Hacker News Reader

Comment by superkuh | Hacker News Reader

More importantly, the headline should read: "Police use high time resolution data from modern water meter against man charged with murder."

Basically this Abstruse Goose comic, http://abstrusegoose.com/553 , about the dangers of utility data logging have come true. The greater the time resolution of such things the more can be inferred. Such high resolution is completely uneeded and complex in order to just charge for usage. The new wireless meters could simply send daily lump sums and still be wireless and easy. But instead they abuse the functionality in order to intrude on the lives of everyone.

View on HN

Isn't it a good thing to record data that helps solve a murder? The police are still getting it with a warrant and presumably from physical access to the device, so it's not the same worry as bulk monitoring.

In the case of a murder, it's a good thing. The risk is always the chance of abuse, and although in the case of water stats it's hard to imagine how you'd abuse it, having more information always means a higher chance.

Ideally we'd want to get the benefits without the abuse, but that requires a solid bureaucracy and strong rules.

You can tell when someone's home or not from utility usage.

Timers and automation are a thing.

Those patterns would be easily detected in the dataset.

I tried logging my electricity use. You can easily see what time I wake up each day, when I get home, whether I watch TV or not, whether I cook using the microwave or the hob.

It doesn't show up on my graph, but I'd bet from the data you can see exactly when I go to bed, since I plug in my phone (+1W) and switch off the last light (-5W).

Within all this, the fridge compressor turning on and off is easily ignored.

Thus you have a training dataset. Train a statistical model, and with enough programmability in your smart home it will be easy to forge a rather normal-looking pattern in your absence. And I'd probably use some special fire-proofed energy dissipation device instead of the more powerful real ones (e.g. the hob).

EDIT: the internet surfing patterns are another story. But you can probably VPN to your home through some newly-installed network interface and access the net as if you were there.

Exactly so. And then there's the corollary of times you come home, walk around back and drink beer on the patio watching the sunset with friends and eating takeout.

Yes, by the exact amount of current used, the logger can tell what you've got on at any moment. It can be correlated with water usage as well. It'll log how many times you flush, and whether you brush your teeth.

I wouldn't be terribly surprised if the government tries to require a device that logs all our conversations at home. It'll be sold as "crucial for law enforcement".

The Echo is a technological marvel. But who's to say Amazon doesn't get a "national security letter" one day where they're forced to send a patch to the machine to have it log everything?

Some day, if I'm bedridden or some such, I'd welcome an Echo in the house. Until then, I'd rather get up and turn a manual switch. I could use the exercise anyway :-)

They're not going to have to patch what you can touch because they already want for themselves the information the government is going to request.

May I ask how you collected this data? I am extremely interested in doing the same. For me right now, my power usage is "stuff goes on, money goes out".

I have a CurrentCost power logger, given to me as part of the British government's insistence that power companies try and reduce power usage. The choice was a power meter or some low-energy bulbs.

I was lucky to move house just after the rules came in, but before cheap, residential versions of the logging devices were on the market -- mine has a serial port! It outputs a line of XML every 6 seconds showing power use, measured with a probe which clamps around the live wire where it enters the house.

I've since moved again, and my newly built apartment has the wires hidden, so I can no longer use the probe.

http://www.currentcost.com/product-envir.html -- but make sure you get one with a serial port.

Ha, reminds me of when I brought my iPhone into repair due to a faulty battery. They took a look at the charts and noticed a high usage of power in the morning and then it would stay charged the rest of the day while I was at work. When the device was unplugged was when I left.

> Those patterns would be easily detected in the dataset

You left off the implied "unless the intent of those systems is to create the appearance of occupancy". You know that a pattern associated with waking up happened. You don't know if I'm home or not. Now add in the case which happens to me sometimes where my morning ritual automation kicks in because my phone is home (but I'm not).

If I recall correctly, this argument has been successfully applied, from a legal standpoint, in the context of traffic from IP addresses on shared resources not mapping to individual user behavior (and, relatedly, in the context of a vehicle being photographed in violation of the law not proving that the vehicle's owner was driving the vehicle at the time the law was violated).

I imagine it'd be a harder argument to make, before a jury in a criminal trial, that the dataset showed the house was occupied because the occupant had rigged beforehand an elaborate system of intentional resource wastage to fake occupancy in the way that would fuzz a police investigation, as opposed to the much simpler explanation "the suspect was home." Burden of proof might be high.

> It doesn't show up on my graph, but I'd bet from the data you can see exactly when I go to bed, since I plug in my phone (+1W) and switch off the last light (-5W).

You'd probably have a viable business model if you could actually do that. Existing options like Neurio cannot reliably detect devices pulling single-digit power draws.

Assuming that digital electric meters are fair, it's quite easy to use them for this purpose. Most houses in the UK are fitted/will be fitted with them.

Circumvention is not the solution. It doesn't solve the problem for Average Joe.

You've always been able to make this guess from lights on/off, windows, etc. It might have become easier to some people.

its a fine line. There is a great fear of false positives from such inference. Let us imagine for a second that there is a legitimate reason for that water use that he doesn't wish to disclose because it would incriminate him for something else?

Hypothetically speaking if you were doing some grey-hat, "illegal" probing/hack at the time of a murder the police question you about; admitting to your actual activity could result in more time behind bars than a murder. If this activity was somehow vaguely recorded and correlated with an activity of the murderer then you have a serious problem.

"A legitimate reason for that water use that he doesn't wish to disclose because it would incriminate him for something else" is an odd definition of "legitimate." ;)

Just because something is legally a crime doesn't mean it's not legitimate.

Actually, that's exactly what it means.

No, it doesn't. Legitimacy (or ethics) is not derived from law (or the reverse). It's possible to be guilty of a crime (by the law) for legitimate reasons. These extenuating circumstances are improper generalizations or other imperfections in the law.

But a protected right in the US.

e.g. he needed to refill his secret swimming pool for his pet crocodile that he doesn't have a permit for.

"legitimate".

I can't refuse a valid search warrant on my apartment because something found within "might incriminate me" for something else.

Take criminal cases where they find drugs in your system and use that to make it look like you're more apt to commit said crime.

>Isn't it a good thing to record data that helps solve a murder?

Recording everyones data everywhere just in case it might be relevant to an extremely rare event like a murder is very dangerous. The use of NSA recording capabilities by employees to digitally stalk victims of their affection (LOVEINT) is a clear example of how badly this kind of collection can go wrong.

Your first statement is correct, but not because of what you said in your second. Google "ubiquitous law enforcement" for some good reading.

That was an excellent read, as Vernor Vinge usually is.

I've certainly used anti-surveillance arguments like "what if some employee misuses the data", and I mean them, but it's not my fundamental objection. On a much deeper level, I worry about surveillance as an inherent harm. Partly as the death of privacy (a right in itself, to me), largely as a de facto power transfer to the state.

I think Seeing Like a State is right when it argues that states depend on 'legibility'. If it can't be logged, measured, or predicted, it can't be handled on the statistical level where states have to operate. And so everything that increases the legibility of our lives eases the real power of the state, without any need to expand its legal power.

> Seeing Like a State is right when it argues that states depend on 'legibility'.

http://opentranscripts.org/transcript/no-neutral-ground-burn...

So how much do we hold back technology and the use of it to protect against people committing crimes with it (as mentioned above)

This data when used properly (which is probably the case 90%+ of the time) helps to make the world better, or provides efficiency for the company who uses the tools.

>So how much do we hold back technology and the use of it to protect against people committing crimes with it (as mentioned above)

Would you like an arbitrary line in the sand? Sure, let me give you one. The crime you're trying to stop has to affect at least 0.0001% of the population before I'll even consider lowering the quality of life of 99% of the population to protect against it.

0.0001% is an order of magnitude over the percentage of the US population that got murdered or manslaughtered in 2015, if you were counting.

But what is your line in the sand for then? Collecting the data? Or allowing people to access it via warrant?

Even having the discussion. It's default "no and fuck off" until then.

Unfortunately, that assumes that "you" have that choice.

Individually, I'm sure you can rip out, mutilate, deface, damage or what have you to devices that report fine-grained time-series data. But the moment you go anywhere else, you're being recorded, watched, monitored, measured, and recorded.

It's already being done, and it's way past time that we do have that discussion. Because without string and stark dialogue, we'll end up with "We're not doing anything wrong, so why should we care?" kind of responses. Or worse yet, people won't understand what something like accurate TS data from power input can lead to.[0]

And as an addition on the above comment, I also wrote a CPU-bound facial recognition app. I wrote it initially for a convention that we have for our hackerspace every year... But it took me a week to do in my spare time, using open source libs (OpenCV, QT). This stuff is just getting more ubiquitous.[1]

[0]: http://www.hennebert.org/download/publications/isspa-2012-ma... MACHINE LEARNING APPROACHES FOR ELECTRIC APPLIANCE CLASSIFICATION

[1]: https://github.com/jwcrawley/uWho

>Unfortunately, that assumes that "you" have that choice.

>Individually, I'm sure you can rip out, mutilate, deface, damage or what have you to devices that report fine-grained time-series data. But the moment you go anywhere else, you're being recorded, watched, monitored, measured, and recorded.

There are a variety of ways that this can be done, most of them revolving around visibility and community education.

Stop using facebook, stop buying voice assistants. Refuse to sign up for accounts that require your personal data. Be offended. Tell cashiers in shops on no uncertain terms that they don't need your email address to sell you things. Be a leader by telling people around you that you have consciously made these decisions, and /why/ you have made them.

Attach your public key whenever you email someone. Inevitably, someone will ask what the attachment is. Use it as an excuse to tell them about encrypted email.

Be the change you want to see in the world.

Early next month I'll be getting an old stock but new in box feature phone from the UK. it's from 2011, it cost me $30 including shipping. It has a browser, SMS, a phone and /nothing else/. I look forward to showing off my new phone to everyone at work and explaining I ditched my Galaxy ace 4 because I didn't feel in control of it. It's partially a token gesture, I can still be tracked via the GSM connection, but it's a talking point. It keeps the problem in peoples faces.

The US government doesn't give a shit about the legality of what they're doing. They have a "secret court" that no one is allowed to ask about that is "authorizing" all the things they are doing. Clearly, you're not going to win them over by town hall debates. What we need is gentle but powerful groundswell opinion against surveillance, the kind that threatens election results.

P.S, you can also move to places where this shit doesn't happen. I moved from Australia to Iceland, and at least 30% of my reasoning to do so was that Australia is actively turning into a surveillance state.

> There are a variety of ways that this can be done, most of them revolving around visibility and community education.

Indeed. But I also think, that much of this community education and "awareness" has gotten people to ignore these types of things. A lot of it comes down to "ehh, whatever" or similar "convenience > privacy".

Facebook's a problem child. You have an account even when you don't have one, because other people put things up there and link to you. Indeed, having an account allows some modicum of control of other user-posted junk, whereas not having an account = no control.

Facebook also has network effect going for it; easy to chat on, and easy to see events and parties around. And being "that guy" means you're likely forgot about events. I'd love a viral solution to claw away from Fb and get my friends on. Ello seemed to be a good candidate but quickly turned into a shithole. I'll honestly have to look into Diaspora and its user-friendliness. Because again, network effects rule.

> Be the change you want to see in the world.

Indeed. I try to be. I have my own IoT infrastructure. Built it myself. I have a few different buildings I do stuff in, and use Tor to tie my network together. I've documented what I've done here: https://hackaday.io/project/12985-multisite-homeofficehacker...

> The US government doesn't give a shit about the legality of what they're doing.

Absolutely agree, being a US citizen. I am looking at emigrating elsewhere. Iceland was a consideration, as was a few other countries in Europe. But it also feels like trading one set of problems for another set of problems.

I think the parent comments are a separate issue from concerns about the NSA. Most people would agree that warrantless access to all data everywhere is a bad thing (smart water meter or not).

In this case, there are traditional limitations on law enforcement (requirement of a warrant) and a specific crime that has been committed.

In my opinion, using a warrant to get the data from the water usage device is no more problematic than using a warrant to search the house.

I think that's a difficult comparison to make because you can reasonably secure your house against unwanted entry and it's fairly likely that breaches could be detected.

Obviously malicious law enforcement could still circumvent that just as they could a warrant depending on their level of capabilities and your house's security, but there isn't a similar path to securing your water usage data against warrant-less searching. Its security is wholly in the hands of a third party and any security breaches are essentially invisible to you.

That said, we're both painting in broad strokes.

So we shouldn't collect data that might help solve a major crime because someone might commit a minor crime like stalking?

We're trading the potential to solve a major crime for the potential for a minor crime of stalking to happen. As long as we require that a warrant be involved in getting the data and there are punishments in place for accessing without a warrant I'd feel safe with them collecting it.

"Minor crimes" like stalking can lead to major crimes like murder. There are 2 major issues here:

1) The information might be used "off-label" for all sorts of personal and political reasons. There are countless examples of this in history.

2) The standard of evidence must remain high. Circumstantial evidence has been, and continues to be, used to convict innocent people of crimes. Giving the authorities what amounts to a mountain of circumstantial evidence could allow them to build a case out of thin air.

"Ladies and gentlemen of the jury, he must be guilty: he entered the bar two minutes after the victim, and left a minute after they did. He took an Uber along the same route and exited at the same address. 3 minutes later the murder weapon was purchased by a man matching his description a block away." ... and so on.

"If you give me six lines written by the hand of the most honest of men, I will find something in them which will hang him." -- Cardinal Richelieu (disputed, but classic)

Imagine what can be done with a lifetime of data from hundreds of sensors?

1) Which is why the data shouldn't be accessible without a warrant and there should be stiff punishments if it is accessed without one. We shouldn't stop all the other advances that come from the technology of these sensors and the help they could provide in solving very real crimes.

2) There's already plenty of avenues for circumstantial evidence already, this is the reason we have lawyers and judges to argue over and prevent evidence that isn't solid

There has to be a line drawn somewhere, lest we end up with putting GPS trackers and video recording devices on each and every person.

That's obviously a slippery-slope argument, but it's not outwith the realms of possibility that if I can think of doing something like that, politicians (especially those!) can think of it as well.

> lest we end up with putting GPS trackers and video recording devices on each and every person.

....phones?

That's at least voluntary. For now...

As long as it's voluntary, it's fine. We should keep educating people about the risks as long as we don't curtail what people can do with the technology voluntarily.

"Voluntary" is easy to say, but the environment heavily suggests certain choices. A (mobile) phone is these days used or needed for authentication: banking, email, real time communication and social media. Having to forgo those things is not without drawbacks.

1. It doesn't come from physical access to the device: the device is accessed wirelessly.

2. The water company collects all this data from everyone with the device.

3. As the Mirai botnet showed, it's not just the government who might have access to that data.

> Isn't it a good thing to record data that helps solve a murder?

You might trust the current custodians of $BACKDOOR, but do you trust all future possible custodians that have dominion over that role? Keep in mind that nothing is technically restricting them to look only at possible murder.

The Amazon Echo records data from all of its users, not just the murderers. It's creeping close to bulk monitoring.

It is bulk monitoring. My four kids each have a google tablet. My autistic daughter has learned a lot through "OK, Google", which means everything said in my household for the past six years is sitting on Google's servers, waiting for America's Stasi to commandeer it. Frankly, it's bullshit that Google didn't know what was going on, or that I didn't either.

When I was a student this murder happened https://en.wikipedia.org/wiki/Murder_of_Rachel_McLean One important piece of evidence was the record of the number of bus tickets sold at the time/place Tanner said he got the bus (there was also a witness that said he was alone). A single ticket was sold, not two he had claimed.

That was 1991.

We either assume the data will more often than not be used for negative reasons, and stop it, or we assume it will be used mostly for positive reasons and allow it to improve the world.

If we do the latter, we treat the minority cases as a crime. Just as we do when police use their access to stalk people.

That being said, I DO think we have poor transparency in these cases of who has access to what. I've always said that if it were publicly available to see how many times someone has run my license place and/or drivers license I would be really upset.

I think it would be important to know if Amazon can or would release information/data which could be used to exonerate someone suspected of a crime.

I wonder what the market is for a utility usage randomizer that fuzzes this data?

While it is likely illegal at various levels, there are probably ways to interfere with the smart meter via RF.

I believe building your own 'storage' units would be better (letting you draw from these as you need, without monitoring) thus having a continuously slow supply of water,gas, electricity coming from the grid.

As long as these storage units don't simply begin to 'top off' when they are drained, the usage could be masked by averaging and predicting demand.

very simplified water day

6am - Toilet flush & hand wash 2 Gal

7am - Shower 36 Gal

8am - Breakfast - Coffee, Oatmeal, Dishwashing - 2 Gal

6pm - Dinner & dishwasher - 8 gal

10pm - flush and brush - 2.5 Gal

So you'd want a constant 2.1 gallons to be fed in from the street per hour, and you'd need a minimum 40 gallon reserve tank to cover that day.

Quite true, but the opportunity for time-of-day rate arbitrage is tempting.

thats why i didnt want them to install the 'smart' meters for electricity. i told them I didnt want it, but they said they would shut off my electricity unless i complied.

You didn't want them installed because you were concerned they might rat you out when you murdered someone? ;)

Reply on news.ycombinator.com