That was all very nice, but then Nest blew it. They kept forcing firmware updates to the thermostat that I had no control over. Some of them bricked the device for a period of time; some merely changed the UI so that options on the thermostat moved around or literally disappeared. I finally got fed up and reset the thing and refused to give it my wifi password. Then Google bought Nest which made me doubly sure I'll never connect the thing again.

Now it just works as an ordinary dumb thermostat. A rather expensive dumb thermostat, but it does the job. The extra features that came from connectivity are not worth the company's disrespect for my right to control the devices in my home.

No thanks. Been there, done that. With a simple device like a thermostat, I don't want the thing changing all the time. Security updates are fine but every update needs to be under my control and the device should be designed with a mechanical switch that restores the previous version of the firmware.

And really even security updates should be unnecessary if the code is written in a type-safe language with good security practices.

I got Sensis too. Best compromise I could find, avoided connecting to the internet until I couldn’t find another way to program them. Suppose I could disconnect them now.

The Nest thermostat allows me to control the heat remotely.

These two things are great for rental properties. Guests often forget to lock the door, I have records of when guests or maintenance people are coming or going, I can adjust the heat to 55 when nobody is there which saves a lot of money and more importantly I can ensure the heat is not turned off (which could cause frozen pipes) which some guests do for some reason.

Outside of a rental, these things aren't as useful. But I'd still get a programmable keypad (I've been running the Yale touchscreen ones in the mountains for years). It's very handy for friends or maintenance and with a Smartthings hub you can unlock the door remotely and give no code. Or give a code and change it later. I got the one with a key backup just in case but I haven't had the touchscreen fail yet.

I do run the Nest cameras outside and they've been nice to check snow levels, bust the spa maintenance guy for not even showing up, expose a neighbor for stealing firewood and rocks, expose guests who brought an actual bus full of people, expose another guest for filming a movie, and plenty of other things that have been helpful documentation.

It's kind of crazy how far the Nest cams pick up audio and how clear it is. So be cautious of these things because they are constantly recording and it makes it easy to scrub for footage (detects and quick links to voices, dogs barking, people, etc). I only listen to the audio when it's applicable (guests are doing something they shouldn't) but I'd imagine that some homeowners are watching and listening for fun.

Of course, the same privacy rights also likely apply to short-term guests as well, but while everyone is skirting regulations, what's a bit of surveillance in addition. I mean you've basically admitted (why people can't just shut up when they break the law) that you record your guests in outside areas where they can expect privacy (sounds like the hot tub). That people make movies in places they expect privacy (shocking), and of course you don't look at any of those skinny-dipping videos wink wink. Funny how unregulated businesses attracting unregulated activities still like to have some way to regulate their guests--and do so likely illegally.

Yeah, I know, some airbnb guests can be jerks, abuse the crap out of your investment, I mean second-home you couldn't afford without hotel-income. But you don't deal with the bad apples by violating the rights of everyone else, at least unless you want to invite scrutiny and regulation into your activities.

If they're doing unpermitted guerilla filmmaking on someone else's property, they're just as guilty of violating someone's privacy and doing unregulated business activity.

FWIW, when I lived in a house where the landlord installed security cameras there wasn't much we could do about it. The law was on their side for outside cameras and cameras in common areas (like a laundry room)

There are regions where is is a regulated, perfectly-acceptable activity, and has been for longer than the founders of AirBnB have been alive.

This is not correct. I pay transient occupancy tax just like the hotels (11% at one, 7% for another) and also have local authorities come out periodically for certifications to make sure everything is up to their standards (fire alarms, extinguishers, lighting, safety rails, notices posted outside, etc).

If I was in a city that didn't have regulations for short term rentals, then it wouldn't be a regulation-skirting business anyway. It's only skirting if the area has TOT and regulations in place and you don't abide. How did you get to this conclusion based on my post?

> I mean you've basically admitted... that you record your guests in outside areas where they can expect privacy

No I didn't. And my listings clearly state that I have cameras outside.

> That people make movies in places they expect privacy (shocking)

Why is this shocking? They were making a low budget horror film. You need a license to film a movie on someone else's property and the city has regulations and procedures for filming. Not to mention they started a large fire in the desert and I could see the fire department come out and have a conversation with them. The guest also lied to me saying they were having a little couples getaway and we even gave them a discount because we were told it would just be 2 of them.

> and of course you don't look at any of those skinny-dipping videos wink wink. Funny how unregulated businesses attracting unregulated activities still like to have some way to regulate their guests--and do so likely illegally.

I have a pool at one property but there are no video cameras overlooking the pool. Or even the back yard for that matter.

You've made a lot of assumptions.

If you don't like short term rentals because some people don't comply then that's fine but hotels have bad actors too. Plenty of hotel owners not paying their share of TOT taxes, peep holes in bathrooms, stealing property when cleaning. Ever see those service spaces between 2 rooms? Bad actors would use those for peep holes and watch guests in person.

> I mean second-home you couldn't afford without hotel-income

Actually I can afford both without hotel-income. Even if I couldn't - the cities approve of it and get income and jobs.

> But you don't deal with the bad apples by violating the rights of everyone else, at least unless you want to invite scrutiny and regulation into your activities.

I comply with regulations and also create a lot of income for people in cleaning, maintenance, upgrades. We pay a living wage to all service providers.

Nor do I violate the rights of anyone.

None of what I wrote was minor either. If a vendor is charging me and not showing up, that's fraud. Bus full of people - over my approved capacity and I can get fined. Filming - I can get fined. Fire in the desert - big liability. Stealing firewood - theft and now my guests won't have any.

Not sure why you are fabricating a story that isn't there.

I would never be okay with my landlord controlling or monitoring my thermostat.

On a flat you're renting out? That's insanely unethical.

https://news.ycombinator.com/newsguidelines.html

I'm an adult, I know I forget to turn things off sometimes and it is really nice to be able to remotely check on things like if a light is on or off or something similarly as trivial.

I bought my Nest before Google bought the company, and I will switch brands instead of buying one of the Google built devices, again over privacy considerations. Hilariously enough, we entered the name of the thermostat as HAL.

You could put a local AI model into the Nest, and it shouldn't need to talk to the internet at all. It can keep the data locally, and be programmed to draw inferences without help from the mothership.

For some reason we've decided it's ok to gratuitously connect everything to the internet "just cause". Sort of like how data maximization has become a thing because the costs are low, so why not?

I did A/C repair in college, fixing it early more than paid for the maybe 50 extra bucks wifi cost. The thermostat itself is on an isolated wifi network,

Frankly, the way these IoT devices are designed to be just dumb terminals to some cloud on the other side of the world, it's just bad engineering, in service of anticonsumer business.

Imaging buying "smart light" thats actually smart enough to do what you want it to do.

I have a somewhat irregular work schedule, and on hot days was lifechanging to be able to turn the A/C on before I got home. Similarly, if you're on vacation and go "oh shit I forgot to turn the A/C off" it's nice to be able to do that remotely. The Nest has a motion detector but it's not 100% accurate. Maybe some non internet connected thermostats also detect presence, but it's nice to be able to confirm the state of the A/C remotely. I also like being able to graph the temperature/humidity of my place over time.

That said, my next thermostat will probably be an Ecobee because I'm pissed off about a number of decisions Nest has made lately (e.g. removing access to their API)

Why do you even need AI for this? This kind of task is solvable pretty easily and is really only an ML task in the most broad of senses.

I like that I can check on the house from my phone if one travels for multiple weeks at a time. I also like warming up the house before I get home late in the evening. Granted, I've now got a VPN / firewall setup that would do this without an internet connection, but I'd still need some form of wireless connectivity to execute the changes from my mobile device.

Not much, because you're probably doing regression and I doubt this tasks needs neural nets. Well, it is definitely solvable without NNs at the very least.

I feel the same way about voice assistants. Recognition can be done locally— playing your playlists or turning the lights on and off should require no internet connection. Updating a todo list or calendar, maybe, but then it should be an API call to that service provider, indistinguishable from an action originating in any other third-party app.

Is that a problem most people have? I bought my house a few years ago and haven't touched my thermostat since I originally set it up years ago

Also, it knows when you're not home, so it can then just shut down everything until temperatures would cause other problems such as frozen pipes.

Additionally, you just dial it to where you're comfortable and if you get too hot or cold, adjust it again. It learns what temperature you like it at at certain times of day, it takes a week or two to really get to know you, but once it does, it just keeps the temperature comfortable and you rarely have to touch it or even look at it after that without you needing to go through a complex programming process like you do with most other thermostats.

That's pretty much what a smart thermostat gets you.

Whether or not you consider that useful or a good use case for your needs is really down to your own evaluation.

I like that my thermostat is smart enough to be able to regulate the temperature in my house without needing to kick on the AC or heating, which reduces my power consumption somewhat and is somewhat better for the environment, but is also better for my wallet.

Knowing when you're home can also be done quite well without any Internet access - just see if your mobile device recently associated to the wifi.

These things can trivially be built without the privacy concerns. The business model might be a bit harder if you can't collect the data, though...

The thing I really liked about the Nest is that I just turned the dial to where I felt comfortable and it just took care of everything else for me without me needing to do anything. As long as you're brave enough to put screws in a wall and hook up a handful of wires to connectors, it took about 15 minutes of me getting it out of the box to it being installed and configured. I can't say the same about any other programmable thermostat I ever had. Seriously it wasn't an awful lot more complicated than the mercury switch thermostats I've installed.

The handful of Honeywell programmable thermostats I've installed had a disastrously bad user experience. In fact, I installed the Ecobee in my current house because I got so pissed off going through the programming cycle of the Honeywell I gave up before I got to Wednesday :'D

My SO's dad is a gadgets guy, and he travels constantly for work. He is also incredibly vigilant about bills. They're the kind of family that can't touch the thermostat until a certain time of year, weighs insurance costs into every decision they make, etc.

Anyways, when he was traveling—even in cold months—he would see the thermostat going up on his phone and panic, turning it down. His kids, who were freezing in the house, would turn it up again because winter, and several hours later the cycle would repeat. They were basically locked in a constant battle to warm their house against their miserly dad and the cloud.

(He’s the “PC guy” from the Apple commercials, among many other performances, and he has a podcast where he acts as a more cerebral and empathetic Judge Wapner, hearing “cases” and making rulings. The show has quite the internal lore, complete with local court rules, standing, and binding precedent.)

The only thing my friends with IoT devices do are party tricks on voice command with their smart lights. Other than that, I find the so called 'problems' these devices solving are extraordinarily trivial.

Source: https://support.google.com/googlenest/answer/9330256?hl=en

Thanks again,

Anyway, I do really want a smart home lock. Just changing the shape of the keys so that they would stop puncturing my pockets when I walk would be enough of a gain. I also want a programmable bedroom light that can slowly dim on at morning.

I don't have those because the costs are much higher than the benefits. I'm sure a smart lock will lock me out of home at some point, with no way in. A smart light will also fail at the worst possible way, and I don't want to deal with it.

Oh you absolutely do, this has been by far the best (arguably only) benefit I've seen from home automation. Sunrise is now whenever is convenient. The lights slowly come on over 30 minutes and it feels like I'll wake up at some convenient time for my body during that period, rather than having my sleep interrupted by an infuriating alarm.

> A smart light will also fail at the worst possible way, and I don't want to deal with it.

I haven't had any problems with mine (v1 hues), they default to working like normal bulbs so the worst case scenario is you toggle the switch off and on and you get an illuminated lightbulb every time. Also (true of any LED light) they will continue to work even after you've shattered the glass by accident.

> changing the shape of the keys so that they would stop puncturing my pockets when I walk

Have you considered alternate solutions such as less-tight pants? ;P

FWIW, a lot of cheap CFL bulbs do this on their own. It's one of the reasons I hate CFLs for most uses, but I've come to appreciate that the light in my kids' room starts at about half (perceived) brightness and takes a minute or two to warm up.

I don't really care about brightness ramping, I just have my bedroom Hue bulbs come on at max brightness at 6500K at 5:00 every morning, switch to 2700K at noon, and 2000K (and dimmer) at 20:00.

It was only hot enough for this to happen at the peak of the heat when I was at work. I was able to catch the problem early and fix it cheaply all for the cost of a thermostat (Emerson Sensi) about $40 more than a non wifi model.

Of course, the next question is always "How do I confirm this?" but if it means I can catch those problems sooner, regardless of whether I'm home, I'd rather do that.

And yet, this article has me re-thinking it. Maybe removing the Nest alarm and getting Simplisafe's detector is the better move (but I need to look into theirs first).

Nothing that a programmable thermostat plus a Home Assistant instance won't get you.

Meanwhile, you just swap your old thermostat with the Nest, tell it how to get to your wifi, install an app on your phone, and you're done.

I was answering a question posted on HN, where the assumption that the average HN user is capable of fiddling with hardware and software is usually valid.

I do have some WeMo switches on some lamps in our living room. I got them instead of fiddly-dip-switch timers before a vacation several years ago, and it turns out I like having them come on an X time automatically, and turn off at 10, when we typically go to bed. We never interact with them; it's great.

But the thermostat is dumb (but programmable). The locks are all regular locks. We have no smart speaker (dear god WHY?). The TV is dumb.

Why? Because I've spent 30 years in software, and my watchword is "don't put firmware in shit that doesn't fucking need firmware."

I'd say that's a question that can be asked of anything that's Internet-connected. Phones, watches, TVs, cars, etc. They probably all create a level of convenience, typically at some cost Like all costs it's up to each person to make a decision whether that cost is too great - usually these questions are really veiled value judgments on others' cost/benefit analyses.

However, it is not a Nest, and only I can control it. It's not connected to any account or cloud service. It doesn't have direct IP connectivity at all, so it can't be turned into a part of a botnet. And it works just fine without the Internet.

to me, that is the killer reason to have it online.

https://support.google.com/googlenest/answer/9330256?hl=en

I don't get this position: my (Android) phone can already record my voice, so how does having a Google Home materially increase the privacy risk?

The GH does have access to mains power and a wifi connection, so that makes it easier to record all the time and upload a lot of data vs. a phone, but it seems like a bit of a stretch to me to think that Google would be doing something nefarious with GH and then absolutely nothing on Android. I've also noticed that when I call out to my GH, my phone will usually respond too, so it's not like my phone can't understand me, even when it's in my pocket, so the capability is definitely there.

So if you do trust Google enough to have an Android phone, then you should probably trust them just as much to have a GH. Beyond that, then it just comes down to bugs: what if GH is too sensitive and thinks you've uttered the wake word often when you haven't? Ok, then some conversations may get recorded that you didn't intend. That's certainly a concern, but "I just can't comprehend it" would seem a bit hyperbolic in that case.

> ... includes transcripts or even audio of all your household conversations/activities over the past three years

This seems to indicate to me that you believe that GH etc. are always recording and uploading every sound it hears, all day long, without regard for the wake word. If you believe this, I guess that is your prerogative, but I guess we're probably not going to have a productive discussion here.

Indeed. Maybe people don't realize that a smartphone is a much bigger privacy risk. A phone can literally record audio and video, track your position, monitor all your online activity, and upload all that without you even noticing it, even when the screen is off and locked. (This can happen when the phone is hacked or if the OS or device vendors are evil.)

I think people perceive devices like the Echo Dot or Nest Mini as more dangerouse simply because they aren't used to them. But smartphones have existed for more than a decade, so people have long accepted these.

I can leave my phone at home when I go out. In another room. Shut it off. I do all of these things quite regularly.

Do I go around and disable numerous smarthome gadgets for privacy or stick with my oh so terrible life without them? Bonus! I can avoid awkward conversations with my guests.

People buy products because we feel they help us be better. There’s little real utility towards bettering myself with a colorful, and extra power sucking, thermostat at home all the time.

I don't understand how the risk is anything but greater with a phone.

I would love to see a parody of Star Trek where all the tech is run by an advertising company exploiting all the crew members individually by capturing all their private data around the clock.

I mean, if you don't trust the computer to look after your interests with regards to your voice interactions, why are you trusting it when it also controls life support and airlocks? ;)

Which is appropriate. There's nothing wrong about locating a crew member on a _military vessel_.

Like come on guys, there's more than personal logs. There should be more barriers before you go there. But to also be fair, it is stuff they know is going to be used in that way and it isn't just a log of everything they've said in their room. So there is a little more consent involved. I don't think you can really give consent to always listening devices.

This is the highest probability of the realistic outcome of AGI as it will advance in capabilities so much and so rapidly that we won't be able to keep up. I personally think it will be benevolent and care for us like we do for our pets which won't necessarily be a bad thing.

We are still very much savages. Clinical psychopaths and delusional narcissists run our institutions and major decisions are made on the basis of fanatical superstitious ideologies and mindless herd behavior.

In short I won't put an Internet connected microphone in my house and I'm very wary of privacy leakage in general because I don't trust the people who run our world (including both governments and corporations). I don't think this is irrational. Glancing around it seems like a decent number of world government and corporate leaders belong in prisons and mental institutions, not powerful offices.

But also portrays societies run by and indeed composed entirely of mobsters (see "A Piece of the Action" TOS S2E17).

Interestingly, while this is certainly the background assumption of Star Trek, it is much less valid in episodes where the crew must interact with their own government.

I suspect the answer is a mix of "we want to invade privacy" and "it makes the devices cheaper to build since you can put really wimpy chips in them."

1) Context is nearly invisible, and is constructed from dozens to hundreds of data points. Holding state on the user (not just from direct interaction with the agent, but from other services the agent can be aware of) makes a lot more context possible. My assistant knows about my calendar, where I've been recently, and my recent emails; it can make much better educated guesses about what I mean if I say "What's travel time to Oakland?"

2) Getting the voice recognition tech itself from 90% accurate to 99.9% accurate requires awareness of the tens of thousands of real-world complicators that never show up in laboratory testing. As Google showed off by bootstrapping their voice recognition tech via Goog-411, they know how to take megabytes / gigabytes of "dirty signal" and use it to refine their algorithms against unforeseen corner cases.

3) Releasing new assistant features can be done atop existing collected data, without requiring months of lag for enough state to build up for the feature to be useful.

(1) is not that hard, just subscribe to feeds or query cloud services from the device. (2) is harder but could be done by treating the devices as a cluster or sending data up only when it can't be easily recognized and storing it only long enough to be useful in training (and storing it anonymously), and as for (3) the device could store data encrypted in the cloud and that would be fine. A new or updated device just downloads and decrypts the index to its data. There are other workable approaches.

I think the best way to change that is legislation. We need HIPAA-type liability laws applying to all privacy-sensitive data: recorded audio, recorded video, location, etc. Leak someone's location or a recording of audio from their house? That'll be $10,000 per person per incident.

That would instantly transform privacy invasion from an asset to a liability and incentivize companies to minimize data collected and when it must be collected to provide the service minimize the length of the time it is stored.

As it stands literally all economic incentives jump up and down and scream "invade the user's privacy s much as possible!" These are:

(1) Data is valuable and advertisers will pay for it or pay for data-driven ad services.

(2) Engineering privacy-invasive systems is easier than engineering privacy-respecting systems. This is a subset of "engineering insecure systems is easier than engineering secure systems" as privacy is a dimension of security.

(3) Privacy sensitive approaches usually require more compute and storage at the edge, increasing device build cost.

I'd like to see regulations addressing this issue directly, in addition to more strict pro-privacy regulation. For instance, I'd love to see it so that data, software and compute become independent - instead of current vertical integration. That is, I want to a) own my data, b) choose where it'll be processed, and c) have the software processing it come to the data - this way the vendor supplying software for c) won't be able to take the data and sell it or hold it for ransom.

It's also a lot easier to debug and improve software running on machines in a cloud one owns than on local heterogeneous architectures out "in the wild." Ask anyone who's done backend web development vs. frontend how fun it is to get "It don't work" bug reports from end-users with no logs and only spotty event metrics to guess what failure mode they got into and how they could get out of it (including failure modes such as "their ad blocker decided to kill half your functionality because your JavaScript's name had 'ad' in the URL").

There's a lot of incentives aligned for thin clients and fat servers.

I agree that if you want to change the model, law is the way to go. It'll hold back progress, but if that's the price we want to force people to pay, we could.

There are plenty of avenues to collect the data needed at that scale, it's just that none of them are as cheap as simply spying on users.

The fact that incentives exist to violate user privacy is not an excuse to violate user privacy.

The second most correct answer is to just opt-out if you feel spied on and get to feel very smug watching people who didn't opt-out get burned by their choices. Honestly, I recommend that solution unless one encounters a situation where opt-out is impossible.

While both were government funded, the "cowboy" approach of ArpaNet and IP vs the OSI network protocol consortia are a great case study.

It's everything after the hotword, not everything.

For a proper Star Trek comparison, imagine how comfortable Federation citizens would be with using ambient computing technology created and sold by the Ferengi, who are also known to be performing surveillance for Romulans in exchange for money.

(For non-Trekkies here, Ferengi is a species of extremely greedy and exploitative capitalists; Romulans are what you'd get if you reified the Five Eyes into a species, and put them into state of cold war with the Federation.)

I actually am less worried about those (since I turn them off) than I do about Netflix/Amazon knowing every movie I watch. Amazon knowing every product I order. My Credit Card and Bank company knowing every business I transact with. My Mac and PC spying on what I do with them (i think that's more Windows than Mac). My PC/iPhone/Android/Rift knowing every app I run. My ISP knowing all the websites I visit or DNS lookups I make etc... I could go on and on but I don't see how Google sticks out here.

It's a much bigger problem.

Also, the Netflix and Amazon examples don't seem to be in the same category, to me. Every time you interact with Netflix and Amazon, you are doing so deliberately. You are choosing exactly what to give them. You can easily pick another service.

Always-on listening devices, on the other hand, are taking data that you did not choose to give them, and other people's data as well. That's the point in the headline of this article. I would never need to tell guests "If you order something from Amazon, they'll know you ordered it," but apparently I might need to tell them to watch their mouths in my house. There's a big difference.

It does bother me when politicians say GAFA and leave out Microsoft (but not Apple??)

#Netflix did play an important part in adding DRM to the Web specifications...

(I guess it's not the only slur it has to avoid with that letter combination though.)

Don't forget that Echo devices are Amazon's always-on listening devices.

If you have proof of echo devices not doing this, you need to yield a source. Wireshark logs, anything.

I suspect most of these comments are from people who don't own, haver never used, let alone sniffed traffic from those devices.

They conveniently ignore all the networks (including my own) which will throw a fit if these devices start sending unusual traffic.

I'll worry the day they embed a cellular modem.

Imagine if AI advanced to the point where a human brain could be simulated in a computer. If you collected enough data on a person, you could create a virtual clone with their exact same beliefs, opinions, intelligence, etc. Just imagine how easy it would be to exploit someone if you could simulate how they'd respond to anything with perfect accuracy. You could literally enslave people.

And while that's currently scifi, the amount of money companies like Google and Facebook are pumping into AI research means we could land somewhere pretty close to that eventually. If we're still not protecting peoples' data by then, we're all fucked.

[1]: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ...

Might wanna read that again, it’s a story of clever and spooky observation but not that clever.

I think allowing any company or government having an open mic/'smart' speaker, essentially bugging your own home/business is just a bad idea.

"I actually am less worried about those (since I turn them off)" - are you sure they are off, had to ask.

I'm not as worried about Netflix/Amazon knowing every movie I watch, product I order or am interested in.

I'm more worried about private conversations and communications being monitored as it applies to freedom of ideas, diverse political movements, freedom to vote and even freedom of religion.

Can you cite where Google says that the mute switch is a hardware kill switch for the mics? I've only read where they say that it doesn't listen but never that it physically disables the mics which means it's done in software and the device could be "muted" and still able to listen.

Unless the mute switch is a physical disconnect for the microphones then it's no different than the Alexa's mute button.

And don't forget Google's Nest Secure contains/ed undisclosed microphones and not physical mute button and Google wanted you to trust that it is/was disabled.

If I google "Google Nest Home Mute Hardware disconnect" for example, none of the results on the first page state it is a hardware disconnect.

The support.google.com links all use vague language that never explicitly says there's a hardware disconnect. Just that the mute switch causes the microphones to be disabled.

Interestingly the last result on first page is this: https://store.google.com/magazine/google_nest_privacy#ada2d2...

And it's claims are known to be false.

> Your device will only send audio to Google if we detect that you or someone in your home is interacting with your Assistant

So I'm genuinely curious where all this information about the mute switch being a physical disconnect for the mics exists.

https://news.ycombinator.com/newsguidelines.html

That was more of a joke, are you sure they are off?

I'm sure someone has or is testing that the hardware mute button works.

I'm not against Nest/Google. I was always sporting Google T-Shirts and hats back in the day, telling everyone to use it and recommending their products.

I'm just not sure on this product.

Obviously there can be defects, but Google is not getting 100% of the audio from your home.

You are using 'Google' in a vague way which may be why you are confused about what my comment meant. The Google device gets 100% of audio it hears, because it can't magically hear a 'wake word' without listening to all audio. It only sends audio to a Google server after the wake word.

(This is at least true for the iPhone, Hey Siri is recognized locally before anything is sent off to the net)

Re reading your comment, I guess you mean: the sound processing goes through the local microphone no matter what, even if the audio is only forwarded on if they lead with certain keywords, right? In other words, just like how every other voice activated system has to work. The privacy implications are that the equipment could be compromised to actually put the recording in longer term storage, like a bug or something.

https://machinelearning.apple.com/2017/10/01/hey-siri.html

I've disabled Siri and Cortana to the best of my ability (and I don't own any Amazon devices with a microphone). Of course I can't actually be sure that that means that they're really not listening, but it's absurd to imply that just because one owns a smart phone they must give up all hope of privacy.

You are trusting that your efforts to disable Siri and Cortana have actually resulted in them being disabled, and that Apple and Microsoft aren't doing nefarious things to keep them enabled anyway. If you trust that, I don't think it's a stretch to trust that Apple and MS would avoid doing nefarious things with a HomePod or [whatever MS's analogous device might be]. You seem to not want to have anything around that can record you without your express permission, so sure, I get why you don't have a HomePod.

I had an Amazon Echo for a while, but decided I didn't trust Amazon, and exchanged it for a Google Home. I have an Android phone with Google Assistant enabled, and I don't see the privacy implications as materially different between the two.

I have a Google Home in my bedroom, and it's extremely convenient for controlling my lights. I'm aware of the privacy concerns with such a device, but out of all the things in the world that worries me it's just not far up my list to disregard the convenience. I imagine a lot of consumers might feel this way. Now this would be a different story if something was forced upon me (like the government mandated some device, and I don't have an active choice in having it) but I'm making an active choice to buy this device and use it knowing that it could very well be recording the conversations, sex, whatever else goes on in my bedroom.

Every day we are making choices about what risks we're willing to accept. In the scheme of things, this is novel but minor.

Always-on listening doesn't seem like a huge deal. What matters is the policy of allowing Google people to listen to captured audio, even if it's for training purposes. This is skirting uncomfortably close to a dystopian surveillance society.

If you want to gather data from people, provide an incentive to voluntarily participate. For instance, on Google maps you can become a contributor and there are benefits and bonuses if you contribute enough. Yes, this group is self-selected so it's not as good as a random sample, but it's better than the dystopian alternative.

https://app.leg.wa.gov/rcw/default.aspx?cite=9.73.030

Secondly, even if it's not normally recorded, so long as the capability is there, you've basically bugged yourself. Now if you become a target for surveillance, all the other party has to do is arrange for (Google|Apple|whoever) to give them a real-time feed from your device and now all of your conversations are fair game.

Smart phones are bad enough, but why put more devices in your home (car, office, etc.) that make it trivial for others to monitor you?

Combine that with a general rise in authoritarianism and living outside of the US where these tech companies are based.. I can see the problems arising.

The reason is "people" lack education in technology. They can barely use a computer these days, with "using a computer" generally means being on Facebook with a browser.

There's a very significant divide in this regard that has been growing for decades. And it isn't getting any better. This is why it is nearly impossible to convince people you know to use better passwords and adopt security and privacy-conscious behaviors. I don't know if you (plural) have experienced this, I know a few people for whom sending a link via email is a challenge. They think they sent you a link and what you get is unusable. No, it isn't about age at all.

Smart phones and tablets have changed the relationship people have with computers. They make things very easy to use and require almost no technological knowledge to operate. In computers, the browser or Chromebooks do the same.

Add it all up and the idea that someone does not even remotely think that a thermostat could be recording every word being said in home doesn't even cross their minds. Going beyond that, thinking about how and where that data is stored and how it could be used it yet another level.

All this says it is the responsibility of the tech community to make sure customers understand the important aspects of the technology we provide.

Which is difficult as long as some peoples income depends on not making sure that customers understand.

As much as I'd like to pretend we are all morally and ethically acting grown ups, reality tells us we are not. Ultimately this needs to be regulated by law and continually enforced. It seems like regulation is slowly rolling in, but enforcement seems to still be far out.

This is a dumb statement. Lots of tech people use all of these devices. In fact smart tech people differentiate between machine listening and humans listening.

A Math, Tech person know that the probability of a human listening to my recording is close 0.000001%.

Humans, including yourself, make far worse decisions every day in your life that have much higher detrimental effect to your overall quality of life than worrying about stupid things

I thought it’d be fun to go back through my two years of recordings to see if it captured any accidental conversations, and it was so boring to go through. They are entirely expected questions like “what time is it” “what’s the weather today” “set a timer” “what’s the cubs score” “when do the cubs play” “how many games back are the cubs” “how many games left are in the cubs season” “Do the cubs even have a chance at this point”

I agree with you that nobody is listening and it isn't valuable data even if they were.. My biggest issue is that the devices are a black box. Nobody knows if they're hacked or compromised which is why hardware disabled buttons. More transparency is critical. I think I should be able to monitor all of my smart devices' activity with as much done on device as possible.

Apple and (now) Google have made strides with on-device ML lately - I hope things continue down this path.

> This is a dumb statement.

How about doing a bit of research before saying stuff like that?

For example:

https://lifehacker.com/this-chart-shows-how-computer-literat...

From the article:

"The results aren’t terribly surprising if you’ve ever worked a help desk. Around 5% of the U.S. population ranks at level 3, the strongest level of computer literacy. Another 26% ranks at level 2, where users can do things like “find a sustainability-related document that was sent to you by John Smith in October last year.” Those who ranked in level 2 likely aren’t developers or engineers, but they get by.

However, for the 69% of the population below level 2, complex skills are still hard to come by. 26% of people weren’t even able to use computers. The full report at the source link below shows how these numbers break down over age ranges. Regardless, the numbers give some perspective on how skilled the overall population is."

And this is about using a computer. Understanding technology in general is a very different matter. The data on this isn't difficult to discover. There have been a number of studies, for example:

https://www.pewresearch.org/fact-tank/2014/12/04/half-of-ame...

The vast majority of the population is utterly ignorant when it comes to technology. Start adding areas like machine learning and AI and the chasm is massive, deep and wide.

Using tech people to claim this problem does not exists and, therefore, calling the statement "stupid", betrays a failure to understand the topic at hand.

The same is true about medical knowledge. The vast majority of the population is utterly ignorant about medical science, pharmacology, etc. Claiming this statement is "stupid" because doctors, nurses and medical practitioners do, in fact, understand the domain, is, again, missing the point. And missing reality.

The example is useful because, unlike in the general case of technology, the medical sector has a range of safeguards, rules and laws in place in order to protect consumers. Very little, if any, of that exists in technology.

You can manufacture and sell a 3D printer that can fail and burn down someone's house. There is not FDA equivalent in order to ensure some level of safety exists as a minimum. If you think FCC, UL, CE and TUV provide those safeguards, well, you are obviously not in the hardware business. This is particularly true for the utter junk that comes from China, where these certifications are given away like candy and ar sometimes falsified.

And so, you can insert a microphone and a camera into a consumer device and there really isn't anything preventing anyone from using those sensors nefariously. This is particularly true if the device's firmware and software can be updated over the 'net. Which means you can ship a device that can test out to be completely benign. Later on the same device is updated over the net to include obtrusive capabilities that compromise privacy, and nobody is there to prevent that.

And then, of course, there's this:

https://www.youtube.com/watch?v=0vL4HLTZQ_Q

and this:

https://www.youtube.com/watch?v=t-lMIGV-dUI

...and who can forget this:

https://www.youtube.com/watch?v=aHGd6LqAVzw

Because it's a lot of fun and the personal risk is basically nonexistant. It's the oldest story in the book: different personal evaluations of risk of activity.

The Star Trek vision of voice as an interface is a good vision. The question is how the regulatory framework protects users in this world and keeps it from being the dystopian nightmare that our "for profit corporations get to do whatever they like as long as you agreed in an 'all or nothing' click through" seem to be headed for.

https://techcrunch.com/2019/03/12/googles-new-voice-recognit...

That's not really observable or enforceable by me though. A question I ask myself all the time before I say yes to some kind of "new data tracking" is: "How do I think company_x will behave in 10 years if their profitability is plummeting?" Because I think that paints the worst case for companies that so far have been good actors.

I guess the TLDR is: That's great, but I think that the solution is legal protections for users. Right now all the power is in the hands of the collecting organization.

Voice interfaces today work like keyloggers by design, and we're just trusting them not to abuse all the data they're getting. It'd be a lot better if we didn't have to.

(I'm on board with legal protections too, of course.)

We are far more likely to end up in a cyberpunk nightmare where corporations are richer than entire nations and technology is used to subjugate and control people.

I don't really understand why the device needs to store my recording and send it over the internet.

I would be willing to pay more for a device that guarantees that my recordings are deleted immediately.

I think a lot of people don't understand how much their data is worth. It is something I frequently hear, that "my data is useless, so why hide it?" or "what are they going to do, see that I like cat photos? HA!". I think these kinds of comments are slowly diminishing as the average person is seeing 1) these tech companies are worth $1+ trillion 2) people get scared and think Facebook is always listening because they every so often hit the mark too well. But I don't think most even know the extend of the data that these companies have (e.g. they think that Google only has their email data and knows nothing about financial), let alone how that data can be used.

Personally, I thought the bigger news (insult) was when he said, "Gosh, I haven't thought about this before in quite this way."

Give me a break. You knew it. I knew it. We all knew it. Maybe the right thing to do is have the device recognize there's an unknown voice in the room and announce that it's listening. That hasn't happened for some reason, but I highly doubt the reason is that no one ever thought of it.

Imagine if I made a self-driving car. Imagine if, in the rain, it simply slowed down and stopped right where it was, even in the middle of the freeway.

Now imagine that when asked about whether this was dangerous and ought to be disclosed to customers, the VP of Engineering said, "Gosh, I haven't thought about this before in quite this way."

We'd all say the VP was either lying, or was dangerously incompetent and was running a dysfunctional engineering group.

I don't need to get into a long conversation about whether this is malice or incompetence: It's unacceptable either way with self-driving cars to claim to have not thought about safety, and in the case of privacy and disclosure of recordings in my home, it's unacceptable either way to claim to have not thought about it.

This company is rotten from head to toe, and so is the system of incentives that lead companies like Google and Facebook to flout all standards of responsibility.

How can someone say something like this and still have their job at 5pm PST? How can Zuck say time and time again, "Sorry, we'll do better next time, by the way, political ads can lie," and there are no serious consequences?

I see parallels to the genetic testing kits: it's all fun and games until police use it to find people who committed crimes. Just wait until investigative units realize FAANGs have recordings from inside houses that they can supeona.

Because big tech/advertising companies don't sell your individual data. They sell to other entities the ability to show people like you specific content. I wouldn't use products monetized in the first way, but I'm happy with the latter approach.

Humans are a lazy bunch and this feeds into these genetically coded tendencies in the worst way.

I feel like there is a growing backlash against these technologies that are getting way too invasive. Myself and a few of my friends have gone back to being as "analog" as possible. Dumb thermostats, dumb tv's, regular old key locks (you can still get some pretty secure, non-electronic locks), dumb fridges, etc, etc. The point is to disconnect as much as possible from the internet, Google, FB and Amazon. Three of my best friends now just carry Windows phones. Internet connected, very few apps, and phone service.

We are of course, a very small minority whose seen the dark side of these technologies. I don't think enough people know or understand the consequences of having these devices in their homes.

What do you mean by this? What exactly is it that you believe smart speakers capable of that smartphones are not? (Modern smartphones are just as capable of listening for and responding to hotwords as smart speakers are, fyi.)

I think the same thinking allows people to overlook their phones while focusing on their smart speakers.

I imagine that if governments start to use Alexa, et. al. to crack down on dissenters, those whole business lines disappear as people dump them all in the trash. I think we have to depend on capitalism and whistleblowers to shout if things change. If someone wants to listen in on the sounds of my bathroom, help yourself. I personally draw the line at cameras. No camera, no Ring doorbell and my phone is usually in my pocket or in a shelf with no visual access.

The other thing that I'd point out is that a lot of people publicly shout their political and other positions voluntarily. This is great for free speech but is an even easier and richer target for government crackdowns.

There is just a disconnect in understanding. With huge efforts they will create archives of 5% special price coupons, but how tech companies extract free money from them is beyond their comprehension.

The implications of having every deed and word quantified is perhaps too abstract a danger to understand.

Jaron Lanier has an interesting plan for renegotiating the "paying for services with data" relationship we now have with tech companies: https://www.nytimes.com/interactive/2019/09/23/opinion/data-...

Because flying cars. The home assistant is the glimpse of future people looked forward to since first Scifi movies. Its convenient and out of way. And people don't understand nor care about privacy.

That said I am still ranting to my brother that he is using Alexa whenever I visit his home.

It may be naive, but I think the device vendors have way more to lose if it turns out they are actually recording all the time, not just listening for the wake word.

It will be the wrong road if we end up at 1984, but I’m hoping for the Culture, or at least the Jetsons...

I'm personally not a very private person and I'm not worried about giving up some privacy for convenience. There are a few bad things I could think of like the government using out of context recordings to frame you or something like that, but the odds of that happening seem so vanishingly small, I can't worry about it.

I find their news product a useful barometer of the extent to which they've abandoned that. Last Monday I sat down at my workstation and the top item - amid war, political instability, and everything else going on in the world - were articles about the Google Pixel 4. I know this wasn't a reflection of my browsing habits - I'm just not interested in phone technology at the product level, and and any online shopping/product research that I do these days happens on a separate account on a separate device, because I don't want to mix up my consumer preferences with my intellectual interests.

You would think that with all Google's brainpower the presentation and distribution of news would by now have seen massive advances, but instead it has simply magnified the worst aspect of print and Tv news. I'm beset with sports, entertainment, 'health' (mostly quackery), and 'technology (mostly consumer products) news every day, even though I used to be able to opt out of those things. If I make the mistake of looking at a sports article on my workstation or phone, I am swamped with sports articles for the next day or two, presumably because the sports market is so lucrative and Google's algorithms get tremendously excited at the possibility that I am adopting the consumer the consumer profile of a sports fan.

Likewise I am constantly being advanced second- and third-hand sources of news based on SEO and existing reach; every other story I read is a half-assed rewrite of original research done by a news outlet with a smaller brand footprint. If I'm lucky I can see that in the first paragraph or two and can click through to the original, if not it's because a big news outlet is just stealing or redoing stories without credit or attribution and I may find out about it later when I read Twitter.

I could rant on and on and on about Google's news product and about their other products and practices, but all my critiques of the firm boil down to this: google has become a company that takes bids (including from itself) on the order in which it provides leaves to consumers, while denying them sight of the trees they are plucked from.

The thing is everyone has a cellphone, even many kids. Invite half a dozen people over and you’ll get half a dozen cellphones.

I have a couple of Amazon Echos, plus iOS devices that occasionally activate Siri for no reason.

I love them. My feeling is that I’m willing to live with it for now. Hopefully, a decade from now, voice technology is perfected and more is done on device.

Once it becomes a solved problem, there will be better open source devices and more competition. On device will be a checkbox feature.

Not every company makes you the product.

yeah, but since that has been a workable angle we'd had an explosion of companies that do make you the product.

I wouldn't assume the inverse over time related to a reduction in effort from better devices and competition; i'd assume further growth.

I agree, smart phones provide the same surveillance capabilities of 'smart' speakers and more.

But supposedly phones aren't always on and listening to all your audio like the 'smart' speakers.

This might just be a divide in society, some people accept surveillance as not an issue and are fine with it.

Would you place a similar 'smart' speaker device in your home if the government suddenly required it?

I just feel a surveillance state is detrimental to freedom and democracy, the freedom we have in the states is not common.

If you look back in history and even around the world today it's pretty rare. I just want to make sure we aren't setting ourselves up to weaken it or even lose it.

New phones are.

Also, AFAIK, Alexa et al. are only sending traffic home after the wake word is heard. I believe this has been confirmed via wireshark.

But it's a setting. I own an iPhone have turned off "Hey Siri". If it would still listen all the time that would be a security issue and breach of trust.

It seems 'smart' speakers wake more easily than Siri.

We hadn't been hearing news stories about FANG employees listening to/reviewing conversations intimate moments until the 'smart' speakers were deployed.

I expect this could have been going on with Siri as well though.

I'm open to using a 'smart' speaker in the future, there might not be any reason for concern.

Currently it just seems like a bad idea to add recording devices you don't control around your home.

That just shows that the media didn't think those stories would sell.

Not new Android phones. Since Android 8, google no longer listens while the phone is locked. If you want to use the google assistant, you have to give your fingerprint or password first.

It's more accurate to say "after the wake word is heard or after a false positive for the wake word is heard".

Even in the presence of hardware interlocks (active when the wake word has not been detected), the accuracy of wake word detection is a really important question, as is the question of whether an adversary can modify the wake word detection code remotely.

For example, just this year my always-on, always-near-me iPhone was subject to an attack where anyone could call me on FaceTime and gain access to my microphone. This attack vector does not exist on my Google Home.

So, why are smart speakers much better at invading privacy?

Streaming audio is not seen as suspicious.

Also they aren not phones and not using wifi, not LTE/4G uplinks, making it more easy to stream audio without hitting bandwidth caps or using too much battery.

It's way much easier to notice if a phone streams audio often and unnecessarily.

For a phone the assistant is an optional side-feature and disabling it takes almost nothing away from the utility of the smartphone. You're still left with a smartphone.

I'm willing to make some sacrifice so that I can get GPS/maps anywhere I go, ability to make emergency phone calls, etc.

Smart speakers really don't have the same value add.