Hacker News Reader

I take a slightly different tactic when arguing for privacy.

https://en.wikipedia.org/wiki/History_of_violence_against_LG...

It wasn't very long ago that the private lives of the LGBTQQ community risked physical violence.

Even today, people risk physical violence for their personal opinions: http://www.wnd.com/2015/11/black-lives-matter-critic-fired-f...

Privacy is the only tool that protects in a society where free speech is not respected. So no, you probably don't have any secrets worth protecting, but others do. It's these victims that matter- and you are turning a blind eye to the defenseless when you are saying your privacy isn't important.

LGBTQQ isn't speech or an opinion, but I see your point. I think the argument is still kind of ridiculous though.

How is it not a speech issue?

While I agree with you, I think I am starting to suffer from a kind of "privacy violation revelations fatigue". I have read the article and thought "so here is another gimmick that steals away a bit of my privacy, so what? no big deal". And that's scary.

The problem is there is no coherent programme on how to counter the privacy-encroachment trend. There is a bit of general dissatisfaction, some idealists urging us to repent and leave the clouds and a good amount of profiteers trying to benefit from that dissatisfaction. Meanwhile the forces behind this trend are organized and powerful.

Another aspect to the problem is that honoring privacy as a human right is a social issue, not a technology issue.

The place to work on solving it (and, like all social issues, it won't be "solved," just "solved for now") is in the public and political spheres. We need to talk about this. We need to have the conversations that describe "what does privacy look like" and "how can we get the benefits from these gadgets while protecting our human rights".

Right now, no one's talking about a world in which you can have your cell phone listening to your conversations, while ensuring that private companies and government agencies can't make use of that willy-nilly. The place where those restrictions have to happen is in the laws we need to develop to keep them all in check.

We can't go back to a primitive, disconnected world, but right now, no one's leading us forward either.

> The place where those restrictions have to happen is in the laws we need to develop to keep them all in check.

IMO we need that, but we also can't rely on that. Things like the NSA have shown that the laws don't really matter. Yes, we should be striving for those laws, but we should also not depend on them, when they can so easily be ignored.

There is a technological aspect, and that's putting the voice processing back into the device itself. We can still get the benefit of mass machine learning on failed voice samples by asking users to submit the failed sample only, playing it back for them so they know what is being sent, and deleting the sample once it has been processed. Then an updated speech model can be sent to the device. Or, just do the retraining on the device, too.

> The problem is there is no coherent programme on how to counter the privacy-encroachment trend.

IMO, there sort of is: FOSS

I have my own "Alexa". It uses FOSS, and recognizes simple commands, like: lights off. Play/pause music. Add something to the grocery list.

Unfortunately, it's a lot less feature-full. I really want an Alexa (or the like), but I personally refuse to get one until I can run it on my own servers, and block it from getting on the internet. And unfortunately, that doesn't look like it's gong to happen anytime soon.

I think the big problem is the inability to run these systems without "big data" behind them, and I'm not sure what's the answer to that gap.

What verifiably private options are there for speech-to-text? It seems like they'd have to be FOSS but most of the "FOSS" solutions I've seen still use a remote server for the actual processing.

It doesn't have to be FOSS, all you need is software designed to run offline, which can be network isolated, e.g. in a virtual machine. Dragon Naturally Speaking (from Nuance) in a Windows VM could perform the speech recognition and relay commands to a Linux VM.

See also the http://www.voxforge.org effort to aggregate OSS acoustic models.

> It doesn't have to be FOSS, all you need is software designed to run offline, which can be network isolated, e.g. in a virtual machine.

It depends on who you ask, some people may only be willing to run FOSS stuff |;)

But yeah, I've looked for a "offline: FOSS or not" solution, and haven't found much.

EDIT: I can't reply to you because thread is too deep I'm assuming, but: yeah, it looks like it does. Unfortunately I probably won't try this because it requires windows, and I'm not really wanting to set up a VM/Wine to accomplish this. But you are correct.

Dragon works offline, http://venturebeat.com/2014/08/22/nuance-dragon-dictate-4-is...

> What verifiably private options are there for speech-to-text?

Unfortunately not a lot. That's why I said it was "sort of" and answer.

I've tried http://julius.osdn.jp/en_index.php and https://github.com/cmusphinx/pocketsphinx in the past, with very limited results. Though it's been a while and they may have gotten better.

FOSS is from another era. Back then Microsoft were the bad guys and FOSS was largely a response to their hegemony. Ubuntu bug #1 was "Microsoft has a majority market share" and it is kind of resolved now but increasingly irrelevant. Now it is a different kind of challenge.

True, if you are a geek, you can use open-source software to cobble together some resemblance of popular cloud services but under your own control. There are even detailed guides on how to do it (see also: https://github.com/sovereign/sovereign). But that's a lot of nuisance and in the end what you get is an inferior version.

> Now it is a different kind of challenge.

Howso? I don't see it as any different, but instead of Microsoft, there are multiple organizations.

> FOSS is from another era. I'm not sure what you mean by this. FOSS is still current, and quite active today. In fact, I'd say even moreso.

> But that's a lot of nuisance and in the end what you get is an inferior version.

I agree. That's what I do now, but it's not nearly as good. But what I was implying was that we have the potential to create a much better version, that's trustworthy.

> While I agree with you, I think I am starting to suffer from a kind of "privacy violation revelations fatigue". I have read the article and thought "so here is another gimmick that steals away a bit of my privacy, so what? no big deal". And that's scary.

Is it though? To be honest, this article is pretty light on substance. I read through it all because the guy is a good writer, but the only revelation here is him realizing the microphone device he put in his kitchen is... gasp, listening to him.

If it turned out Amazon was actually recording and storing everything, that would be quite the revelation. If it turned out your data wasn't being deleted when Amazon said it was, that would be quite the revelation. Otherwise, there's just not a lot here other than a semi-humorous story about accidentally insulting a robot.

Sure, there is nothing of substance revealed in the article, but still the gist is that we should be vaguely concerned. My point is that I am tired of being vaguely concerned.

I don't have that view of HN, it's just that most of us understand that corporations are not evil by default, they are not part of a bigger scheme that wants to enslave the world.

That said, I believe that most HN are very privacy-aware. Indeed one of the reasons Echo wasn't a huge success is privacy invasion.

If you read the latest threads regarding Paris Attacks, you'll see that most HN users are ready to give up security for freedom (if there ever was such a trade-off, I'm not even sure anymore).

> Indeed one of the reasons Echo wasn't a huge success is privacy invasion

Most modern phones passively listen to your voice too, and people carry those with them everywhere. They are present in the most intimate situations, rather than (e.g.) just in your living room. So I don't think there's a reason for people to feel differently about Echo than about Cortana on Xbox, or "Ok Google", or any other voice activated interface.

> it's just that most of us understand that corporations are not evil by default

Perhaps not evil but they are amoral.

If a course of action is not illegal and would increase revenue then a company really should be doing it. Particularly if their competitors are doing so.

> it's just that most of us understand that corporations are not evil by default

That's not an objective fact; it's kind of condescending to present it that way. I would say that they corp's are amoral.

>they are not part of a bigger scheme that wants to enslave the world.

Corp's act in the self-interest of the directors without regard for whether their actions increase/decrease "world enslavement".

There are also legal differences in corporate structures, e.g. benefit corporations, https://medium.com/@ystrickler/resist-and-thrive-1d36819853c...

"We want Kickstarter to be similarly in sync with society. Earlier this year we became a Public Benefit Corporation. This means we are legally obligated to consider the impact of our decisions on society, not just our shareholders. Though we are still a for-profit company, as a PBC it’s very different from the expectation that for profit companies maximize shareholder value above all. It acknowledges and embraces that you are a part of a larger community."

Yeah, I know, there are exceptions and special cases, I'm talking about your ANSI standard corporation, and I think that skepticism is prudent until something like the PBC becomes the dominant corporate structure (or until you can buy iced tea in Hell, whichever comes first).

Privacy and security are quite similar when you think about implementation.

With respect to both, it's impossible to achieve it 100%. As the amount of applications running and the nuisance to which they run increase, it gets trivial to break either.

Developers have long agreed that security is an arms race you really can't win. It's easier to break into than it is to secure. What mostly stands in the way of security breaches is that it's generally frowned on to do and, failing that, is illegal.

The same might need to be true (or truer) for privacy. Privacy is harder to make people understand, though, given the muddiness of the area.

Privacy breaches will never be reined in, therefore. We need to deal with the new reality of ultimate transparency. This could be okay as long as everyone has the same access to that information.

Private information access inequality is the issue.

I don't see it as people wanting to be listened to and watched. I see it as people automatically accepting new technology if it provides a novelty and / or a real benefit without considering consequences. It's also a status symbol to have the newest and coolest tech. Smart homes will simply be the new social status indicator. Technology is way too liberal, and it's leading us in a direction that we might not be able to revert back from.

>This is a direct path to a totalitarian surveillance state...

My critique of this argument is that it has no bounds. Totalitarian surveillance states are very possible without modern (2000+) electronic devices and have existed for a good half century. To add to that, there are many other technologies that are essential to the kind of centralized bureaucracy that totalitarian states generally use. Without electronic communication, it would be extremely difficult to disseminate policy. Without electronic storage, it would be very hard to collate descriptions of suspects and centralize information. The parts of the Catholic church of the middle ages would very much have liked to run a totalitarian state - but it was beyond the technology of the time.

So, if you think we should oppose technology that allows states to survival and terrorize their citizens, there is a longer list to go after: police, vehicles, computers, electricity, etc. All of these technologies gave rise to fears that they would restrict freedom and those fears were ultimately realized. In many ways, we are far less free than people who lived in societies without electricity or cars or any of those things. However, we are also safer from many groups that used to use their freedom to take advantage of others (though not all, obviously).

I agree that this kind of electronic surveillance (or, more specifically, the ability for machines to collect and collate information) is transformative. It's similar to other technologies that changed things - the telegram, electricity, plumbing, etc. It will alter the balance of power and we, as citizens, should think about how we want this technology used and not used. However, I find the idea that this technology _in particular_ is any more or less totalitarian than ones that have come before it to be silly on its face.

Pre-automation surveillance states were people and paper-intensive. They didn't scale. The most powerful analytics were out of their reach. They couldn't sweep up everything and discover what was important. And, in the end, they were overthrown by a combination of discontent, internal sabotage, and a variety of external influences.

Be that as it may, the FBI isn't above reaching back to the old ways. They run 15,000 paid informants nationwide, backed by the information dredged up by pervasive surveillance. Parallel construction is just the beginning of what that cyberspace/meatspace combination is capable of.

With the NSA and pervasive surveillance, it is possible to build an eternal, indestructible static system, governed by an elite, and with the appearance of representation.

>Pre-automation surveillance states were people and paper-intensive. They didn't scale.

Tell that to people who lived in eastern block countries.

I agree that this technology, like all communication technology, makes it easier for governments to repress their people. However, I don't think there's any reason to believe this is the straw that finally subjugates all mankind under tyrannical government or that this technology will be any different than previous advances. As I said before, this is only one of many technologies that are necessary to run the theoretical nightmare scenario you've outlined.

The main difference, from my point of view, is that that other technologies are older and the doomsday scenarios predicted with the rise of telephones, computers and other advances have not come to pass. At the same time, we've had plenty of time to appreciate the advantages of older technology and balance its use in society.

For the recent Paris terror attacks, at least some of the planning was done over text messages. The text messaging system is a better fit for the kind of dystopic observation you're describing: the infrastructure is more centralized, there is lots of legal precedent for observation, there are weak safeguards and no fancy voice recognition is necessary. Of course, the only way we know about the text messages is that they found a physical phone - because despite 20 years of commercial availability - such monitoring is not in place for text messages.

> The problem is with a big trend in society that we want to be listened to and watched, because it makes life more convenient on a personal level.

I would argue that it isn't convenience that attracts people, its narcissism. Society has been surrendering their privacy for years with Facebook. We are comfortable with it. We have seen a direct connection between sharing more (location, pictures, videos, etc) and getting more (food, taxi rides, sex)

So of course the next frontier is not having to share anything at all. The next frontier is listening and deciding. Its concierge services that bring you an array of options, specifically tailored to you. You are the center of their world, so as long as you keep making purchases and supplying them with information to better serve you.

There is a significant difference between private companies collecting data and the state doing wholesale surveillance.

I don't particularly care for the first one, I do very much for the second. These are separate concerns. If we can't have data-driven gadgets because the state may get that data, we're solving the wrong problem: we need to get rid of the state.

> There is a significant difference between private companies collecting data and the state doing wholesale surveillance.

I used to agree, but don't we know for certain that this is false now? With the Snowden documents, we know that the USA government, along with others, knows how valuable these private companies' data stores are and will hack their way in so that they can get access (if they're refused when they first ask politely).

I no longer think there's much real difference at all between the private companies collecting the data and then actively/passively sending it to the government, vs. the government doing all the collecting and analyzing themselves.

And the point that I was trying to make is that even if the private companies can collect and analyze the data safely (they can't), that still sets up a society that is willing to be watched. That is the hardest part for totalitarian governments to overcome; if you hand the USA or China or Russian governments a silver platter of a willing population that will accept the surveillance from one entity, they will accept it from another. That's a dangerous place to be.

A catchy way to say it: If you build it, the governments will take it.

>There is a significant difference between private companies collecting data and the state doing wholesale surveillance.

Not under the current rules where the data are treated as "business records" and subject not only to spying while en route, but also to administrative subpoena and all sorts of other "legal" and questionably legal collection methods employed by various divisions of the state.

>we need to get rid of the state.

For sure, the part that wants to collect all the data needs to go.

i understand your argument and the distinction it makes, but unfortunately i think in a societal context, that ship has sailed.

the people chose to carry smartphones with them all the time, everywhere. that decision was given up without a fight, in fact it was welcomed as the best thing that has happened in the technology revolution to date.

not only has the developed world already made this decision, the developing world continues to make this decision on a day to day basis, and their choice is clear; devices on everyone, everywhere.

So in order to avoid a totalitarian state, no more personal decisions based on my own analysis of the risk/reward. Right, got it!

> But it's NOT a personal decision.

> When we have a society that so willingly accepts and even desires to live in a surveillance state, well, that's a problem. This is a direct path to a totalitarian surveillance state that would be much much more invasive than anything written about in 1984.

The cognitive dissonance in these two sentences is staggering.

How so? I don't think so. Perhaps I wasn't clear. If you make a decision about what devices to buy, on a personal level, and you think that decision only affects you (a "personal decision"), that's not correct. Your decision to buy an Echo or use Siri affects me, too, and the rest of the people around you.

Every time someone decides to trade some of their privacy for some convenience, they are taking away other people's privacy too, including those of us who do not share the view that it's worth it for the convenience.

If my echo is listening to you, it's because you're in my house. So uh, keep out.

Oh, I didn't mean to suggest that your Echo was listening to me. What I was trying to say is that every purchase of an Echo is a signal to Amazon and to other big corporations that there is a market for these products. So they will improve the products and expand the market. I don't want that to happen; so it affects me in that way. Additionally, how long will it be until I basically can't go anywhere without there being an Echo in place? Must I stay at home and keep all electronics turned off, in order to avoid corporate surveillance robots? Will every cafe and grocery store also have them?

Edit: I'm not allowed to answer any more questions right now, sorry to Pyxl101 and others. Hacker News tells me "you are submitting too fast, please slow down". Maybe I'll be able to come back later and answer your questions.

Big retailers are already running analytics on their video surveillance.

This is sort of a cheap shot but I am honestly curious, how do you reconcile this point of view with your goal of building an all seeing barometer?

Yes, that is actually a very good question. It's true that I create weather forecasting systems that rely on barometric pressure data that's collected from smartphones, and that those systems are privacy-sensitive.

It's a difficult line, of course. My goals are to reduce the effects of global climate change by better understanding the atmosphere. If we can provide more accurate wind speed information to wind farms, more accurate solar irradiation information to agriculture businesses, etc, then we can radically improve the world's preparedness for large scale global climate change.

For me, the issue here is deciding which problem is more important to work on. Mitigation of disastrous effects of climate change, or choose to not use any sensor data from smartphones due to my privacy concerns. So far I have aired on the side of climate change being a bigger problem, and that I have a helpful set of work that can make a difference, so I will do my work and hope that it helps.

I think it's a fair question, and I'll remark that I think it's a shame that we have to distrust everyone by default because there are a number of potentially wonderful things that can be done such as pressure.net with this kind of data. As far as that goes, I definitely prefer and would be more trusting of someone who is mindful of privacy concerns than someone like Larry Ellison who I believe would squeeze every nickle of profit out of the same types of projects without regard for anyone's privacy.

There is already a huge market for modern, high end mobile phones that passively record the environment constantly, in order to wake up and respond when their user says "Hey Siri" / "Ok Google" / "Hey Cortana". Users tend to carry those with them everywhere.

I see that your company plans to "build[] an open platform for atmospheric data collection using smartphone sensors". It looks like you support Apple and advertise their products on your company's website: https://www.pressurenet.io/blog/iphone-6-has-barometer/

"We’re excited to start including iPhones in our study of the atmosphere!"

If you believe that "every purchase[] is a signal", then what signal are you sending about Apple iPhones?

I have not researched it in detail, but it seems like you're relying on data collection from large networks of smartphones in order to put together accurate weather forecasts. That's cool! But I find it hard to reconcile with the view that smartphone data collection (from people who volunteer to participate, or who use a product that has a collection feature) is invasive.

Meteorological data collection is probably opt-in, and pressure+gps data is less invasive than speech+gps data.

Variations of this argument are everywhere though, for every product you may not like. Everybody who buys a burrito at Chipotle sends a message that it's ok to put cilantro in rice. What happens when there aren't any cilantro free burritos? What happens when they put cilantro on pizza? Cilantro may or may not be a human rights violation, but it's a weak argument that people shouldn't buy a product you don't like because then that product will continue to be sold.

For a while it was considered creepy to have a camera on your phone at all times. There was no way to know if you were filming/photographing at any time.

People in certain secure facilities or attorneys in many courtrooms weren't allowed to bring in phones with cameras because of respect for the privacy/security of the information and people within.

Now there are no phones without cameras and the insistence on privacy in many of these facilities has been relaxed to saying "please don't take pictures". Not because it's not important, but because of consumer decisions on the whole demanding cameras, to the point where they are in every viable option and it's not feasible to ban them. And privacy/security are eroded.

So when you make a consumer choice like this you're giving our society a tiny little push in that direction. You're pushing always-on always-recording voice monitoring software in the most intimate places of our lives to become the new norm.

It's not that simple in the United States, because of the 4th Amendment. A major factor in determining what would be an "unreasonable" search and seizure absent a warrant is what people reasonably expect when it comes to privacy.

If enough people put in listening devices that report to third parties in their homes, that can shift what is a reasonably expectation of privacy to where the government can put listening devices in homes without a warrant.

There was a good essay on this a while back by Alex Kozinski, the chief judge of the 9th Circuit Court of Appeals: http://www.stanfordlawreview.org/online/privacy-paradox/dead...

Spoiler warning: that contains a major spoiler for the terrific Isaac Asimov short story "The Dead Past".