I No Longer Have Any Trust in the Nest Protect | Hacker News Reader

I No Longer Have Any Trust in the Nest Protect

phoronix.com

81 pointsprotomyth63 comments

Some things are fine just being "dumb" products. Smoke alarms have worked (well, too!) for years long before Nest arrived, adding likely thousands of lines of code to a device that literally only has to beep when it detects smoke or CO.

I have a $20 hunk of plastic on my ceiling. It beeps when it detects smoke, stops when I press the button, and requires battery changes once or twice a year. It doesn't have a smart phone app, wake me up in the middle of the night unnecessarily , cost $100, or add a ridiculous amount of complexity.

This wave of disruption is getting long.

I have a $20 hunk of plastic on my ceiling too. It beeps whenever I cook a bunch of things, won't stop until I get out the stepladder and literally press a button on it, and when it needs a new battery it helpfully lets me know by beeping just infrequently enough that I can't tell if I'm imagining the beep, or tell which smoke detector it's coming from.

The Nest seems like overkill to me, but let's not pretend that old-fashioned smoke detectors are some kind of awesomely perfected tech.

I disagree, these 'dumb' smoke detectors are spot on.

Learning to cook, ceiling height regulations and hallucinating aren't problems a smoke detector is trying to solve.

Also, if it beeps when there is smoke (from your cooking), it is doing it's intended job.

Fires from cooking (2013) take up nearly 50%!! of residential building fires.

SEE: http://www.usfa.fema.gov/data/statistics/

its doing its job. not being able to cook without burning stuff is not its problem. you will thank it when you leave stuff on the stove and it stops you from burning down the house

Dumb smoke + CO detectors aren't going anywhere, but neither are smart ones. Networking the devices together is a huge boon in larger homes where a few closed doors and space can make an alarm less likely to be heard (such as when you're sleeping, and it's even worse if you're a heavy sleeper). They'll improve with future generations and the price will continue to fall. They'll still cost more than the hunks of plastic, but that's OK: consumers can decide for themselves whether the increased cost is worth it for them.

I had a "smart" smoke detector that's older than nest. It would connect to other smoke detectors, and tell you "there is a fire in the garage". Or any other room or building. Which you wouldn't get from a regular smoke detector.

It's not really for residential use, but there are definitely use cases for remote smoke detectors.

Nest's primary strategy with the Protect was never to improve the user experience of the smoke/CO detector. Like you said, they are devices that you ideally have zero interaction with until the day they are required - at which point they mustn't fail.

The whole point of the Protect was to get a speaker, microphone and some sensors in every room in a house. It's an obvious requirement for any company with ambitions of winning the connected home. Tweaking lights and thermostat settings from your phone is clearly a stopgap solution. The endgame here is Star Trek style interaction: "Computer - turn the AC down a few degrees."

It was a brilliant ploy and would have worked perfectly except they flubbed the core functionality with various bugs. Normally, that's not a big deal with consumer electronics, but the market is very unforgiving when safety products fail.

This happened to me a week or so ago. Unlike the OP, I searched for "Nest Protect false alarm" and was sent directly to this helpful page: https://nest.com/support/article/Nest-Protect-sounds-false-s...

After blowing the visibly-accumulated dust out of the interior everything was back to normal.

I wonder if the OP uses similar debugging tactics on servers that send alerts? "Hmm. 'CRITICAL: disk full' alert? Guess I'll hit it with a hammer until I stop getting paged."

Did it happen to you at 3 AM with your dogs barking?

Honestly I'd have gone the sledgehammer route too.

Yeah, the difference with the disk full alert and the alarm is that the alarm is emitting a very obnoxious noise continuously while you're trying to debug it; and in the meantime, not actually providing a useful service, like the server is. Trying to debug a problem at 3 AM when you just want to go back to sleep because there isn't anything actually wrong, just faulty alarm hardware, is not nearly so useful as clearing out some space on a full disk when it is in active operation.

I've never seen a Nest product in person. A quick search [0] makes it look easy to replace the batteries in all models. Can't you just remove the battery plate and pop one out quickly before debugging?

[0] https://nest.com/support/article/How-do-I-replace-my-Nest-Pr...

In the industry, we call it "percussive maintenance"

Here's a good video demonstrating a similar experience:

https://www.youtube.com/watch?v=BpsMkLaEiOY

It is amazing how many of his attempts to silence them fail, until he finally gives up and packs them away in multiple coolers in his garage to insulate away the noise.

With a little bit of editing Brad could have made a fantastic terror short film out of that video.

I felt like https://www.google.com/search?q=I%27m+sorry+dave all along.

I've heard from prior owners that the depicted experience is authentic and very typical.

Back when I worked at Best Buy in high school, I heard this basic story -- "A single unit of a particular product went bad, and therefore this product/brand is untrustworthy and bad, and I will never buy it again." -- from probably half the people who walked in.

Sony/Toshiba/Panasonic/Dell/HP/Compaq/Apple/Canon/NEC... seriously, every brand of anything, there was someone who was prepared to swear up and down that it was all junk, because one thing broke on them one time.

There's something interesting about that, to be sure, but I don't think it has much of anything to do with Nest.

It's slightly different when a product is supposed to be significantly more reliable than consumer entertainment. And even in that space, where there are multiple similar options, it's not crazy to use the "I've had [good/bad] luck with brand X" as one of many shopping evaluation metrics.

A single unexplained false positive is rightly enough to ruin an end user's confidence in a smoke detector. And the rate I hear about Nest Protect failures is enough to shake my confidence in the product even without experiencing it first hand.

> single unexplained false positive is rightly enough to ruin an end user's confidence in a smoke detector.

Wait what? You want false positives. You definitely don't want false negatives.

> You want false positives

Absolutely, 100%, no. This is terribly wrong. I work for a company that manufactures smoke detectors, and false positives are almost as dangerous as false negatives. False positives encourage people to disregard the alarm as "just a false alarm", and ignore it, which can, and has, led to people die needlessly.

Also, if one of our products woke people up in the middle of the night because of a false alarm and they were unable to silence it, I would join them in calling it crap.

I agree, but in this case we're not talking about reasonable false positives (e.g. toast burning or candles being blown out). I wouldn't put up with a £10 smoke detector waking me up in the middle of the night for no discernible reason, much less one I paid £89 + shipping for.

If a new TV breaks in normal use then o well. If a new knife breaks when chopping vegetables then most people are going to call the company crap.

Smoke detectors fall under the second category where the is basically no excuse if it fails.

There are asymmetric costs with smoke detector failure. A poor false positive and you are awoken in the middle of the night. A poor false negative and very well could be dead.

Because the detectors aren't perfect, you want it to make one of these errors much more often than the other.

This was more than just "a single unit went bad"... it was "a single unit went bad and there was no user-friendly way to stop it from disrupting my life". It sounded like if the damned thing had an "on/off" switch on the back that worked, he'd have been much less dissatisfied.

It's simple, designing a smoke detector for convenience is a really bad idea because it adds unnecessary complexity to something that has just one purpose: to save lives. The primary design pattern is failing safe. In the case of warning devices, this means loudly and spectacularly so that ignoring the problem is more unpleasant than fixing it. Convenience features are all centered around preventing this.

On the other hand, what sounds like a feature, one unit triggering all the others, doesn't add anything in a residential setting because smoke detectors are required to be really loud and annoying and awaken people when their is a danger of smoke [1]. If building wide notification matters, there's no substitute for hard wiring.

[1]: Mortality from dwelling fires is largely from smoke inhalation not burns. That's why smoke detectors rather than heat detectors are used.

You are the most correct and relevant person to have commented so far. Adding consumer-level smarts to safety-critical devices, is a very very bad idea.

I do like hard-wired one-triggers-all in residential settings, but my house has three floors and if I'm soundly asleep in the top floor, an alarm going off on the ground floor won't wake me up :)

Nest seems to have a big problem with false alarms. There's an article in the San Francisco Chronicle.[1] It's so bad there's a class action lawsuit against Nest.[2] There's no reason a photoelectric smoke detector should be so troublesome. (Nest is photoelectric only - no ionization detector.)

Photoelectric detectors are supposed to automatically compensate for dust accumulation. See UL standard 217, section 9, "Automatic Drift Compensation for Smoke Sensing". Someone should take a Nest smoke detector apart and figure out how they do their compensation. Here's a paper on drift compensation for smoke detectors.[3] This is a solved problem.

Maybe Nest got too fancy and tried to handle it in the "cloud".

[1] http://www.sfgate.com/business/article/Nest-smart-smoke-dete... [2] http://www.girardgibbs.com/nest-smoke-detector/ [3] http://www.systemsensor.ca/es/docs/guides/A05-0340.pdf

So, let me get this straight:

* One unit sounds a false alarm (because normal smoke detectors never ever do that)

* The rest of them relay the alarm

* He then takes a sledgehammer to the unit that generated the false alarm

* ..And because of this, he has "no trust" in the nest protect.

And this guy writes for Phoronix?!

Yknow, downvote me if you wish, but that post just comes off as remarkably ignorant, with a side of "all that newfangled tech these days..." ludditism subtext. And the fact that he took a goddamned sledgehammer to the smoke detector is just the icing on the cake. It was stupid behavior, as is the suggestion that we take that post seriously.

Devices fail. Find a person, any person, and they'll tell you that they won't use X brand of thing because they had one unit of said thing fail on them one time. This is that fallacy, writ large, but less forgivable because this person writes for a tech blog of some note.

Just to counter his anecdote, I'll give mine. I have two hard wired Protects, since they were launched. They've gone into their low threshold alarm modes twice from cooking, and both times I've had my kids silence them. They know how to "go wave at the alarm" because they're around when I test it (monthly) and I tell them about it. I've never had any unusual behavior, I look out for the green "everything's okay" response every night when I turn off the lights, and I've been very happy with the Nest Protect.

The Nest Wave feature has been disabled since April 2014 https://nest.com/support/article/Nest-Protect-Safety

Guess that goes to show you how long it's been since I've had a false alarm. Makes me feel even better about it.

I can attest to similar feelings during multiple middle of the night false alarms with dog barking and kids crying. I've followed online support articles including regular dusting and all units are hard wired. I've spoken with support on multiple occasions and exchanged 3 of 5 units with Nest. The units have improved over time with software updates, but continue to be problematic. During one recent false alarm we were not home and the neighbors called the fire department who had to break in to verify there was no fire and disable the alarm. I contacted Nest support after that incident and they were unable to determine the cause of the alarm. We never had a false alarm with the previous dumb detectors.

To be fair, getting some "false positives" is quite common even with normal, dumb smoke detectors. The real question here is how frequent such behavior occurs. Personally, I'd happily tolerate some false positives if in exchange I can be sure that the alarm will always go off in case of real danger.

Also, when working with an imperfect measurement device the engineer designing the system will always be forced to choose an operating point that fulfills the desired characteristics of the device best. In case of a smoke detector the most desirable characteristic is probably that it ALWAYS goes off when there is some smoke in the room, so choosing a working point that might incur some false alarms but that will never (or only under very unlikely circumstances) create a false negative is still the best choice. Of course there are other strategies for dealing with failures in a given component (e.g. by using several, distinct and uncorrelated sensors), but those tend to increase complexity and cost as well (after all, no one wants to spend a fortune on a fire detector).

Smoke detector behavior is heavily regulated. Much of it is not in the engineer's power to decide.

On the other hand, if it goes off when your self-cleaning oven is cleaning itself, it might be too sensitive

Lots of people are throwing around the words "false positive" in relation to dumb smoke detectors, and I suspect a lot of you mean "the alarm sounded, but I wasn't about to die". If it detected smoke, it told you there was smoke. That's a true positive :)

While visiting my parents this summer I realized they had no working smoke or CO detectors. I ordered some 10 year lithium units on Amazon (not Nest Protect). Two weeks after we left one of the smoke detectors went bananas and wouldn't stop beeping. My father had to pull the "permanently disable" tab to get it to shut off.

My point is no technology is perfect; not even normal "dumb" smoke detectors.

My point is no technology is perfect; not even normal "dumb" smoke detectors.

There is far less to go wrong with those, however. Many are based on a single analog IC with no microcontrollers or firmware at all.

We've had interconnected smoke alarms for years in our houses and it has been my experience that what's describe with the Nest happens with every brand of interconnected smoke detector I've ever had. In the last 20 years or so we've had at least a dozen alarm events, many at night and during the winter where all of them will go off. It turns out that there are two sensor types, optical and ionizing radiation. Both are susceptible to interference by bugs, specifically small spiders as well as their waste products and webs. Moths also can cause a problem if they get inside a unit and cause dust to be kicked up. I suspect false alarms happen more in the winter because spiders and other critters might seek out the additional warmth at night that the units provide. Typically when our alarms go off, I trace it back to the originating unit, pull it down, disconnect it from the mains and take the battery out. If it happens in the middle of the night I'll wait till morning to take it out to the shop to blow it out with an air-compressor. I won't buy any unit that doesn't allow me air-gap the power - that's why I walk right past the Nest units at the local hardware store. To mitigate this problem with my existing detectors, I've started lining the inside of the units with double-sided sticky tape when they have a false alarm event. My hope is that it will trap insects. It seems to work, but your mileage may vary.

Doesn't poking a vacuum cleaner nozzle at them once a month / week work?

Two weeks after we left one of the smoke detectors went bananas and wouldn't stop beeping. My father had to pull the "permanently disable" tab to get it to shut off.

This is, of course, the correct failure mode for a smoke detector. If something goes wrong with it you want it to go crazy and beep a lot. Not sit silently doing nothing for 10 years.

If something goes wrong I want it to email me.

I have pulled out and disassembled two generations of landlord-installed smoke/CO detectors, because waking me up at the random moment they decide their battery is low is just not OK.

Reading the title, I expected to read about some deep technical design flaw or a bad direction decision or something...

Cache: http://webcache.googleusercontent.com/search?q=cache:1sJ0j7O...

A false positive is better than a false negative. And OP's actions have presumably precluded any further diagnostics.

I think once it'd had the heck smashed out of it with a sledgehammer, it's out of warranty and Nest/Google/whatever won't be in a hurry to RMA it.

Which is not to say I'd have necessarily done anything different, or that it was OP's responsibility. I'm just saying it's not very helpful.

It could have just been something daft and it just needed to be blasted clean. Or it could be something serious and systemic, in which case getting the manufacturer to take a look would have been wise.

Still, if my whole house was turned into some sort of chamber of loud alarms in the middle of the night, I don't think logic would have been my primary response either.

I think the Nest smoke detectors are overengineered pieces of shit. The dumb whitebox kind you buy from Walmart are well understood, you know where you are. If any one goes off at random at night you push the silencer button, throw the thing out and replace it for very cheap. But they won't turn the whole house into Pandaemonium and tempt you to fetch the sledgehammer for an exorcism. It's another instance of Chesterton's Fence.

This space is just bizarre in general. The fact that a smoke alarm was Nest's #2 product was strange given the advantages over standard ones are marginal. Then they did a poor job executing (remember accidental wave off...). In the same space, all these VCs pumped butt loads into this mysterious company, Leo, who was making an audio/internet bridge for smoke alarms (I am sure they had more eventual value planned but that was the pitch to the consumer). I think they laid off over half the team when reality set in.

I had the same thing happen to one of mine recently - I put it in the microwave (not on) to silence it. No smoke, no fire, nothing - and it went absolutely bananas.

>Once smoke reaches emergency levels, Nest Protect can't be silenced- by regulation in the US and Canada. Address the source of the problem, and Nest Protect will quiet down when the smoke level gets lower.

https://nest.com/support/article/Nest-Protect-Safety

Due to their nature smoke alarms have to fail in the most annoying way possible, and be difficult to disable. I decided a couple of years ago to sleep by a window and take my chances. Now I can eat steak without that cold bead of anxiety sweat as I wait for the cooking detector to go off.

"and be difficult to disable"

Difficulty to disable doesn't really benefit you in a fire.

Having a disabled fire alarm doesn't really benefit you either in that case.

> Due to their nature smoke alarms have to fail in the most annoying way possible, and be difficult to disable.

Why? Why not make them so that you can easily press a button and disable them for X amount of time universally? The one in my apartment has no silencing mechanism and trips every single time I am cooking on high heat. I literally have to rip it off the wall to get it to stop, and the result is that I now cover it while I cook which is probably the most likely time an actual fire will develop. Yes we really don't want false negatives, but if something designed to protect against a statistically unlikely event throws inordinate amounts of false positives with no recourse for silencing, people will just disable it.

My unit had its path light on all night and wore out the battery's on a nightly basis. It went from full battery to chirping in less than a night.

Support wasn't very helpful, so I just disabled that feature and haven't had an issue since.

Some detectors respond well to blowing fresh air into them when they misfire- holding them in your hand and swinging swiftly back and forth in fresh air. I don't know about these devices specifically.

That only helps when there is a cause to the false alarm, steam from a shower or cooking smoke typically.

They can also trip with things like dust and some other ways of fouling the detector or circuitry that I haven't yet discerned. Again- different system, but I've had success.

I got to stay in a house that had a nest thermometer. The AC compressor happened to be broken and the nest perpetually said 1 hour until cool. Nest isn't smart enough to detect a broken AC.

Some guy had a problem with his Nest Protect. Why did someone rush to put this on HN? Will we need to hear about his car problems, too?

Google Derangement Syndrome guarantees that all such articles reach the top of HN.

I've never had any trust in the Nest Protect, because I've never used it.

I wonder what Nest Protect is using to detect smoke that it false alarms, in a pot with a lid on it, without any smoke anywhere in the vicinity.

Unsmart smoke detectors use minute amounts of radioactive materials (such as Americium-241[1]) and almost never false, although they can be annoying if the source of smoke is otherwise legitimate.

My chemistry professor in college pointed out that a big stack of smoke detectors in a hardware store was not regulated and yet would collectively emit radiation that would probably exceed legal limits for people walking by. I thought about that the last time I was in Home Depot and there was a huge stack of $10 smoke detectors. (Yep. I bought three.)

1. http://www.world-nuclear.org/info/Non-Power-Nuclear-Applicat...

Reply on news.ycombinator.com