Nest Thermostat Glitch Leaves Users in the Cold
nytimes.com
nytimes.com
The furnace turns off almost immediately after changing the WiFi key once the Nest loses connectivity. At first I thought it was coincidence but now that I rotate on a strict schedule it's a joke in our office. However, what's not a joke is what happens when the internet goes down. So does our heat.
What the hell, Nest?
Relevant: https://twitter.com/internetofshit
Which hit me on the 8th at exactly 9:29AM. It was FREEZING in the house. Dead nest. Took me about 10 minutes to look up the issue and 5 to hook up common.
Any code that touches battery or recharge methods should throw a red flag immediately and have a 1 month wait time slapped on them. Also, there should be a dedicated Test Nest for each release known to be "in production".
It's that the software was not thoroughly tested. Nest cut corners in R&D.
Consumers are also at fault for not taking the time to understand the hack job Nest has done to trickle charge the device when no common wire is available. Those with 'mission critical' needs (babies, elderly, 2nd home) rolled the dice and are as much at fault as Nest IMHO.
As more devices become 'smart', the companies that invest in design & testing will become more trusted. But also, consumers need to appreciate that a device is only as smart as the infrastructure it's connected too. Any limitations in your house need a fallback plan.
(I'm actually glad that trickle charge hack has come home to roost)
I can't disagree with this more. When you are trying to replace a thermostat, the onus is on you to do right by your customers. And that may be (is!) difficult. But the customer is never at fault for trusting you, but you can be at fault for abusing their trust.
Most thermostats are trustworthy because they are simple. I bought a fancy thermostat back in the early 90s that had a small LCD screen on it and it would run a schedule if you set it. I eventually replaced it with a regular thermostat because it was a pain in the ass to adjust the temperature with button pushes and the hall that it was in was dark most of the time. This is the extent of customer trust given to a company.
The ability to control the Nest with a dial and from my shell on my computer are great reasons for me to trust the Nest and buy it. That said, I'm also cognizant of the fact it needs to connect to the Internet to allow control from the CLI, do updates and set the schedule correctly.
The onus may be on a company to do right by its customers, but the fact is that mistakes can and will occur if the requirements are such that a remote update and control system be in place for the product.
I think @radiorental is correct that customers don't take the time to understand that having a device which provides X features comes with Y risk associated with it. Nest will always be at risk for "being at fault" for a problem that will occur again in the future. Customers need to understand that a time savings on one side may mean a time cost on the other, when Nest isn't able to cover it for them.
I might point out that anyone who disagrees with this assertion assumes that Nest should not make mistakes. Assuming a complex system is perfect, or can achieve perfection, is irrational.
Nest, in an effort to widen the install base went out on a limb. And that's fine but a little research would reveal not insignificant user feedback with installation issues
I'm putting some onus on the customer to understand all the pros & cons with a given product. With the state of smart devices in the home at the moment, homeowners are effectively system integrators. You can't have your cake and eat it.
Would you put t-mobile at fault for selling a phone to a user who didn't check the coverage map first?
Do you read Amazon reviews for any given product you purchase there, or assume all manufactures are selling the absolute best mankind can produce?
The Nest is not fit for installation in a huge portion of homes. The trickle charge tries to side step lack of common wire. When it works it's fine, but it's not hard to find reports of weird installation problems.
It sounds like they could have just added some generic code that prevented the device from going into deep sleep.
The first assumption was addressed by several heat and air contractors on Youtube, so I hooked up the common 24v wire to ensure it would charge and it immediately started working again. This required me to hook up the wire on the heating unit itself, and the Nest. Given I've worked with electronics and have built parts of houses, done wiring, plumbing, etc., it was a 5 minute job for me.
The Nest was built to provide a feature by which it could charge itself off the latent, low power current that runs through the fan wire while it is running. This is one of those cases where more features equals more things to break when you push.
But it's just beginning. Give them the ability to interact with other connected objects, and your potential problems proliferate. Add a smart garage door opener, and you may not figure out how that's connected to your thermostat cranking the heat too high, or who's to blame.
Seriously — listen to this: http://iotpodcast.com/2015/12/episode-36-a-total-breakdown-i...
Any non-hip, non-fancy, non-SV thermostat control system doesn't have these bugs.
Sure, those systems don't come in fancy design, but they will work offline, they will work in every edge case, and they'll cost twice of what Nest costs.
Treating IoT devices as if they'd be consumer software, with updates every week and constant internet access has never worked, and probably never will.
is this just a matter of complexity growing beyond
its masters' capacity to manage it?
Back in the 1990s if Microsoft released a new version of Windows they would test compatibility with other vendors' software - and often adapted Windows to maintain compatibility of vendors were relying on some eccentric behaviour.
What you do is build a test lab. You buy 100 routers covering a range of chipsets and manufacturers. Plug them into a network and a software-controlled power switch. Every time you do a revision of your hardware you keep a few back for the test lab. Then every time you do a revision of the software, you deploy it to all the hardware, and you test it one at a time on every one of the routers. Customer's having problems with hardware so old you can't get it for your lab? Ask the customer for theirs, and offer them something better in trade.
You do the same with furnaces - although you probably need many fewer units in your test lab because the interfaces are a lot simpler.
The main reason not to do this is cost - a company might think spending $200,000 on a test lab has lower returns than spending that money on adverts or product discounts.
That's very odd. My Nest functions fine during Time Warner outages.
As a couple of other folks have noted, that's not normal. You might want to get on tech support with them.
My Nest just keeps running its last program when it looses internet.
Seems pretty damn similar.
Criticizing Nest for having tech support is just silly. Much better to criticize it for the actual stupid failure being experienced by the original commenter.
this also allows you to finally get around to rolling out eap so the scourge of shared wifi keys can die out :)
* WPA2-Enterprise is not universally supported. (Ferinstance, no Chromecast can connect via WPA2-Enterpise. :( )
* Which EAP? EAP-TLS? EAP-MD5? [0] EAP-MSCHAPv2? Or perhaps EAP-GTC? [1][2]
> this also allows you to finally get around to rolling out eap so the scourge of shared wifi keys can die out
Unrelated to the issue at hand... I really wish OS X and Windows supported EAP-WFA-UNAUTH-TLS (and that EAP-UNAUTH-TLS would become an "actual" EAP method, rather than languishing as a vendor specific extension. Securely encrypted zero-configuration public WiFi can't come fast enough! :)
[0] God no! ;)
[1] It probably sounds like I'm trying to imply that WPA2-Enterprise is too complicated. Frankly (speaking as a guy who's implementing his own RADIUS server as a hobby project), it's really not. There are just many, many options available to you... and not all of them are supported by all RADIUS servers.
[2] Anyway... given what very little I know about EAP-GTC, I would suggest that -if we're going for a near-zero-configuration deployment- it would be the most suitable for the situation that we're talking about... The trick would be hooking the verification logic for the security token in the device into the RADIUS server. :)
EAP-TLS ideally. You need a CA infrastructure, a radius server, and a way of doing automated enrolment, revocation and install of client certificates. A lot of work if you haven't already got those parts (though many places already do), but it's fairly hassle free and secure once up and running.
For bonus points you can extend 802.11x goodness to your wired desktop ports as well.
You probably need at least one other network/BSSID with standard wpa2 for non-managed devices as you say.
* OS X machines
* Android devices
* A Nest Thermostat
> For bonus points you can extend 802.11x goodness to your wired desktop ports as well.
Though, that only works if you have either a Smart or a Managed switch that understands that it needs to tag frames from that port with RADIUS-dictated VLAN tags, right?
Android - MDM (I think the stock google apps MDM can do client Certs and 802.11x network config)
Nest - would not bother; stick on a dedicated standard WPA2 network, segregate from rest of network then leave well alone.
Wired network - yes, you'd need a managed L2 switch (but then you'd need that anyway to trunk multiple tagged vlans to your APs).
Spent a weekend troubleshooting, Nest ultimately sent me a replacement, which I was sure wouldn't fix my issue, but it actually did! Bad chipset.
Then I had the common wire issue. Then I fixed that.
Then 6mo later I spent a few hours on the phone with support before I found out I actually had something in the furnace fail over the summer. I was amazed it was not a Nest issue.
Also, my Nest loses its internet connection when it goes into the infamous "power save" mode, which Nest insists means your access point is not 802.11 compliant, but I have the certification from Meraki that their APs are fully compliant.
It's been a fairly colossal pain in the ass, even though at least one issue was not Nest's fault.
From cruising the forums, they just break shit way too often, don't give a downgrade option, don't allow you to choose whether or not to apply software updates, and.. just generally treat an important home appliance like it's a cheap toy.
I worked at home which means the smart scheduling was useless but the aesthetics and ease of use were great. Not to mention turning it on/off while away from the house was nice (I know, not limited to Nest).
Predictable. Avoidable. Pitiful.
It is by far the most fragile device from a network connectivity standpoint. I'm so happy that I won the Nest or I would be so much angrier at the whole thing.
I wouldn't buy a nest again because their API is locked down and they seem to think they own all the data that they collect in my house and won't let me access it, but they've been reliable for me and I have never had a problem with my heat going out.
Apropos of anything else - I don't think Fitbit has ever tried to market itself as a medical device.
Makes me wonder if it would be possible to intentionally drive a company out of business by convincing a bunch of people to use their products incorrectly.
In this case, would courts find that a reasonable person would assume that a heart rate monitor can accurately monitor their heart rate? My guess is, yes.
If I buy a desk lamp and it doesn't work reliably then you sold me a lemon and I want my money back. It doesn't matter why I need light on my desk.
Remember, I'm talking about situations where the company actually says "not for medical/diagnostic use". They're not trying to get away with selling lemons.
To use your analogy: If I sell you a small LED (say for use as a power indicator on a computer) and you try to use it as a desk lamp, you shouldn't be able to sue me on the grounds that "you should have known I was going to use it as a desk lamp and made it fit for that purpose". That's what seems unfair to me.
They totally advertise it as a working heart rate monitor. If they have a problem with accuracy, they should not be making it. Otherwise it's like making a tape measure that has marked distances within +/- 20% of the real value (i.e. 1cm becomes anything from 0.8cm to 1.2cm), and marking it with a small caption that it's not for construction use. Oh, and having the product website say "it measures distance" and elaborate on the benefits of measurements for building stuff.
--
I'm conflicted about it. On the one hand, I feel like people should be free to build whatever trinkets their like (as long as they are safe). OTOH, I feel like they should totally not be free to sell crap tools.
On a totally unrelated note, the image sensors in optical mice are actually low-resolution (16x16) 2D arrays.
https://pickandplace.wordpress.com/2012/05/16/2d-positioning...
It's not like Fitbit has gigantic text on their web site and product packaging saying "NOT FOR MEDICAL USE." In fact, I challenge you to find any such statement anywhere on their web site. I couldn't, but maybe I suck at it.
They could try a little harder to educate their buyers, here.
You don't do that unless you want people to believe you're selling a working heart rate monitor. So if it turns out the monitor is not usable for its stated purpose, then they either fucked up the engineering, or they're literally scamming people. In both cases the comapny is supposed to either address the problem or close shop. That's exacly what should happen.
I'm curious if there are legal precedents on how accurate a thing must be to advertise as a being a measurement device.
Everything has some level of quality or tolerance expected. Car parts break all the time, it is just a normal part of owning a car. You don't get to sue a car company when your battery dies and you need a jump. You only get to do it when there is a significant or dangerous breach in the guarantees that they actually made, even if only made implicitly. Fitbit explicitly does not rate themselves for that purpose.
With the same logic as Fitbit, my watch tells reads my heart rate. Right now it's saying my heart is beating 9:21am. Sure, that's not as accurate as a real medical device, but it's a number so it's in the right ballpark.
Fitbit's HR monitor fails to provide accurate readings whenever you move on almost everybody. What they don't do is advertise that you have to be still to get an accurate reading. Fitbit is marketed in such a way as to make you think it tracks HR during motion, I'd say it's not the same as a car parts that break at all. It's more like lying by omission.
Not bullshit.
The whole employee/contractor thing is all about payroll taxes. You have to pay them yourself if you get a 1099, but your employer pays them if you're on a W-2. Employers have a much better track record of paying that, so the government wants to avoid too many 1099s.
I've been a contractor before. I got a W-2, but no benefits. Now everyone is happy.
The ecg monitors built into the precor bikes and counting my wrist pulse for 30 seconds and doubling are generally within 2-4 bpm of each other. My fitbit consistently measures 130-ish when the truth is 175-ish per the above two methods (my cardio hr target).
This outcome was predictable from the experience of those early adopters.
EDIT: Why couldn't Nest fall back to being a dumb thermostat? Lots of smart device makers should consider having such a fallback.
The Heat Link has a big button on which lets you turn the heating on and off, overriding what the Nest is telling it. That's not as economical as a regular thermostat, but it means you won't freeze: https://nest.com/uk/support/article/How-do-I-turn-the-heat-o...
Forced hot water systems are very common in the colder parts, and they use the same thermostat wiring as forced air. Some parts of the U.S. even use diesel fuel for heat (it's called #2 heating oil, but it's really just diesel fuel with pink dye added so they can fine you for using it as a vehicle fuel...it's taxed advantageously).
Also, boiler-radiator systems can absolutely work with standard thermostats. The classic design has no electrical control at all, just bimetal valves (like car thermostats) to regulate the flow of heated water. But the modern revision uses the same kind of electrically controlled valves (24VAC, IIRC) that forced hot water does.
All low-voltage thermostats with standardized wiring layouts.
I didn't know the UK ran high voltage to ordinary thermostats though, that seems crazy if it's just a control circuit. Presumably the full mains voltage?
Old U.S. "radiant" (electrical, non-hydronic) heat ran 220/240 VAC to big rheostats in each room, but that was actually the supply to the heating elements. It feels like something that wouldn't meet code today, but it used to be popular in some parts of the country where electricity was inexpensive.
http://www.cnet.com/news/nest-thermostat-finally-arrives-in-...
https://nest.com/blog/2014/04/02/the-uk-just-got-a-little-mo...
The US version won't work with line voltage systems. It looks like the UK version only works with line voltage systems.
This also explains why in the US changing a thermostat is something pretty much anyone can do but the UK version requires a Nest Pro to hook up.
Nest US: "It's easy to install the Nest Thermostat. Everything you need comes in the box"
Nest UK: "A Nest Pro installer should install your Nest Thermostat. They will identify cables from the boiler, zone valve or junction box to connect to the Heat Link."
Though I have no need for a thermostat whatsoever, I kinda like nest. However, I'd love if people went to court with a class action suit against them.
These 'no class action' and 'arbitration' clauses are becoming increasingly common, but I think they probably don't hold up in court - and nor should they, given how much power they put in the hands of the company.
NYT has been on the case recently: http://www.nytimes.com/2015/11/01/business/dealbook/arbitrat... http://www.nytimes.com/2015/11/02/business/dealbook/in-arbit... http://www.nytimes.com/2015/11/03/business/dealbook/in-relig...
http://www.nytimes.com/2015/12/23/business/dealbook/sued-ove...
This can also come up in landlord/tenant situations. If the landlord put in a Nest, and its failure caused the tenant problems, like having to spend the night in a hotel, the tenant can probably sue Nest.
The reason you see it everywhere is because it was deemed legal.
The same thing happened with Dropcam. Once they were bought by Nest/Google they stopped making any real changes for the last 2 years. It's sad and infuriating as a customer.
This seems quite strange to me. Here in Germany it's extremely common to (automatically) turn of heating all together in the night (called Nachtabsenkung). One reason is to save energy, but lower temperatures also help sleep:
http://www.nytimes.com/2009/08/04/health/04real.html
The article says between 60°F (15.5°C) to 68°F (20°) is best.
I have a dumb thermostat. It doesn't even have a clock. It just has two buttons, one two increase temperature, the other to decrease it. In winter, during the day, I put it at 20C. At night, I put it at 15C or lower (which stops heating, as the house temperature decreases to about 17 or 18 degrees during the night, without heating). It works like a charm. We are not cold at night as we use pyjamas and duvet. We also have a baby. In fact, if I forget to turn off the heating at night, I can't sleep well and I often get up to turn it off.
Energy costs also tend to be a LOT lower in the States.
http://blog.sealy.co.za/wp-content/uploads/2013/06/baby-wint...
http://well.blogs.nytimes.com/2014/07/17/lets-cool-it-in-the...
I guess it doesn't get too cold where I am (high of 9 C, low of -1 usually in winter), but my place doesn't have any insulation at all, or double glazed windows, and I use the heater for only 30 minutes a day usually in winter, because I just put on more clothes.
To be fair, the issue isn't that there was software involved; digital thermostats have that. I think more telling was the level of complexity, which allows for a lot more to go wrong. Combined with a poor process behind update validation, this is what happens.
Keep it Simple. If you're going to use new technology, keep the old stuff around, ideally working in parallel as a failsafe.
And generally, anything you can do to distribute the failure points and allow for overlap (such as backup spaceheaters that don't normally kick on) should be considered if someone's life is on the line, or property is at risk. Stuff can always go wrong, but the more you can do to reduce risk the better. This applies to any system dependency; mechanical thermostat or nest.
Startups try to treat IoT devices as a standard arm or x86 processor with some additional GPIO pins where they read info, deal with it, and write info.
No real-time guarantee, a level of QA similar to websites or consumer software, etc.
And here is what previous companies building smart devices (say, SIEMENS) treated them like: As extended microcontrollers.
Put all the critical software on one tiny microcontroller, do static verification that that microcontroller is always acting correct. If you need Internet access, add a second chip that does that, and provides the results of that data to the first microcontroller — but don't depend on it.
There is a large divide between the Engineers designing these devices in the past, and today's crowd.
(I see the commercial and civil IoT possibilities, but an awful lot of them involve being able to monitor and somewhat control things that are physically very dispersed out in the real world. I don't have a house that big.)
A selling idea of the Nest is that it keeps your house warm when you're there, and not-frozen when you're not, and it will do so efficiently, so that the house is warm when you arrive home but not long before.
Now of course you can just turn up the heating as you walk in the door wait a few minutes to heat up, it won't hurt. But if technology can do it reliably, there's clearly some value. Compare, for instance, lighting a candle when you get home with just flicking the switch for those newfangled electric lights. The candle works but the technology is more convenient.
Of course, electricity doesn't give you failwhales these days, and there's a bit more engineering involved than "move fast and break things."
In fact, I’ve actually designed something similar to that, but a lot simpler. A simple barcode scanner where you check in new products you buy, and check out products you throw away.
Your homework, before you start too far down this road, is to start doing that right now if you haven't already. You don't even need a real scanner; just pretend with a stapler or something vaguely the right shape and weight. You should also go through your cupboards and just scan everything once to simulate the initial load. Then, I don't know whether you mean to design this for yourself or for selling to others, but if it's the latter, consider whether your customers will actually do this for any period of time. I'd also recommend faceoffs between two people, one pretending realistically to ask your system where something is (either literally typing it in somewhere or doing real-enough voice recognition that you can see the correct query came out), and one just looking. Even if I don't know your kitchen, if it's at all sensibly laid out and you give me a quick look around first, decent odds I still beat your system finding something on average, even if it's not my kitchen.
If you completely eliminate 100% of the time in a year that I spend looking for kitchen goods that I don't know where they are, you've brought me maybe $20/year in value total; if you make me scan everything going in and out, I'd pay at least $200/year not to do that. (Probably more if I really faced that problem.)
(If this is just a personal project, go nuts. Very educational, lots of valuable skill building involved, and it's always good to scratch an itch with it. But if you have any ideas of making money with it, then I suggest taking it seriously and doing some heavy market research and proofing before the tech.)
The situation was this: Was at home, had no idea what was in the cupboard. So I scanned everything with the app.
I stopped using it, mainly because scanning via phone is far slower than via a real scanner, and I don’t want to spend money. And because my parents always ignored it anyway.
But I’ll start using it after I move out ;)
I tinkered with something like this too in my app; but not for where things are, I already have pre-defined homes for things.
Instead I was more worried about food going bad before I ate it. but I backed away from it, mainly because barcodes don't carry expiration data. Instead, my app knows whats on my shopping list, and send me an email afterwards that lets me +1 my inventory of something, which ties into a general expiration for that product (like milk is ~week) and then sends me emails when something I bought is close to expiring. For milk it's not such a problem, but other things I buy and forget.
Other IoT possibilities I see tossed about have even worse possibilities; Nest is probably already the biggest possible winner. If you eliminate 100% of the time I spend turning lights on and off, you've basically had no impact on my life. If you design a glorious IoT refrigerator that somehow requires 0 additional time out of my life to feed it data (which is a negative) it still has to face up to the fact that my "display" showing me what I have which I can get to by simply opening the door is superior to any practical front-mounted display. There's very little room for any sort of IoT water-use optimizer, certainly nothing a startup could wedge into and make money. What can the IoT do for my washer and dryer, play tunes off of Pandora while I'm loading them? My cell can already do that.
I don't need an Internet of Things. I need a Robot of doing Things, and if it's hooked up to the internet I rather expect we'll still see it as "a robot" rather than "an IoT device". (I don't think I want my robot live hooked up to the internet anyhow.)
And just to be clear, I'll say again I totally get it for commercial and civil use, so I'm not just down on IoT in general. (I'm down on IoT security in general, but that's a separate problem. Sort of. Close enough for now anyhow.) It just seems to me that the vast bulk of the IoT story involves being not physically proximal to the IoT device (and, indeed, note how the core Nest use case of "turning off the heat when you are not there" fits that to a T), and therefore, unless you live in a mansion, it's solving a problem a homeowner mostly doesn't have.
But your argument can be advanced against electric light. I could do just fine with candles in my house. But in 2016 I'm not going to go back to candles, even if electricity isn't really saving me that much time or money.
Imagine in 2050 never having to turn on the lights again because they will turn on automatically and that will work well. Never deal with carrying things and reaching over with your elbow. It won't add more than $200 of value to your life, but will you actively opt out of it?
It's not that "IoT" will make your life vastly more efficient and save you tons of money. It's that when it's easy enough and good enough, it will become the new normal as mains electricity is now. Currently it's a buggy gadget, but so was every technology once.
IoT is about finding some sort of just-shiny-enough bauble - that may or may not have an actual useful purpose - that you find interesting enough to buy without paying attention to the data you're handing over to some company. Occasionally you will find more traditional vendor lock-in plan, but these "smart"/IoT devices usually use some amount of surveillance-as-a-business-model.
Sure, there are lots of possibilities. We have barely begun to explore what is possible with {,inter}networks. I'm sure there really is a lot of very interesting ways to mix traditional products and network access. For now, though, like any new technology, the market is mostly misguided designs, scams, and safety risks.
The difference is one of market strategy. The IoT devices are deliberately low-threshold purchases (cheap, easy to install) because they want to build a customer relationship that leads to extra sales. Nest cam, ring subscription, etc... Traditional home automation products instead are targeted towards a one-time sale of a larger system, so they focus more on ability to design the system how you want, with more standards support and more ability to integrate with other products, but also a much higher purchase threshold and a much more involved installation process.
In the end, you have to choose what you want. Do you want a single gadget or do you want a whole smart home system?
I want a whole smart home system built piece by piece through buying various smart devices of different vendors and having them talk to each other. This is how it's supposed to work for the "personal use" market, and this is what we need to push for.
I worked on an anesthesia machine where the control systems were implemented with micro controllers and RTOS. The UI was Windows Embedded running on a separate x86 board and the whole thing would still safely operate even if the UI subsystem crashed. There were completely mechanical backup controls too.
1. It doesn't honor DHCP lease expiration [1]. It will happily keep using its IP without renewing its lease. I've had situations where it's gone long enough without a lease renewal that another device has received its old IP and caused a conflict.
2. It constantly disassociates from the AP, and decides to come back on the network when it feels like it, not when there is traffic for it. And note, this is with a constantly powered device, so there should be no power saving causing it to power down. Also, the AP is about 6 feet away and there are zero other WiFi issues. Manually kicking the device off the AP causes it to come back immediately.
3. Their backend goes down all the time, usually for 1-2 hours.
4. They had a perfectly working method for turning heating on in advance of your setpoint with the goal of reaching your desired temperature at that time rather than just turning on at that time. Then they broke it [2]. This led to nice surprises like waking up to a completely cold house. Instead of rolling back the broken code, they refused to respond on the aforementioned thread for well over a month.
It's really inexplicable. They have nice industrial design, good brand recognition, no shortage of money behind them, and they screw up at every conceivable turn.
I'm going to try out the ecobee thermostat because I very much appreciate the convenience of being able to remotely adjust the temperature.
[1]: https://community.nest.com/thread/10316 [2]: https://community.nest.com/thread/11438
Ipso Facto. They can't not screw it up.
A critical component of infrastructure should not be IP addressable. It should not be "on your wifi". It should not have a login or a password. It should be as dumb as possible and as simple as possible - and as anti-fragile to all of these events as possible.
It's not possible to have this "feature set" and not have this pain. It's a fools errand to pursue it.
Have a microcontroller that stores the time table for when to set which temperature, and applies it.
Have a second processor, with internet access, that updates this table.
Simple as that.
It is possible. Their competitors seem to achieve it at the same price point to a much better degree (case in point: the British Gas myHome that I own and its newer cousin, the Hive). It helps that they don't shrug their shoulders about software bugs and in general don't roll out untested cowboy code on critical home appliances...
I don't know what you're using as a DCHP server, but the servers I've used try hard not to hand out the same IP address to a second device until after handing out all possible IP addresses. Which means my network would need something close to 256 different DHCP clients before having a problem.
I do have some flexibility at home because I run my own dhcpd on OpenBSD, rather than relying on something in a router. Which means extra flexibility. So I lock down the IP addresses anyway. E.g. I have entries like this in my dhcpd.conf file:
host foo {
hardware ethernet 00:1b:63:95:1c:4f;
fixed-address foo.example.com;
default-lease-time 3600;
}
It's a little bit of hassle to set things up this way, but there are never problems even if devices are stupid about when they renew their leases.
That's not to take anything away from your comments, it's just a workaround that has worked for me. Your entire list seems like annoying problems for the Nest.
I've since ditched my cable boxes and thus the ISP router and now just have untrusted devices in a separate VLAN walled off from my other networks.
Why make things more complicated than they need to be?
And having the ability to change the temperature from my bed is nice too.
My energy bills dropped about 30% with the Nest.
It does a number of neat things to save money, including running the AC fan through the chilled ducts for a period of time after stopping the active cooling.
For a forgetful person like me, it's also very handy to be able to turn the house into vacation mode the day after I left for vacation.
So 60F is surely fine.
Anecdotally, I was once without central heat for three days in 0-20F, and nothing froze. The basement got down to 38.
I had two pipes freeze and burst after 3 days of freezing temperatures (first time anyone can remember being more than 24 hours below freezing). One was an outside faucet, properly protected and insulated. The other was frozen inside the slab, near the wall.
60 Fahrenheit is practically a comfortable day-long office temperature. It's a _long_ way from freezing.
1. Economic - it easily paid for itself in one summer by not running the A/C when no one is home. I have geofencing set up to do this automatically.
2. Comfort - my office/bedroom temperature is all I really care about as it's where I spend all my time. My thermostat was not wired to there, and temperatures fluctuated wildly in my office. Ecobee has remote sensors so that I can set the rooms I care about to the correct temperature.
Overall, it has certainly made my life a lot less complicated; I never have to touch the thermostat again to get comfortable in my office, and I save a lot of money too.
I don't really care for the "learning" feature but the 2 other features I find critical are 1) ability to schedule, and 2) remote control.
It's nothing you really need, but it's nice to have. I can program the temp to go 10 degrees lower overnight, and have the heat ramp up 30 minutes before we get up. I can have the blower run 15 minutes every hour without heat or cooling to make sure air circulates around the house. You could do that by hand but it would be more manual and cognitive work.
While I don't have a Nest, I put a similar timer on my water heater and saved a lot of money ($20/mo).
You could set two pairs of heating times, and control which (or both/none) were used. There was an override until the next set point.
This was 1980s stuff, or maybe even earlier.
The interface was far easier to use than any digital one I've seen.
It actually wasn't super intuitive (in cooling mode it inverted the sense of energy saving mode, if I recall correctly), but it was easy and never failed me once I set it up.
Doesn't make sense, unless you're willing to live with the potential failure of those functions...
The ubiquitous programmable thermostats do different things at night, in the morning, during the workday, and in the evening. The idea being that the temperature is allowed to drift when no one's home, but is brought back in line by the time someone is in the house again.
However they are a PITA to program (tiny cryptic LCD, 4 buttons) and basically useless if you don't follow a consistent daily routine on the weekdays. Nest's value proposition is to be smarter about when the house is actually occupied/going to be occupied. Also when you're traveling (or if you have it in a rarely-used property) you can remotely start the heat/AC when you're several hours away so that it's comfortable by the time you get there. I have none of these problems, so I don't have one, but they do solve a problem that some people really have.
In fact you will save money & energy with manual control because the heating will not turn on when it's cold but you are wearing enough layers.
It seems like a GIANT bug to me that the Nest won't come back online till the battery is almost fully charged.
I "fixed" it by putting our old thermostat back in then I put the nest back a few days later.
But hey it's the Internet of THINGS! woot!
The ecobee is much smarter, mostly because of the remote sensors. My house is far more comfortable. It's also nice that ecobee has an API and IFTTT support. It has no issues running normally when the internet goes down. (My internet was down completely the other night and there was -25F windchill. It followed schedule just fine.)
I'm guessing ecobee coming from Canada they know the value of a functional thermostat...
First problem I had with mine, and it was a great support experience.
Anecdata and all. But combined with this story, makes me wonder if they replaced it because they knew it was likely to happen again.
https://www.youtube.com/watch?v=BpsMkLaEiOY
[Warning: contains extremely annoying Nest smoke detectors beeping. So prepare to lower your volume]
Brad looks very annoyed and it was no fun for him, but on an artistic level it is really well done, it is an epic battle between human and machine.
Both the Nest and Honeywell require the infamous C wire, both are similar installation, and both cost around $200.
[0] http://www.amazon.com/Honeywell-RTH9580WF-Touchscreen-Thermo...
PS - And, yes, the Nest has a theoretical setup without the C wire, but you'll lose key things like fan control, or if the boiler doesn't run enough the entire thing can die. On both it is ultimately C wire or don't waste your time.
Pretty sure my thermostat was sub $75 on Amazon.
And most house insurance does not cover flooding / water damage.
I would guess that probably 85% of "user-generated content" on mainstream outlets like Amazon is the result of astroturfing or other paid schemes. It's practically impossible to compete in any online marketplace without turfing the crap out of every slightly relevant outlet. Your competitors are doing it constantly, and the disadvantage will be really big if you don't play ball too.
One example: our objectively superior product with a great design and great functional foundation was substantially less popular than the dirty rip off that had their site built with cheap oDesk labor and designed with cheesy stock photos, but spent ridiculous amounts of money on SEO and fake Facebook comments. We were substantially cheaper too. There was literally no reason to ever use the other person's offering, and there was no reason for them to show up ahead of us on searches, but they did.
Don't believe Google's line of BS that to get to the top you just need to put out good content. It's total nonsense. What you need are backlinks, which are bought and laundered into a network of sites controlled by the same SEO group, and social media clout, which is bought by paying third-world contractors $1.50/hr to make fake Facebook and Twitter profiles and talk about your company.
My heating system (hot water baseboards) doesn't really respond quickly enough for it to be super useful for things like away detection, but as a programmable thermostat, for the most part, it works better than I expect most software to work.
edit: I'm talking about redundant systems. A wood fireplace, for example.
You should have a backup of any form of technology you rely on to maintain your life. We regularly do it with far less critical applications.
The Nest is so obviously a product designed in warm California—not anywhere that sees truly hard winter. If it were designed well, it would fail safe (furnace on). I'd rather have a hefty heating bill from my furnace running 24/7 than have all of the pipes in my house freeze and break. If I ever install a Nest or other smart thermostat, it will be in parallel to the bimetallic-strip-and-mercury-switch from forty years ago.
You're absolutely right that furnaces aren't designed to run 24/7. But your thermostat can request heat 24/7 and unless the furnace control board is broken, it'll the cycle the system as necessary to keep up with the requests.
Most people never learn this because the systems are designed for peak needs and most days don't get anywhere near peak.
I have a hot water boiler. Even when constantly calling for heat, I can hear the element relays click on and off to avoid over-pressuring. The recirculation pump stays on the whole time.
(I'm agreeing with you; just pointing out that my system does cut out under normal operation, while calling for heat.)
In the event the house is too hot, you could open windows, leave, or if it is above freezing outside simply turn off the boiler directly. If it gets too hot while you're asleep, it's not "dangerous": you would wake up sweating. Granted, infants or the infirm may have issues, but in general they face a multitude of dangers that need special consideration or supervision.
What Nest should have really done is include an old-school analog control mechanism as the fail safe (reed switch, magnet, and a bimetallic strip?) that would maintain a safe baseline (say, 50ºF) if the "smart" controller fails.
It seems like a safety-net would be high-school easy, and at least let them claim they've learnt from their mistakes.
So the hardware is solid enough -- it's just up to the thermostat to not drop the ball.
My heating is controlled by a time-switch that directly enables the boiler in the morning and at night, and as far as I know this is how most people I know operate.
We do have thermostats but they are largely ignored boxes that sit in the hallway and/or landing.
Is this just something to do with living in a cold country? I would have thought a nest equivalent that directly operates the boiler would have been a better device ...
That's why the main thermostat doesn't do much.
And all for some supposed improvement that is marginal at best. At worst, it simply doesn't work; taking over and setting some uncomfortable temperature that you never requested. My old thermostat worked just fine. Set it on exactly what I liked, so there was no need for it to "learn". I literally gave it a specific number. Don't really get the concept of wanting it to second guess me.
When I Googled and found others with similar issues, they seemed to be largely insulted on forums, as if the users were somehow unworthy of the wonderful Nest. They just hadn't invested enough time in understanding what the Nest wants.
The whole thing seems like a hoax--an elaborate parody of the times.
I'm sure it's just a grammatical mistake, but I lol'ed.
We have a lot of problems with the Nest thermostats at work. I've got a 10 year old basic thermostat at home that works exactly as it should. Follows its schedule perfectly.
No "learning" necessary.
The way to look at it is that a company with an arbitration clause might as well not be in the US.
I easily added a few controls to my own home automation system and now I can control it from anywhere and I don't need a 3rd party (other than the internet) involved. What do I like the most? Since I've set it up I haven't had to physically touch it. Programming it through json or an app is easier than the buttons on it. Now if you want a smart/learning t-stat, you'll want something else.
It could not see any wifi ssid's: not mine, not my neighbors', nor my iPhone's hotspot held right next to it. It was fully charged, not sluggish -- none of the problems listed as 'typical problems'.
I went through the menu system and did a reset. I think that was a soft reboot. Still no success.
Then I found you could reset it by pushing it in for 10 seconds until it turns off. It seemed more like a hard reset, and voila -- network back up.
It's anecdotal, but this firmware update doesn't seem to be just about the charging. They seem to have messed up the wireless capabilities until a hard reset is performed.
That is, have a user settable manual switch that fails to "heat on" if there is no response from the main unit.
The house would be quite hot - but probably much less damaging than frozen pipes.
The liability costs for frozen pipes could be staggering.
Does anyone have ideas for cool hardware or algorithms for something like this?
Most units just use bang-bang control around some known limits. Something I've always wanted to try is to have a "periodic recirculation" mode for the times that I'm relying on passive solar heating, just to turn over the stratified air.
> Something I've always wanted to try is to have a "periodic recirculation" mode
Yes! That is one thing I want to try too.
I did an experiment in an office setting during summer where I found that if set on "auto" someone would inevitably turn the thermostat down to 65 to cool some spot that was in the sun or where the air had gotten too warm. When I switched the fan to "always on" I would set the thermostat to 70 and nobody would feel the need to adjust it.
I suppose depending on the cost of running the compressor vs the fan my approach may not have saved money, but it was interesting to see the difference in comfort offered by both.
A bug like this would be financially catastrophic.