Smart homes and vegetable peelers
ben-evans.com
ben-evans.com
Self-hosted is the only acceptable alternative here where even the raspberry pi has more processing power to run hundreds of homes.
A simple service to route mobile apps through the potential NAT is the only thing the cloud should be needed for. Obviously downloading "drivers"/configurations are a valid user case but that's part of the setup-process.
Not saying that everyone should tinkle with a RPI but it is trivial to create a more user friendly device. The fact that Amazon, Google, Apple etc. don't even attempt this is just a consequence of them being hell bent on making money from crap.
You are completely correct. But 99% of people don't know / can't be bothered to host their own. Googlezonbook will create a near frictionless experience and suddenly you'll need an internet connection to open your front door.
You still need to connect the power cord and connect to ethernet or wifi (end regardless of whether it is a cloud device or self-hosted device you can configure wifi from your smartphone like the chromecast).
A smart-home-controller is a physical device. You buy it, you provide power and connect ethernet and configure wifi. Now, either the controller relies on the cloud or it doesn't.
If it is completely reliant on the cloud you will have to muck about with registering, if it is completely self-hosted you don't. Simple as that.
EDITED REMOVING A REVERIE TO MY PROFILE, my apologies, often carried away when I'm enjoying the topics here, it's not my wish to spoil the flow as I almost did.
This said, I will surely be delighted to hear from anyone regarding my rather wistful wishes, to which I applied months of health enforced downtime last year. I still think the qualified opportunity isn't a total wash and would be willing to invest in a no tears no recriminations study, later in the year.
About the necessity of a WAN and firmware etc updating your home automation system: are we not in cities now at sufficient density where peer to peer hops in a as hoc network, has become a technical possibility that maybe lacks a sufficient motivation?
Various smart features should fail safe anyway, if that is true then "reliable enough" doesn't have to be all that high a bar. Of course most people won't put up with rebooting their house once a week, so it isn't that low of a bar either.
A raspberry pi has more processing power than's needed to search and store email. ... etc etc
This trope is absolute bullshit. What people want is for stuff to work and, if stuff doesn't work, for it to be someone else's fault and/or be easy to fix.
Self-hosting fails massively at the second criterion there. Right now, if a cellphone breaks, you can just get a new one and your contacts, messages, apps, etc all sync back to it. Done, you're back on your feet.
If you have self-hosted email or smart-home-stuff, the failure modes are much worse, and you're far less likely to have someone else to blame.
Also, a raspberry pi does not have enough power to do good voice recognition, not to the level of siri/alexa/google.
It also doesn't have enough storage to cache all your music, videos you may wish to play on the TV, etc, so some external cloud integration will be needed regardless.
Self hosting has failed. People use gmail, not postfix. People use spotify, not cds or local mp3 players. People use netflix, not blurays. People use google search to find previously seen info, not a locally saved and curated archive of websites or a home-grown card catalog system. People use facebook to communicate, not diaspora.
> it is trivial to create a more user friendly device
This is obviously false or else devices like this would have been trivial to make and seen some success... but things like sandstorm.io, yunohost, and others have been incredibly complex projects and have not really succeeded. It's clearly not trivial since those people have collectively thrown millions of dollars and dozens of man-years at the problem with no glimmer of hope.
The closest we have to success in this space, I'd say, is NASs which ship owncloud... and those have not caught in for the average household at all.
None of these things will kill you if they happen not to work or have serious drawbacks.
Software update to Nest could have literally killed people and caused thousands of dollars of property damage. https://www.nytimes.com/2016/01/14/fashion/nest-thermostat-g...
Or maybe everything you say is being recorded and sent to Google all the time. http://bgr.com/2017/10/11/google-home-mini-spying-on-user-fi...
I'll stick with my dumb appliances thanks.
FWIW, dumb appliances could have done the same thing.
A furnace that stops working in the dead of winter while away can still burst your pipes if they get cold enough, a broken appliance (and an operator unwillingly to change the situation) can still cause someone to freeze the death or suffer from a deadly heatstroke.
This particular issue cited first probably has mostly to do with the fact that nest really wants you to have a 24V power source for the thermostat (A "C" or "Common Wire"). They've figured out how to make it work without for most systems, but that relies on an internal battery.
Wasn’t that C wire issue causing Nests to continuously turn on the furnace causing homes to get too hot?
Spotify is a good example of something that could work identically to what it does today self-hosted. No change anywhere, no extra setup at all, except for the fact that storing the entire spotify library is impractical.
When I talked about a more user-friendly device I meant something of a smart-home controller. Not a general purpose script that glues together some open source packages that each require a weekend of setup from an enthusiast... No, that none of those has taken off doesn't surprise me the least (that doesn't mean they are useful or that I could see myself using one, but it is not for average Joe).
Of course the average "smart" lock is unlikely to be engineered with such considerations in mind and I doubt that the extra effort would pay off financially, but it isn't completely impossible.
This allowed them to (a) not run a power wire to every door; (b) use a long (12 digits) passcode; (c) reprogram passcodes in a few seconds; (d) immediately give feedback on whether you had typed in the right code or not.
IIRC they cost upwards of $500 each. Given that there was already a 24/7 human guard to check you in and out, I'm not sure that they made a smart decision over physical metal keys.
Meanwhile, I have a non-networked, non-wireless electronic combo lock on the front door of my house. It's not hackable without physical contact; if you want to break in, smashing a window is a better route. It's powered by a 9-volt battery and lasts about 4 years per battery.
The big win is never forgetting to carry a key, with a smaller bonus of being able to give friends individually revocable passcodes.
Personally I don't see much use in smart locks for my home, but if I did have one I imagine I'd still use a regular key and carry it with me all the time. The smart lock could still provide benefits such as allowing visiting friends to enter without key etc.
Being able to remotely control and monitor the heat and water is awesome for the vacation house, but that is only in addition to using the physical knobs. It should not increase the risk (but in practice the security in such devices are terrible and you need to be somewhat knowledgeable to even assess the risk introduced).
It's not understandable nor acceptable if my appliance within my home with no inherent external dependencies doesn't work unless that link to somewhere else is working.
> This is obviously false or else devices like this would have been trivial to make and seen some success...
Rubbish. There's simply more economic incentive for companies to create devices which siphon off usage data for them to sell.
I’ll buy cynicism around the data collection aspects of these things but one only need to look to the desktop application marketplace to see the broader trend.
I’ve supported both self hosted & cloud hosted products for money. I know which one I prefer to work on and I’m surprised anyone in tech would disagree.
If my nest isn't connected to the internet, I can still go over to it and turn it and the heat will come on (or go off). If my smoke detectors can't talk to the internet, they'll still work like a regular smoke detector. And even control my Nest through it's own mesh network (If I want them to). I can't speak from experience, but I'd bet that my garage door remote will still operate the door even without a wifi network to sell my open/close times to the highest bidder.
Regardless, I don't think Craftsman is interested in my garage door habits; they're interested in selling me a new garage door that I otherwise wouldn't think I'd need to upgrade. Smart home functionality is just one way to get that sale in. And it apparently works.
You can't troubleshoot problems in every single person's home, but you can troubleshoot problems in the cloud.
It makes total sense to go with a local solution like yours, however when all the control is done through the cloud (http://home.nest.com, Siri, Alexa), it kinda negates the usefulness of it, except for momentary internet connection loss. And it makes it 12x more complicated for the Average Joe ("What's a Raspberry Pie?", "Local network configurations?", "I don't even know my router's password...").
It probably already relies on the “electric cloud” and perhaps the “ gas cloud” already so what makes a dependency on a “computing cloud”any different?
You can still press the light switch if the net goes out.
Chance of a Revolv IOT Home Hub working for the next 5 years: 0.000% this has literally already been shut down.
Caveat emptor.
It's not like we'll all wake up Jan 1 2019 and have nothing but cloud-connected home appliances. It'll be an iterative process, and slower than phone/VCR adoption because fridges and the like are slowly-replaced so-called "white goods".
As opposed to ordinary users running their own services?
The only reason I don't see more routers with default passwords is that people will use whatever comcast and the like provide to them, and that provider will randomize passwords.
If people cannot even configure their router, how are they going to configure self-hosted solutions?
This has two causes: the prevalence of tea and the availability of 220V 10A kettles in the UK vs 115V 13A kettles in the States. Almost double the power means a boil in little more than half the time, making the electric kettle insanely useful.
New kitchens in the States generally have 20A outlets[1] so why won't anybody sell me a 115V 18A kettle? Make one for me, I will give you lots of money! The fact that it will have a funny plug is a feature to help you sell it by emphasizing its uniqueness and help you sell it, not an obstacle!
1: https://www.homedepot.com/p/Leviton-20-Amp-Commercial-Grade-...
Smart locks, smart lighting, smart ovens etc may save you seconds at best, and redundancy (code/physical key access, backup light switches etc) are much less obvious, and may make the product less aesthetically appealing, not to mention the privacy/security issues that have been plaguing the IoT market. Perhaps we are, as the author suggests, yet to discover the areas where smart devices will eventually become successful and ubiquitous, but using his approach of finding such areas fails me.
IoT today is like the tablet market pre-Ipad. Eventually, a big company (maybe like Ikea, their new smart bulbs are very good and cheap) will figure out a proper skookum IoT thing that actually solves a real problem.
Another good example is electric windows in a car. They're slower than manual cranks, break down far more often, yet electric windows have become ubiquitous in cars.
But the point is that this is the same as the IOT market: at first glance the drawbacks seem to be worse than the benefits so there are lots of naysayers. Yet there are benefits, and stuff sells.
How?
My kid is too young to open car windows as it is but I grew up with parents who owned a car without electric windows, and the next version of their car had electric windows (both Saab 900). I could open the back window in both cars, but only slightly (mechanically very limited; see pictures).
Now, kids can't control the front seat windows either way. If you close the back seat windows while they hold their hand out, it gets crushed. If you tell your child they may open or must close the window then you give them responsibility. If they don't take that responsibility you can act upon it with consequences (I suggest minor consequences at first sight though). So I think it very much depends on the implementation of the back seat windows.
As for IoT/smarthome market, when Nest (= Google) acquired Revolv they disabled the $300 hardware soon afterwards [1]. That serves as a warning what cloud computing does for you when the overlords decide they no longer want to support your product. At the very least you need an offline backup. Same with smarthome depending on an Internet connection or wireless connectivity (I wonder if vulnerable to KRACK, too).
[1] https://en.wikipedia.org/wiki/Nest_Labs#Intentional_disablin...
I used to be skeptical about that as well. However, in the Netherlands, where paying by card is ubiquitous, we recently got the ability to simply hold those cards near the machine (contactless) rather than inserting it and having to enter a code. Saves you mere seconds, yet it's a massively better experience that quickly saw mass adoption.
But the lock doesn't have to be super smart to fix these things, or at least most of them.
> Smart locks, smart lighting, smart ovens etc may save you seconds at best
It's not necessarily about saving seconds but making these things easier to use. I liked the article's example of using a smart oven because everyone I know has a digital programmable oven but none of them ever use the extra features. I love setting up my oven to cook something while I'm gone, timed to finish around the time I'll arrive home. Make that feature easier to discover / use and perhaps more people will experience the same joy.
I've found smart lighting to be far more practical than it sounds at first glance. I can't get too excited about having lights do things when people come or go, tho I do use those features... but having bulbs individually controllable lets me create "scenes" to provide the ideal lighting for the activity taking place. My TV Viewing scene turns off all the bulbs that reflect on the screen and dim the rest -- a feat that I could not accomplish without re-wiring every fixture in the room, adding dimmers, etc. For the kids at night we turn off most of their bulbs but have a few color-changing bulbs that turn red and go dim as night lights. Etc.
And speaking of kids, when Alexa says it's time to shut off the electronics and go do some chore they actually go do it without the whining and complaining!
I have some battery-powered PIR LED lights in my bathroom that turn on automatically. Very useful (especially at night), they don't need to be connected to the internet, and you can buy them right now.
Edit: These ones: https://www.mrbeams.co.uk/lights/universal-light/ (other brands are available)
I wonder why presence (as opposed to motion) detectors are not more common. Seems like a few ultrasonic distance sensors would do the trick.
For example, automating an irrigation/fertilization system in our small backyard garden for those summer days we're away seems a pretty good idea. But, at least for me, caring for a garden asks for a little more that keeping it alive automatically. Of course, that not maybe the case for many people and this is a grey area. But why automate, say, the toilet flush or soap dispenser unless you're in a hospital or such environment?
Many of these solutions sound like the next startup pushing something in the hopes of leveraging a hype into profit.
And I also find it a little uncanny that this kind of articles make no mention of security or privacy concerns.
You mean the voice assistant that literally failed 3 times during the live presentation?
I cannot imagine many things that I would like less in my house than voice-assisted smart devices. Huge spiders, possibly.
There is a relevant difference though. The button to adjust the mirrors in my car carry no significant disadvantages, other than a very nebulous and easily-dischargeable moral hazard. (That is, even if you are worried about the "laziness" of using a button to set your mirrors, you can easily negate this simply by using the time or effort saved on some other worthy goal; even my great-grandpa couldn't really argue with that.)
But a lot of these "smart" devices carry several non-trivial disadvantages: Almost every one of them is actively spying on you; a non-trivial number of them are smart for the sole purpose of spying on you because there is no other current economic reason for the device to be smart. They generally add a dependency to an external cloud service, which in many cases has a shorter expected life span than the device itself. I'd submit that corporations would be far less enthusiastic about "smart devices" if they were not planning on abandoning them after a couple of years, and instead had to book a 10-20 year liability on to the books to account for future support, even just security updates for any network-attached devices that don't hook to a "cloud". They add interface complexity to what is often a relatively simple device, at least prior to its smartification. Some people may love their "smart lightswitches" but there's just no way to beat the light switch in terms of complexity.
I don't think the "you're just the old fogey of the future" argument here, along with frankly being a bit audience-hostile, works. I'd submit as further evidence for this that a lot of us who are most worried about all the smart devices and most resistant are the neophiles who have been surfing the cutting edge for a long time. I've been a neophile for a long time, and I can present evidence that I'm still a neophile in other contexts, but whoa, nelly, I'm not filling my house with this stuff. Even my phone and I have an uneasy relationship at times.
(I used to be excited about the thought of having a home robot for various tasks. But now I can expect that home robot to be hooked to the cloud, and literally spying on everything its sensors can get at, which is everything in my house, and turning the full "nudging" power of every company involved into manipulating me and sucking dollars out of my wallet with every scuzzy trick anyone has ever thought up. It'll bring the non-ad-blocked browser experience into the real world. I'm much less excited now. Perhaps I'll be able to afford to pay extra for the ones that don't do that, but we're still talking a social problem here for those who can't.)
Let me end with a re-iteration of the fact that the whole industry excitement about "smart" is almost certainly entirely predicated on those industry's fully-justified belief that they can toss a product out into the world, and abandon it the instant it ceases to be useful for milking their userbase of advertising dollars, which is "immediately" for some things like light bulbs. If the industry had to account for long-term support, the industries would be singing a completely different tune... how could a smart lightbulb carrying 10- or 20-year liabilities, often with a non-trivial black swan chance (think Mirai here) of requiring very swift and very serious reworking of the firmware, possibly compete with a dumb lightbulb where the companies only carry very calculable and relatively brief warrantee obligations? There's a non-trivial way in which this excitement about smart devices is predicated on screwing customers naive enough to buy these things.