Consumer Safety Notice for Nest Protect: Smoke and CO Alarm | Hacker News Reader

Consumer Safety Notice for Nest Protect: Smoke and CO Alarm

nest.com

73 pointsuptown63 comments

Not waving, but choking. Many comments seem to be cutting more slack to this announcement than perhaps they would for other companies' products. Nest says they discovered this for themselves, which indicates that there is some measurable effect and it came up on post-release testing, months after release. Even though some people had been discussing a potential for exactly this scenario when the product features were first announced. Given this not unexpected error condition, Nest really should have run those same tests before releasing a product whose sole function is safety.

Things that were once unknown unknowns are always easy to point out in hindsight...

That is true in general. But in this case, what new information arose in the intervening months? There were no reported cases. Presumably the only thing that changed was that Nest's QA team eventually got around to doing what they were likely too busy to do before release. So this appears to be a classic case of trade-off between safety and time to market. I don't presume to correctly judge what would be the right values to plug into that equation, but I think it is fair to raise the question.

It could also be that their international expansion imposes different requirements, or doing it all under the supervision of Google's attorneys influenced the change.

If you think QA is ever complete, then it is clear you have never worked on a QA team. You can always come up with new ideas on how to break things.

I quite agree that one can always come up with new tests. But the pertinent question is whether in this case Nests's QA would have anticipated, yet not tested for, this particular error condition. I believe that on balance the answer is yes, probably they did. A question which follows then is, assuming I am right, what is a reasonable amount of blame for failing to test prior to release?

Yea, just like the 4 recalls for my Honda Fit. They should have totally tested all those aspects before. And since I'm only paying $100 compared to the $20K for the fit, I should expect better safety measures than the automobile industry. Because the number of people that die in house fires is minuscule to the number of people who die in automobile accidents.

Honda is selling a single purpose safety device with no moving parts called the "Fit"? And it's had 4 recalls you say?

Won't that lead to confusion since they sell a car of the same name? With millions of parts? That moves down the road by controlled explosions with chemicals stored onboard the vehicle?

It's not automatically unreasonable to expect attention to safety comparable to an automobile. Yes, you pay 1/200th the price, but the complexity is also much smaller. I would wager more than 200x smaller. Probably even 2000x smaller.

More generally, I think there is also a sentiment that if a device whose SOLE FUNCTION AND REASON FOR EXISTING is to improve safety, then perhaps safety should be a high priority for the development team? In other words, if you don't want to concern yourself with safety, don't sign up to design a safety device.

The support page for the Nest Wave feature is interesting

http://support.nest.com/article/What-is-Nest-Wave-and-how-do...

>One of the most common things you see when a smoke alarm goes off is people waving their hands, towels or magazines to air out the room. With Nest Wave, that instinctive motion will actually silence the alarm.

A person's natural response to the first sign of danger might silence the alarm, inadvertently, and lead them to believe that they are safe. That is troubling.

However, they have thought about this problem:

>Depending on the situation, you may need to wave more at Nest Protect in order to silence it: Non-emergency situations require less waving, such as cancelling a manual test or listening to a Nightly Promise message. For your safety, you’ll have to wave more at Nest Protect during a Heads-Up or an Emergency Alarm. This helps ensure that you are deliberately trying to silence an alert.

Obviously, the liability load on this product is far more than most other gadgets. I guess that even though they anticipated this sort of issue, a paranoid level of caution is required from Nest

(Makes me think of the attention that Tesla received over one or two battery fires, or that bitcoin receives over company failures. New products are held to much higher standards than 'traditional' products)

It’s very important that you immediately connect your Nest Protect to your Nest Account so we can remotely disable Nest Wave.

Does anyone else find that unsettlingly creepy? If they can disable some feature of their product remotely, presumably via firmware updates or a similar mechanism, isn't it not much of a stretch to assume they can disable the entire alarm remotely?

And then they can get your Android phone to overheat and start a fire! Or slow down your car remotely when the police are chasing you!

Yup, that's the 21st century for you. Most of the cool things we have today are here thanks to being able to be controlled remotely over the Internet. I wouldn't be surprised if the next product Nest comes out with is a security system. ADT has had its run of shitty and insecure systems for too long. It is time for someone to create a better version where you can monitor everything remotely. This will have all the same ramifications: the manufacturer can lock you inside your own home.

The fact that they can do it remotely is not as creepy as how they've worded it and seemingly provided no instruction on how you can do it yourself.

I'll use a security system controlled by google shortly after a pig flies in through my window.

The feature I really want in a smoke alarm is a "snooze" button which I can pre-register. I don't have a good fume hood over my kitchen stove, and the smoke detector is essentially in the kitchen (it's an open kitchen/living room), and the smoke alarm is on the kitchen side of the living room).

I want a way to easily hit a button prior to cooking and not have any smoke alarm (but maybe a CO alarm still) for the next 15-30 minutes. As long as I'm still in the kitchen cooking, I don't need the smoke alarm to tell me if there's a fire.

On the other hand, it's also entirely likely someone could step out of the kitchen for a few minutes, kitchen could catch fire, and bad stuff would ensue if the alarm were disabled.

One option might be a pre-alarm for 5-10 seconds, during which you can snooze it. Unfortunately, if anyone ever died in a fire where 5-10 seconds might have made a difference, massive liability would ensue.

The right solution is to move somewhere with a proper high-flow external-venting hood, of course, and only heat/CO alarms in the kitchen area.

I thought about this for a bit. Best solution i came up with would involve sensing the room a person is in and deactivating it locally, but even then there's the "edge case" of a fire happening "behind your back" (e.g. in a closed stove) and you don't notice. Basically, even though the possibility of a fire going unnoticed will become increasingly small, the cost if it does happen is so high that IMHO it doesn't balance the amortized cost of having to deal with an annoying fire alarm. (unless that serves to "train" you to ignore/disable it... This may actually be an interesting behavioral question)

When I first red the snooze idea I thought the UL guidelines would say that it is not kosher but from a quick scan it looks like a "presnooze" button might be acceptable.

UL 217 Smoke Alarms: https://ia600608.us.archive.org/3/items/gov.law.ul.217.1993/...

Carl Malamud is fighting the good fight. Without resource.org that doc would set you back $400-900.

CAVEAT: That is the 1993 edition. It looks like there is a 2005 update.

I just want something which is 1) not designed by utter morons 2) reasonably modern in UI/UX 3) priced above "contractor cheapest minimum compliance option", with value to match. This seems impossible -- I can get absurd commercial fire alarm systems, or I can get stuff like Nest which isn't really designed for life safety, or I can get the cheapest things from Home Depot.

What I really want is someone like Assa Abloy to make a high-end residential/small office product. Generally I want this for building management, access control, security, life safety, etc.

Like I said in another thread, I'm an engineer for a smoke alarm company; so my advice may be biased....

There are smoke alarm alternatives available, but they're very expensive. Nest is a high priced gadget / toy that must be thrown out after 7 years. You're better off buying a First Alert or Kidde photoelectric smoke alarm with a 10 year battery. Then you'll never have to worry about it for 10 years. They'll cost around $30-$40 each.

If you want a higher-end serious fire protection system, there are alternatives if you look around. Crossfire alarms are very good (although their website may not be...) They are made in USA (even the plastic injection molding, circuit board assembly, etc).

They are wirelessly interconnected, photoelectric smoke alarms with a 10-year battery. They have a heat alarm for the garage, kitchen, furnace closet, attic, etc. These are places you can't put a smoke alarm, but also places where fires actually begin. You have smoke alarms in your bedrooms, hallway, and living room; but fires rarely start there. They also have a carbon monoxide alarm; and all the alarms set each other off. If there's a fire in your garage, the alarm in your bedroom goes off.

They're also coming out with products that will send you a text message and call the fire department, and products for the elderly, children, and hearing impaired.

These are very high-end systems, and are very expensive (even more than Nest). They're sold only by dealers who professionally evaluate the home and install them. This also adds to their cost - you're buying an alarm plus a salesman plus an installer.

Companies like ADT and AT&T offer good systems that can call the fire department and have other features. But, they're expensive, have a monthly fee, and are usually based on normal cheapo alarms.

There's a few options; but none of them are perfect...

Have you considered doing a (YC) startup...? :)

consider moving it the current rule of thumb is 19 feet to reduce/eliminate a nuisance alarm

Why spend time developing a feature when the solution (moving it to a different position in the room) is relatively simple and also safer? How many people are going to hit snooze and then be called away for something and not get alerted when their kitchen catches fire?

In the UK smoke alarms are generally permanently installed to a ceiling. Moving it isn't an option.

Really? I'm in the UK and they're just fixed up with a couple of screws.

Only has to happen once for big liability. Doesn't even need to actually happen, just "preponderance of the evidence" could have been a factor in a major fire, or enough risk of that to force a settlement.

First, I'll preface this that I'm an engineer for a smoke alarm company, so I may be a bit biased. The simple answer is that the rules forbid a smoke alarm feature like you (and many others) would like. Fortunately, if you switch to a photoelectric alarm in a proper location, you should not get false alarms from cooking.

Now here's the long answer: Essentially all smoke alarms in the United States are tested and listed to UL217, which has specific requirements and restrictions for smoke alarms. UL217, Section 7 discusses alarm silencing means. Among many other things, it says

"It is not prohibited that each single and multiple station smoke alarm be provided with an automatically resettable alarm silencing means that has a fixed or variable time setting which desensitizes the alarm for a maximum of 15 minutes. Alarm silencing shall not disable the smoke alarm. Sensitivity shall not be reduced to more than 4 percent per foot of obscuration."

So we're only allowed to slightly desensitize the alarm for up to 15 minutes; never disable it.

Fortunately, the whole problem is moot if you switch to a photoelectric alarm (around $25) instead of an ionization alarm (<$10). Ionization alarms will false alarm from cooking byproducts far too quickly; photoelectric alarm generally do not. You also should not place a smoke alarm in the kitchen (this is a perfect location for heat alarms).

While ionization alarms will go off far too quickly from cooking, they also fail to response to slow smoldering fires. There are many videos of rooms and test boxes filled with smoke that fail to trigger an ionization alarm, while the photoelectric alarm is working fine. Some states have even outlawed ionization alarms.

Buy a decent photoelectric alarm with a 10-year battery, like the ones made by First Alert or Kidde, and you probably won't have to worry about your smoke alarms for 10 years. That's even better than Nest that needs to be thrown away after 7 years.

I don't work for any of the companies mentioned above either...

Great response from Nest. But honestly - their algorithmic "trust us, the device knows what its doing" approach has always made me uncomfortable. The thermostat for instance - I don't want my thermostat to learn my habits and predict what temperature I want, etc. I just want to tell it what temperature the house should be and when, and leave it the hell alone.

Same thing with the smoke detector. False positive? Push the button. Its the rational ux given the criticality of failure modes involved.

I was under the impression that the Nest did work the way you would like— you tell it a temperature and thats it. The only "learning" involved was noticing you were gone and letting the temperature change a bit more, and when you come home it goes back to what you set. (There's also some bit about reducing the temperature a bit during peak load).

As with a smoke detector, I would never expect it to disable its smoke detecting function unless I pressed some button.

Very awesome and direct response by Nest. Kudos.

Oftentimes, recalls or other safety notices do not adequately spell out specific issues, merely allude to a vague "safety issue." Transparency is good.

  > we observed a unique combination of circumstances that caused us to
  > question whether the Nest Wave could be unintentionally activated.

You think that counts as "spelling out specific issues"? That is not very transparent to me. Compare Nest's "specifics" with cperciva's response to a security bug in tarsnap:

http://www.daemonology.net/blog/2011-01-18-tarsnap-critical-...

Fair enough, but for a consumer product, it's specific at least from the recalls and safety warnings I've been exposed to as a consumer (car recalls, child safety products, food, etc).

Excerpt from most recent smoke alarm recall from CPSC:

  > In  cooperation   with  the  U.S. Consumer  Product   Safety  Commission
  > (CPSC), Universal  Security Instruments Inc.,  of Owings Mills,  Md., is
  > voluntarily recalling  about 34,000  smoke alarms. The smoke  alarms can
  > fail to  alarm when smoke or  fire is present. A capacitor  in the alarm
  > can burn out, releasing smoke and melting the cover.

  > Universal  has  received  three  reports of  alarms  failing,  releasing
  > smoke. There has been one report of minor  smoke damage to the wall of a
  > mobile home. No injuries have been reported.

  > Consumers can identify  the recalled models by removing  the alarm cover
  > from  its base  and  looking for  the manufacture  date  code and  model
  > number. The  smoke  alarms  are  made  of white  plastic  and  have  the
  > manufacture date  code and  model number  imprinted on  the back  of the
  > alarm.

https://www.cpsc.gov/en/Recalls/2000/CPSC-Universal-Security...

In other words, a smoke-emitting smoke alarm?

Must be made in Britain. Their fire extinguishers have a tendency to ignite on use. (According to the IT Crowd tv series)

And their realistic screensavers.

What a mess. My existing CO and smoke detectors don't need a patch.

As per my comment yesterday on Nest's thermostat, this product doesn't need to exist. To have it proven as dangerous suggests that they have their priorities wrong here. I don't want poorly engineered toys keeping me safe.

As for the press statement, it's wriggling nothing more.

Their products remind me of Zorg's desk in the Fifth Element.

It's only a convenience feature anyway, it seems.

Wave seems like the type of feature that should probably have some audio/light indication that it was invoked, though.

Between this and Google, though, remind me never to name a product or feature Wave.

A convenience feature that could have killed you.

I know some engineers who design life critical safety systems for a living, and they all said something like this was going to happen when this product first came out. The example one of them gave was it was like a AED that could also let you recharge your phone, a moderately worthless feature that could only ever cause the failure of critical components.

Or perhaps a car airbag that can be disabled by spoken word

It does, the light ring changes to indicate you're seen by the Nest Protect and when you wave to silence a smoke alarm the Nest Protect says "Smoke alarm hushed."

http://support.nest.com/article/What-is-Nest-Wave-and-how-do...

So what? If you're asleep and wave at the disturbing noise in half-sleep, that voice won't wake you up. GAME OVER. NO LIVES LEFT.

I'm really wondering what unique circumstances could cause this. Flapping your arms out of fear when you see a fire?

Probably running/walking past the smoke alarm.

There's a fine line between too sensitive and not sensitive enough.

If I burn something while cooking I do NOT want the smoke detector to go off (it's loud), and the Wave feature takes 3 or 4 seconds before it registers.

Plus there are a lot of configurations to account for. Some people put the units on the walls while others on the ceiling. The Nest has to guess how far away people are when the walk by and how far people will be when they wave. Putting them on the ceiling with 10+ foot ceilings verses putting one on the wall in a small hallway will require different sensitivities for the Wave feature.

If you have multiple Nest Protects then they will all go off when one detects smoke or carbon monoxide, so the Wave bug won't matter too much.

Any by all accounts the sensors are ultrasound - cost effective, but not a famously high resolution sensing mechanism.

[1] http://www.ifixit.com/Teardown/Nest+Protect+Teardown/20057#s...

I think there are also a couple of other sensors which also help detect movement and waving. They call them "occupancy" and "activity" sensors.

http://support.nest.com/article/Learn-more-about-the-Nest-Pr...

The movement of smoke can trigger motion sensors. I believe they tried to account for this by also using a heat sensor, but if the smoke from the fire was within the threshold for "body heat" it could falsely assume a person is waving at it.

Funny, when I first read about the Nest Protect, my thought was that the Wave feature could be dangerous if it was unintentionally activated. I figured they would have really tested it to make sure that couldn't happen, but I still wasn't comfortable with the feature in general.

I'm searching for a combination of "totally normal" smoke detectors (for safety) and a possibility to trigger an additional remote alert on my mobile phone. Is there some arduino project for something like that?

Personally, I wouldn't trust an Arduino for anything safety-critical...

Do you know of any better approaches?

Kind of awesome that this will, at least temporarily, be addressed via a software update assuming that you're nest is connected to a nest account via wifi. Puts me at ease.

How do you know it will be a software fix? (Other than just disabling the feature.)

I would assume they would use a new sensor to do a more discriminating job of detecting the motion.

I strongly doubt they will replace the sensors, maybe it's influence the hardware in a 2nd version but they have plenty of sensors in the current Nest protect.

It's more of a matter of better measuring the room to allow the unit to guess when someone is walking by and when someone is standing in front of the unit and waving.

The sensors are almost working perfectly. Because this device is relied upon to save peoples lives and there are also legal repercussions to allowing a known bug, they have to make sure that the waving and walking can be distinguished 100% of the time.

Only in a very rare situation (like having the unit at shoulder level or in a hallways with a very low ceiling combined with multiple people running past it) would the unit malfunction.

Not directly related to this issue, but I'm skeptical about the Nest Protect in general. It includes a CO detector, but CO is a denser-than-air gas, so stays near the floor while the Protect is mounted on the ceiling to detect smoke. Also, most CO detectors need to be replaced every few years because it wears out. Does this mean you have to buy a new $300 Nest Protect every 5 years?

I have a Nest Thermostat that I love, but the Nest Protect doesn't seem nearly as cool or useful...

That is very true. I have one of those older CO detectors and the instructions say it should be about a half a foot from the floor. I have no idea why the Nest CO detector wants it mounted 10 feet from the floor. Seems useless.

I believe any combo smoke+CO detector gets a pass to be up high. And isn't CO roughly the same weight as air anyway?

That being said, I wouldn't trust Nest for a safety device where the goal is to have a simple reliable device, which contradicts having a wifi-connected uC in control. I don't know that they have the most engineering-minded culture either, based on the horrid input lag of their thermostat.

carbon monoxide is less dense than air - according to wikipedia, density at room temperature is 1.15 kg/m³ for CO vs 1.275 kg/m³ for dry air.

The nest protect retails for US$129.00 and has a service lifetime of 7 years.

I recently checked all the smoke detectors at my parents house and found that they were all well past their 10-year replace-by date. (I replaced them, but not with Nest smoke detectors).

Fun little tangent; I don't know how old my smoke detectors are, but the one by my front door gets a little test every now and again. One of my neighbors burns a fireplace on cold days. Invisible (but smellable) quantities of wood smoke pool under the rafters of my porch; the next morning, when I leave through the door, the smoke alarm always chirps for a moment! It amazed me how sensitive it is, when I first figured out what was going on.

> CO is a denser-than-air gas, so stays near the floor while the Protect is mounted on the ceiling to detect smoke

Research on this topic seems to indicate that CO dissipates pretty evenly:

"Contrary to a significant amount of public opinion, CO did not layer on the floor, float at the middle of the chamber, or rise to the top. In each case, the levels of CO equalized throughout the test chamber. It took longer to equalize when CO was infused at the top of the chamber than the bottom, but levels always became identical with time."

http://www.ncbi.nlm.nih.gov/pubmed/21536403

CO is about the same density as air, but it will rise with warm air. Placing CO detectors on the ceiling is allowed by the NFPA.

Nest protect costs $130 and yes you're expected to replace it after 7 years. It's actually a legal requirement to replace CO detectors after 7 years in many jurisdictions.

The primary gas in the atmosphere is nitrogen, N2 with a molecular weight of 28. The second most prevalent gas is oxygen, 02 with atomic weight 32.

Carbon monoxide, CO at 28 is the same molecular weight as Nitrogen and therefore it diffuses similarly. This is part of what makes it so serious a hazard.

The CO sensor in Nest is designed to last 7 years. The Nest has a built in timer that will start chirping after 7 years, forcing you to throw it away.

Most modern CO alarms last 10 years. Nest chose the smaller sensor, which dies sooner. Very poor decision on their part.

For more detail: Nest uses a CO sensor by Figaro, TGS5342. This is a smaller version of Figaro's TGS5042. The CO sensor is filled with a water reservoir, with some air inlets to let gas enter. Over several years, the water evaporates out of those air inlets and the sensor dies.

Nest's smaller sensor has as smaller water reservoir, which means it doesn't last as long as the other alarms with a larger reservoir. First Alert uses the larger TGS5042 sensor, therefore it lasts longer. Kidde uses their own sensor, which also lasts 10 years.

Where's the consumer warning that their data is now accessible to google and the NSA?

Reply on news.ycombinator.com