most folks just want a little convenience.
Of course, it helps that I'm generally distrusting of businesses and of how behaviors will change after my departure. If you want your products to be ethical, the ethics must be built in from the beginning.
What I heard from other people in the industry is that when Facebook had that huge multi-day outage, it was partly because their smart locks couldn't open externally without an internet connection, and it was relying on Facebook's DNS which was the source of the huge outage.
As I liked to say, I wanted to build a product that I'd feel confident about giving to my grandmother. She should be able to use it even if she lacks familiarity with current technology, and I should feel that she's safer for having it.
I don't want to name my company mainly so my coworkers don't learn my HN account, but basically every product feature involves a consultation with a very well known white hat hacker.
The lock still works fantastically well. I admit the new ones with homekit support are way slicker because you can set it up so it unlocks if you hold a paired apple watch up to it. That's slick dude. Hella slick!
I did about 20% of the article's setup which took about a weekend of time, then randomly after Home Assistant update certain things start not functioning, i.e. I bought some $2 Xiaomi BLE temperature sensors and the key I setup just somehow stopped working, then its another half day wasted on trouble shooting them.
I regret my time wasted on them, especially given Home Assistant's data retention is not as great as I was expecting, i.e. default storage is only 14 days for temperature sensor and that sqlite DB is huge with this limited data.
Just thought I would mention it.
Regarding lights, I understood any spot lights are nearly impossible to plan ahead correctly, and are used rarely. what is really needed is multiway switch for the ceiling light, and 2 levels (1 dim bulb for dim light, more bulbs for bright). The rest -- local spotlights, dimmers, adjustable hue -- add very little value, but need maintenance and are hard to repair (try finding a rotating knob for a dimmer). Use table-top lamp or one with a clamp to save space, with yellowish light for local lighting, same for bed light. That's it.
Using smartphone may have seemed ok when they were new in 2010. Now I see they're pain to use, apps live their own life, you have many distractions, and it often takes equal amount of time to walk to the switch to check/turn off, as to browse to it in the phone. Same if you have kids -- you get distracted out of the phone, but phone won't save you time, it's much harder to use smartphone with kids nearby.
The biggest thing with Apple is they get that houses have more than one person living in them, and provided everyone has an Apple device, it all just works.
Initially I didn't think this was THAT big of a deal, but the more experience I got with smart homes the more I'm thinking the ecosystem's are doomed to plateau while Homekit keeps trucking past them since its reliability problems are not as severe. Amazon and Google are building castles on sand.
I guess but cannot confirm that Google has continued this practice. It didn't seem like the problems you mention exist, unless Google did something stupid and disabled all of this perfectly working functionality.
In any case, I’m holding out for a HomePod mini refresh because if they throw a s9 chip in there it’s going to be able to do basic voice recognition offline
Maybe matter can save Alexa/Google home and I’m too pessimistic but I don’t like how they work now.
Just the other day after a power outage, three Nanoleaf bulbs in my home refused to rejoin HK and I had to repair them. That meant deleting them from HK first. Fortunately I was able to use Controller to restore most of the scenes/automations that involved those bulbs. But it was tedious and didn't restore everything.
At this point I'm seriously considering switching to Home Assistant to act as my HK automation hub. That would give me the best of everything: local control, HA for programming, and using my Apple devices to turn things on and off.
And other Apple restrictions. I've got a lot of stuff looking really good with this combo. And then I added a Brilliant (https://www.brilliant.tech/home?tab=system) Smart "Outlet", particularly for my living room, which has six can Hue lights, and a lot of ambient stuff around the TV - it's great for driving a lot of scenes, daylight, evening warmth, media (where the front row of cans are off), etc. But it could also so a lot more that would be nice (Christmas lights, which are off a smart switch etc.) - except Apple doesn't allow external devices to initiate HomeKit actions. So a lot of potential usability is lost.
Our place came with an Ecobee thermometer. I threw it up on Homekit and my partner had it on their watch and phone instantly. Too easy.
Homebridge is a great addition to a HomeKit setup, too. https://homebridge.io/
¹ EDIT: Removed iPad mention, since as @js2 notes, "iPad as a home hub" support is being deprecated.
You can of course use HomeKit devices with Home Assistant in both directions. HA can obviously serve as the automation hub for HK devices ("accessories" in HK parlance) and HA can expose itself to HK so you can control HA devices through an iPhone, iPad or Mac.
Or a HomePod mini, or a 2nd gen HomePod, or a supported 3rd-party device that can act as a Thread border router.
I will say that having my wired AppleTV act as a hub, I've had no trouble at all. I have maybe 50 accessories and a bunch of automations, some fairly complex. e.g. my garage door automatically opens for me as I near my home as long as my iPhone is in CarPlay mode.
"Computer, dim lights by 30%"
"Computer, open curtains."
"Computer, vacuum the floor."
"Computer, play some jazz"
All of these are phrases are actual literal phrases that I use in my home, some on a daily basis. It can only be done, with current consumer tech, through the internet. All you need is a roomba, some 4$ smart bulbs, a spotify account, a 30$ auto curtain robot, and an alexa. For google home you can do the same but you can't call it a "computer."
In the future they will most likely be hooking up some LLM super computer to the home. In that case it can only be online.
The idealistic version of a smart home involves what we see in Star Trek TNG. Talking to the computer and only an online version of that can fulfill this vision currently and in the future. If you go offline, what you get is what I term a "programmable home" more than a "smart home".
Well, not necessarily. The open source LLMs i.e. Mistral/Llama 2 can be run offline if you have a good consumer grade GPU.
I'd like to see an open source LLM control home devices Alexa style.
Probably starting with ESPHome to control my heat pumps over IR.
The remote works but I forget to turn it off when not in one of the rooms, and it can't do smarts like keeping temperatures in desired bounds.
An ESP32 running ESPHome with high power IR LED, a presence sensor, temperature sensor, and some pre-built library to control heat pumps[1]. it looks like it should be fairly simple to implement with Home assistant.
Now if I could only disable the beep from the heat pumps receiving commands.
I would like to do smarter lighting, but the lack of decent smart light switches that support local control and are also licenced for use in AU are a blocker there.
[1] https://github.com/crankyoldgit/IRremoteESP8266/wiki#ir-send...
When Amazon discontinued them and actively pushed an update to brick the devices, it was a nice inexpensive wake-up call to not trust phone-home IoT devices.
Sometimes the phone-home comes in the guise of over-the-air updates that are supposedly good for security or add features. But sometimes they actually remove features.
I think the author of this article has it right — the best security and most trustworthy IoT setup is to host it yourself; never connect things to the internet at large. If your gizmo doesn't support that mode of operation, don't buy it.
- HA have a slick UI, but it's a MONSTER, less monstrous maybe then OpenHAB but still a monster, I HATE it's YAML config, it's use of SQLite is well... And it's less horrific way to deploy, by pip in a virtual environment is meh, at best;
- wires are needed for reliability, but while ShellyPro and WaveShare seems to be reliable (in few years almost no issues) they are still limited, damn expensive and not much comfy to keep them up to date automatically and centrally. In general most IoT products are crap at best, badly documented, buggy, dysfunctional and damn expensive;
- most sound reasons for home automation lay in energy saving/automation, I normally see no sound reason to switch on the lights from a mobile phone instead of a physical switch for instance, but most appliances seems to be designed without any clue on how they can be used and integrated. A small example, I have a Daikin/Rotex heat pump for domestic hot-water, it's not really a small economic device, well: just to read water temp to makes decision I need a 300+€ extra card, my take was a far cheaper pt100 inserted via the hot-water return intake I do not use. Remote piloting can choose "do not heat", "heat full power, heat-pump if possible + classic resistive heater", "try heating with the heat pump for 15' than goes full power", "choose automatically what to do".
WHY THE HELL not offering a simple:
- try heating with ONLY the heat pump
- try heating with ONLY the resistence
- do not heat
- heat full power
leaving the extra bit for "choose internally what to do"? I still have not answer, and to heat only with the heat pump, for instance because I have just 600-800W spare from my p.v. I have to switch any 10' "do not run" for a minute, than "start with the heat pump" in a loop.
My VMC is even worse: it's only ModBUS, native, but various registers store/read different values than those documented, a small set of them are totally undocumented, some fails to read or write randomly so I need a bit of extra logic to try again few times.
IoT can be a nice addition, but so far there is simply too much crap both on electronic side and on the digital side.
The article is about offline Smart Homes.
Using something like HomeAssistant, and products supporting integrations with local control you can eliminate data going to third parties.
Get a stopwatch and try it out.
Setting it up for others to use it is another. The comment about Apple Homekit "just working" has some merit. If it's not automated in the background on a schedule, and on demand, there's some merit to it.
Maybe there's a sweet spot of not having to hack every little thing.
My smarthome setup for example is centered around becoming an appliance of the home that stays with it, and making design considerations for that.
More scrolling on HN? Yeah, big win.
I get that much more time with my kids or towards something else I want.
It's free time for the taking for a fixed investment, I'll take it. It might not make sense for you and how you value your time, but in time it might.
The thing about doubt like yours is nothing new will reach a brain that starts closed off. I generally avoid doubt worshippers who validate their doubts by painting them on others instead of openly entertaining a viewpoint that isn't theirs. I got no problems being straight with you the way you're willing to behind a keyboard. I understand the fun part of being a devils advocate and a skeptic, but not everything has to be for everyone, nor does it have to be validated by you to be acceptable.
To answer your question, HN is limited on a few devices, a few times a day, with the procrast settings, and in grayscale. The hosts file has been blocking it for many years. After a while I pick up and put down the device I have for reading.
What do you do that works for you?