I assume storage of this data for them is nearly 0 cost, reputation is not an issue as most >99% of people seem to simply don't care about privacy and there is probably no legal limit.

There is no reason for them to do it. In fact if they did, it would be considered a loss of value. Its valuable dataset.

Even deleting stuff in 'big data environment' is generally considered too expensive rather than just un-indexing data points that are not relevant anymore.

There is in the EU. Keeping data indefinitely is illegal. Expert opinion varies on how long you can keep it, many think you're save with 12 months, most think 36 months is too long.

So "Alexa, order more beer" in itself can be kept for as long as you want. The audio recording may be a different matter because of voice recognition.

As to how long personal data may be kept, well in the UK one can sue on a civil matter for up to 6 years so there is a good argument that anything related to the service you provide and in relation to which you might be sued should be kept for 6 years. I expect it's similar in other countries.

One day it will all be compromised or misused, because that's what history teaches us.

They just need to destroy that link and can then keep the transcripts for as long as they want.

Is that unreasonable?

I mean, I cannot possibly expect myself to go through my backups and delete a folder from everything, down to the multi-year incremental.

Once you start tampering with a backup, it stops being a backup.

All of that data is wiped when you initiate a delete request under GDPR.

I don't believe that, but it is shocking to me that anyone would want one of these things.

It’s not shocking at all. I find my Echo to be incredibly useful and I’ve decided that it’s worth the trade-off. Just like I did with the microphone I have on my wrist and the one in my pocket.

The "smart speaker", on the other hand, is 100% useless if the microphone is disabled. It also has microphones designed to hear as far away as possible given its design.

Finally, the phone in your pocket and watch on your wrist likely handle the audio data very different than that speaker manufacturer. You might find it useful now, but don't discount now you'll ever find it intrusive.

>My expectation

This is the key phrase here. Just as you have your expectations, Alexa/Home owners have their expectations. And if Alexa owners find out that Amazon/Google is using their inputs in ways they don't want it to, they'll drop it or find an alternative.

Different people have different expectations. I was late into the whole smartphone thing, and refused to buy one for years because of the privacy leak it represents (heck, I refused to buy a dumb phone for years for the same reason). Back in those days, I would have been as unsympathetic for your rationale of owning a smartphone as you are of those owning smart speakers.

Do you not see that there are different levels of comfort, and not everyone has the same level as you do? Furthermore, do you not see that on that continuum, even you have made compromises?

Without being aware of what is going on, people can’t make these choices for themselves.

The Snowden leaks revealed that this had been an attack vector available and used for years.

The method they were using has since been patched, but one would be foolish to think that it has been impossible for interested parties to have found another way to do the same.

I'm also not disputing there are baseband controls a nefarious entity could abuse but I'm saying that people who choose to trust Amazon with always listening devices vs a smart phone are not comparing apples to apples.

Both do local processing for the keywords until it hears one then sends the following data (and I believe a few seconds prior) to the servers for processing.

My phone works the exact same way with the Google assistant. I don't know if other "assistants" (like Bixby) work the same way.

I enjoyed the conversation with Alex Stamos (former CSO at Yahoo/FB), where pointed out that had George Orwell been told that people would willingly pay for a tracking device and keep it on them at all times and powered on at all times, he would not have believed it.

I should add, any time I say anything even remotely negative about cell phones, I get beat up here at HN pretty good. This is a very good psychological indication that people have cognitive dissonance and denial around this topic.

There are of course more subtle ways to do this, but not for real time monitoring. You can transcode the audio with an encoder that does advanced mathematical quantization and batch upload it. If all you care about is human voice, then you could upload an hours worth of audio in a couple hundred kbytes, if even that much.

Days worth of text could be compressed and sent off in literally dozens of kilobytes. And that's not even accounting for side-channels (a hypothetical targeted attack could conceivably look for "target phrases" and send a single bit of information using a side channel to some server to alert that the phrase has been said).

Back when I used to help the FBI monitor cell phones, they were very slow processors and the user could tell when it was sending anything due to the heat. Best we could do was play with MWI on/off to get cell locations and use the mainframe (cell switch) to do audio monitoring, but that was very limited at the time.

The coprocessors are really just ASICs designed to run neural networks efficiently. The real cool shit is in the compression and simplification of the networks that allow constant recognition from a small power-sipping chip.

They're also doing a bunch of work with federated machine learning processes. So all of the training data can be kept on the device itself, but they can (i'm grossly oversimplifying this) basically "merge" a full neural net from somewhere else with data learned on-device.

It's that last part that I'm extremely interested in, as it means we can have all the benefits of machine learning systems, but also keep all of our privacy in that all the processing happens locally, our training data doesn't need to leave the device at all, and the algorithms can still adapt and improve for each person individually. It apparently also has side effects that the neural network on your device can adapt to your usage individually much faster and more accurately than a "master" neural network ever could.

Are there now co-processors running all the time that could do full speech recognition (not just keyword recognition, which is way easier, but full speech recognition) all the time without such significant battery drain that it would be noticed? I ask not because I consider this some sort of technical impossibility, but because if that is the case, I'd like to know so I can update my understanding of the situation.

You are correct that the actual processors that are "always" listening are fairly limited (They describe it as a DSP, so that should give you an idea of its capabilities), but when layered the way they are in Pixel phones, it makes it possible for the DSP to detect "something" interesting is happening, then wake up and pass the information to an "AI processor" (they call it the "Pixel visual core" but i find that a misnomer because it's a pretty capable TPU on it's own) which would then do the transcribing, and go back to sleep.

I'm making some assumptions that I haven't researched myself, but I'm assuming there are patterns, frequencies, and other simply detected "signals" present in most speech that make it easy to detect "someone is speaking", and from there can wake up a more powerful (but still surprisingly efficient) processor to do more work.

And Google also has some simplified and compressed neural networks dedicated to transcribing speech which are designed to run on lower powered devices[2]. I don't know if they are actually using the PVC in the Pixel phones to do audio transcription or if they're just using the general CPU for that, but I have a feeling that a TPU would be able to scream through buffered audio very quickly and would go back to sleep very quickly leading to minimal power draw.

In fact I'd be willing to bet good money that an "always transcribing" system could be done today with current generation phones in a way that wouldn't impact battery life enough to be obviously noticeable or cause the user to start looking for what was draining it in most cases. And considering how often I'm personally actively speaking in a day, I bet it would have virtually no impact on my personal device if implemented this way.

(and man did it feel wrong to type that last paragraph out!)

[1] https://ai.google/research/pubs/pub46522

[2] https://ai.googleblog.com/2019/03/an-all-neural-on-device-sp...

And of course anyone within earshot of any of those devices also deserves to have their privacy obliterated, even if they themselves don't own one?

I really don't understand why people don't get the utility in these things. They're useful. Being able to control lights in the house, control the TV, set timers and reminders, listen to music and the news, manage grocery lists, and more are all super useful.

I have Google homes, and I (and my family) use them on a daily basis, they're useful. That doesn't mean we forgo all privacy, that doesn't mean we deserve any violations, that doesn't mean we are okay with information being collected against our will.

My family and I are comfortable with the tradeoffs that the device makes, we trust that the company won't break a ton of laws and won't be "listening" when they explicitly say it's not. We are able to delete information we don't want Google to keep, and we are even able to set the account to automatically delete all recordings after a specific period of time if we want. And we trust that the company isn't just lying when they very explicitly say that when you delete something it is actually deleted.

I also support more laws and regulations around this, what they can store, how long they can store it, what they can do with it, and mandated ways of controlling your information.

You may have different priorities, and that's okay, but I absolutely hate this idea that people who want the benefits of these devices deserve anything, and the implication that only those who don't care about privacy have them. And a token "I don't believe that" doesn't negate the sentiment you wrote right above it. I'm an adult, I'm able to make a conscientious decision to trade some of my privacy in a controlled way to get benefits from it, and that is no more wrong than you wanting to protect your own privacy.

Besides gathering a huge "spoken language" data-set for machine learning.

You did forego all privacy and clearly you are okay with information being collected, otherwise you wouldn't have those devices.

That being said, you do not "deserve any violations".

Absolutely not! I would not be okay with people being able to listen to the audio they record without my consent, I would not be okay with a transcript being publicly available, I would not be okay with the history of the commands I say to it being publicly available. I'm also not okay with audio being recorded without me triggering it to record.

Making limited tradeoffs with my privacy in specific situations is very different than not caring about privacy in general. And "privacy" isn't just one thing, there are tons of levels in many different areas.

For example, I am not okay with cameras that upload anywhere off site IN my house. That's a line that I'm uncomfortable crossing. Not because I think that there is someone viewing every second of uploaded footage, but because the risk of a security breach is high enough and the information that camera would collect "private" enough that I don't think the tradeoff is worth it. But cameras that record the OUTSIDE of my house and upload it somewhere? I'm fine with that. The expectation of privacy is different on the other side of that wall for me, and i'm more than comfortable with the tradeoffs and risks of that going public.

For you those lines might be different, but neither of us forego all privacy because of those decisions.

Something needn't be public to violate your privacy. Clearly you agreed to google listening to and recording to all sound being made in your house, leading to privacy violations. Also, the risk of security breach pretty much isn't if it happens, but when it happens.

>Making limited tradeoffs with my privacy in specific situations is very different than not caring about privacy in general. And "privacy" isn't just one thing, there are tons of levels in many different areas.

Agreed, however I didn't say you didn't care about privacy. Just that you traded it away for some (stupidly minuscule IMO) benefit.

I think this is the disagreement. You are treating privacy as a binary thing. That once you trade any amount of privacy away, it is entirely gone.

That is not how most people approach privacy. Most people view privacy as a spectrum. Someone can choose to sacrifice some (but not all!) of their privacy to get a benefit.

What's key is they should be able to prescribe the limits as the poster above has described. We should be able to severely punish companies that lie about their behaviors, because that would make it impossible to make informed trade-offs about how much privacy we are truly giving up.

I can't agree with this more, and I really hope we can see some legislation in the US that creates very strict penalties for lying about this stuff, and possibly some regulations that can make storing user data more risky to the company doing it, which would hopefully put pressure on them to reduce or eliminate data storage where possible.

There is of course a difference between _everything_ and _nothing_ being private, but once _something_ ceases being private it ceases being private forever.

If you let some company, which will do who-knows-what, record voice in your house, you no longer can have a private conversation in your house, since the company will record them, possibly listen to them, and ultimately probably leak them. You completely lost that part of privacy, without for example losing privacy regarding visuals.

>What's key is they should be able to prescribe the limits as the poster above has described. We should be able to severely punish companies that lie about their behaviors, because that would make it impossible to make informed trade-offs about how much privacy we are truly giving up.

I do agree there need to be more regulations around this (yay GDPR), although even if you kill the company dead, you cannot un-ring a bell.

I strongly disagree. If I tell something to my attorney it is still relatively private, but obviously not if I had never told anyone.

If I have a deep secret and I tell my significant other that secret asking for their confidence, it is still relatively private but not as private if I had never told anyone.

I agree that, if you absolutely need total control of information, then the only way to 100% guarantee that privacy is to tell no one ever.

But I think being absolutionist when it comes to privacy expectations means you may have a very hard time convincing others of your opinions.

> If you let some company, which will do who-knows-what, record voice in your house

This is exactly the point we are arguing. We should be able to find out from the company what they will do, and if they deviate from promises they make they should be held accountable.

I should be able to let some company record my voice in my house, with strict conditions on what it will do.

That's not how these devices work.

Stop and think how ludicrous it is that your being able to say to the room "pick a random song and play it loud" requires fiber on ocean floors and huge buildings running servers and the rest of it.

We're being socialized into thinking privacy is an impediment to progress.

It is just a very bad idea to create a turn-key surveillance society and hope and cross your fingers it never gets used.

Yes, but I don't want to store every single movie, TV show, song, and news story locally and find a way to get them to myself without using the internet.

I'd love if more stuff happened locally on these devices, and I experiment with various other systems every now and then. I even have a fully self-hosted home automation system running on open source software using local-only radio systems (z-wave) to control things like lights, garage doors, and door locks.

There are things that do require some kind of internet access though, and other things that become much more reliable and easy to use when internet connected.

>Stop and think how ludicrous it is that your being able to...

I have, and so has my family. And we have made the decision that the tradeoffs are worth it for us.

Why can't people understand that I'm an adult, everyone in my household is an adult, and we are able to understand and comprehend the situation, and make a conscientious decision to accept the trade offs.

I'm not an idiot, I know there are risks, but the risks are low enough for me and my family to make it worth it. Just like how there are risks simply just getting in a car, and there are risks every time I turn my oven on, but they are all acceptable.

Maybe you should stop and think just how ludicrous it is that you seem to think it's impossible that someone else could find value in these devices, and that some are okay with limited tradeoffs with their privacy in return for devices and services which improve their lives. And maybe you should listen when someone tells you that they've weighed the pros and cons and decided it was worth it, rather than quickly jumping to the conclusion that they are wrong, didn't give it any thought, or are somehow being tricked.

You are claiming I am saying this.

I am saying (a) we can get these capabilities without damaging society so lets give it try, and (b) your "choices" affect the rest of us.

It is like second hand smoking.

But I think it's wrong to throw the baby out with the bathwater here. I agree that there absolutely needs to be more regulation around this area. I'd personally like a law that mandates a tone or sound (and lights) when one of these devices is recording. I'd love some laws on the books in the US that gives those who don't want in this ecosystem a quick and easy way of ensuring they won't end up in them.

I also think that there are a LOT of things that all of the big players in this space can do better (like the by-default recording and storing of all the voice data indefinitely). But I also think that there has to be compromises. Saying it's ludicrous that these devices use the internet is the pendulum swinging way too far in the other direction, the internet-connectedness of these is basically the biggest selling point.

But I'd be over the moon if they did all of the simple commands "locally", and I'd love if users were given more control over their information (something like a weekly/monthly/quarterly email asking if Google can use the following list of recordings to train their models would be awesome IMO).

There is so much hyperbole and just straight up wrong information gets thrown about so often with this stuff, and there's a lot of extremism when it comes to privacy, especially on this website (I know this is a loaded word, and I don't mean to include any additional connotations with it, I just don't know how else to describe this). I also don't want to enable turnkey surveillance, but at the same time I don't want to live as a luddite to prevent it. There has to be some middle ground, and if that means it makes dragnet surveillance easier while improving the lives of many people, then I might be okay with it.

For very sound and empirical reasons: historically, intrusive surveillance has always gone hand in hand with tyranny.

Good for learning to analyze voice on the other hand, but your point is still valid. We will get a compressed voice service that easily can be held in a small device at some point, that makes this ridiculous data exchange unnecessary.

https://www.youtube.com/watch?v=b9LQX9cLnZk

Meanwhile my mother's flip phone in 2003ish also had on device voice control, and it worked pretty damn effectively

If you don't warn others about a listening device you may even be running afoul of wiretapping laws, depending on your location.

I do not deserve to have my privacy obliterated.

I'm blind and in a similar boat, but at least I can tie all my appliances into my Home Assistant instance, and never have to worry about being stuck out of reach of a keyboard or phone. Even so, as a blind person, finding accessible weather websites is an absolute pain in the ass. I can get daily forecasts, no problem, but sometimes I need an hourly forecast to plan bus commutes, and nothing beats "Alexa, what's the hourly forecast at 4 PM" for getting a quick answer without having to check half a dozen sites searching for actual textual information vs. a damned infographic or radar display.

I'd love to build her a Mycroft-based setup, but she doesn't need a cluster of RPis and USB microphones hanging around on her desk, and even so that ecosystem isn't ready yet. I say that as someone who literally spent days piecing together a HassOS Mycroft addon and hardware setup that barely hears his commands from a few feet away and triggers all sorts of false positives, and even so I don't think its weather skill does hourly forecasts. I own an Alexa as well, but it stays unplugged unless I need it.

TLDR: It sucks, but don't just classify all Alexa owners as dumb and deserving of victimhood.

Edit: Rereading your comment, I see that you don't really believe that. Apologies for the kneejerk reaction of my own, but I've seen this reaction enough from folks who do that I was finally tempted to speak out about it. I really wish there was an accessibility-oriented privacy-focused voice assistant, or at least, I wish I could pay more to Amazon to opt out of their more invasive practices.

You don't have a smartphone? Or a laptop with a mic and/or cam but no hardware switch?

I have one of these things precisely because it is not as easy to overlook as a smartphone's assistant app always listening in.

Even your dictations on Windows is there too (Win+H). https://account.microsoft.com/privacy/activity-history?view=...

[1] - https://blog.discordapp.com/how-discord-stores-billions-of-m...

A more substantive critique would be that these services don't let you set a retention policy. Google Vault lets an organization automatically delete old email, but other than Snapchat, you don't see this for consumer accounts, so you have to do it manually.

When gmail was first created, I was one of the people to first whine about them trawling through messages for URLs and crawling them. Some friends and I would link to files that didn't set content length and would feed them data forever. They eventually caught on to that.

That seems spectacularly unwise (or at least useless) for all but a minuscule portion of people. Either you stop feeding them data after a short while, and don't care, or you keep feeding them data, and I'm pretty sure Google cares less about (and pays less for) data transfer than you do.

Best case scenario you're trickling data to them and just keeping the connection open as long as possible. Even then, Google likely cares about one connection and the resources it takes far less than you do (even if you care little).

If it makes you feel better, it's maybe worth it, but it feels akin to carrying a marker with you so you can cross out Google's name in any newspaper you happen to be reading. It's an act of protest, but its impact leaves a lot to be desired.

Their servers save the data forever using Cassandra.

You could do it like Snapchat and let the author decide the retention policy, or like email where each recipient separately decides what they want to keep, or like a mailing list archive where you're basically out of luck. What makes sense for a chat app?

Arthur: On display? I eventually had to go down to the cellar to find them.

Mr Prosser: That’s the display department.

Arthur: With a torch.

Mr Prosser: The lights had probably gone out.

Arthur: So had the stairs.

Mr Prosser: But look, you found the notice, didn’t you?

Remember when the CIA gave unknowing citizens LSD? Remember the unclassified docs that revealed their intentions to be able to basically warp a mind, and control it at their will? Well, today, we hand them the data for deep examination. Our emotions, intelligence, etc. will be revealed by a quick scan of our face. Just imagine the databases these government agencies have available. I imagine labs filled with scientists and engineers trying to crack the data.

It sounds conspiratorial, but come on. Imagine if you held humanities brain within your hands. Would you not study it, trying to figure out how to make it work more to your personal needs? Or maybe you'll tell yourself it's to make society better, to make it safer.

Just a note: I'm not accusing the CIA exactly of doing this. There is no doubt in my mind, though, that somebody is out there doing this (the CIA seems most reasonable). Sorry for rambling, I've been awake for 24 hours (is there irony in that?)

Then realised it was going to take like 6+ hours (if not days) to get it to play music and set timers, so gave up on it

Ecosystem might be more mature now, when i looked there weren't any good spotify apps you could get for it