Has anyone worked on fighting back this kind of telemetry/spyware of essential consumer appliances?
I'm thinking something similar to what https://adnauseam.io/ does, but but amplified:
1. Someone reverse engineer what does the device send to which address.
2. Block the particular device to access internet (and make it easy for others too).
3. Constantly send bogus data to the manufacturer so the personal data they get overall loses value or is unusable. Make it easy for a lot of people to do it as well, or even just rent a bot farm.
There's too many legit and good services that end up being turned down due to abuse and DDos, and they don't even bring anything good to the attackers. Why not using these techniques for something actually good to consumers privacy?
In my ever-cynical view, I imagine in most cases manufacturers don't, themselves, especially care about the data from their devices. I see various other motivations:
1) Price increases. It's "smart". Pay us more.
2) Planned obsolescence. You have numerous new points of failure in your product + make repairing vastly more difficult.
3) Monetize collected data by selling it to interested parties. The data quality, or lack thereof, is a secondary concern.
I've worked in a big consumer electronics company and 2. is just a conspiracy theory. Nobody wants their products to be unreliable or difficult to repair. They want them to be cheap to manufacture and unlikely to fail during the warranty period. Everything else stems from that.
Your other two points are broadly correct though. Also data collection is helpful for seeing how customers use products, which genuinely does influence development.
> Nobody wants their products to be unreliable or difficult to repair.
Hard to believe when a printer manufacturer moves a commonly failing part from cartridges to the printer itself, then makes it impossible to repair that specific part without taking apart the whole thing and buying a replacement part for 160 bucks.
> Also data collection is helpful for seeing how customers use products, which genuinely does influence development.
I'll start calling it helpful once it actually improves product quality and usability.
They didn't say it improves the product though, just that it influences development. The incentives of producers are not aligned with consumers' wishes.
Well calling it "planned" adds a conspiratorial element, maybe there's a better word.
The effect comes from many sources. When the engineering science is new, tech tends to be way overbuilt because the creators don't know what tolerances are. There's also the ratio of product cost to disposable income of the users. And to their wish for devices over time.
In any case, the natural tendency of a profit-optimizing system of competition is to throw away values (eg., repairability) that consumers at-purchase dont value. In this case, no biz is going to spend extra to include long-life features at a loss to competitors who don't.
The remedy here is regulation, which is the mechanism which sets a floor on competition: since consumers at-purchase are at an extreme informational disadvantage, the gov. should require all biz to build in long-life features given consumers do want them, long after purchase.
It's not a conspiracy, more of a market failure. The market doesnt produce optimal outcomes in cases where these extreme information asymmetries are built in, ie., by the time the product is around for 10yr to be reivewed, it's retired. There isn't a way of getting that information to the consumer.
Perhaps incentivised is the better word? Improving tolerances and repairability so that products last beyond the common 1-2 year warranty periods costs time and money, and potentially means that customers aren't buying newer versions of those products or paying for first party repairs. I'd imagine that many engineers who actually design/build the products would like to make these things more durable and repairable, but are more often then not told to cut costs to improve margins/work on new feature X or product Y, etc.
If repairability isn’t a design goal, it won’t be easy to repair.
Ease of assembly is often in opposition to ease of repair. Reliability engineering optimizes failure rates during warranty and extended warranty periods. Why would they optimize repairability? It not in their interest.
The defective Apple MacBooks are great example of this effect. Politically powerful design guy prioritizes design over everything else. Keyboard technology is defective and un-repairable. Result: Billion of dollars of warranty claims.
Was that a conspiracy? No. Just the result of poor priorities, resulting in a bad outcome.
I can see you've never designed anything! I can think of plenty of things I'd want to monitor on a fridge if I was a fridge designer.
What temperature do they set the fridge to? Do they ever change it? How often do they deice the freezer? Do they use the ice maker / filtered water thing / whatever? How often? How long do they leave the door open normally? What are the temperature / power consumption profiles over time?
That's before we even get to any of the smart features (I have no clue what they are).
Sure you're right there is plenty to monitor and collect, it's just not very useful for the majority of people I'd imagine, unless you like other people to tell you when to buy something.
>I've worked in a big consumer electronics company and 2. is just a conspiracy theory. Nobody wants their products to be unreliable or difficult to repair.
Nothing cynical there: you are simply correct. There's overwhelming evidence that companies profit from selling user's data, engage in planned obsolescence and deploy numerous tricks to disrupt the second hand market.
I mean, this has been going on for at least 97 years if you include light bulbs:
“The cartel lowered operational costs and worked to standardize the life expectancy of light bulbs at 1,000 hours (down from 2,500 hours), and raised prices without fear of competition, in what has been described as a "classic example of planned obsolescence". The cartel tested their bulbs and fined manufacturers for bulbs that lasted more than 1,000 hours.”
It literally saved customers money to buy the bulbs with lower lifetime because they are so much more efficient. Shortly before incandescents were phased out, I saw markings on packaging like "this bulb uses $10.81 in electricity over its lifetime" on a 40 cent bulb.
They could have standardized on a power efficiency metric, but there's an intrinsic inverse relationship of efficiency versus lifetime for tungsten filaments.
Back in the day companies did not try to eat every tiny bit of consumer surplus, so this cartel was a win-win situation.
Now that we have LEDs, which were marketed to last "forever" when they were introduced, but clearly don't; and running LEDs at lower specific power actually helps efficiency, but they don't.
Phoebus being about efficiency was merely a coincidence.
LEDs have an average life of about 10 years from what I've seen. Yes, their efficiency decreases, but I'm not sure that's such a big thing.
And at this point who really needs "forever"? We're getting to a point where the average person will at most change their lighbulbs 2-3 times in a home (I doubt the vast majority of the population lives in the same place for more than 20-30 years).
Actually:
> Report Highlights. The average length of homeownership is 16 years, with lower-earning and less educated householders more likely to remain in their homes longer.
>And at this point who really needs "forever"? We're getting to a point where the average person will at most change their lighbulbs 2-3 times in a home
Why does that seem fundamentally more reasonable to you then each occupant replacing the walls just once?
I seem to remember that this story is a distortion? Don't recall the details, but that Wikipedia article does say the British government looked into it and concluded there wasn't evidence that the ceiling wasn't justified in technical grounds.
> Monetize collected data by selling it to interested parties.
then they do care about the data, just not the quality of that data. But the fact they are collecting data is definitely of importance or they would not be collecting and selling it.
I’d just want to know who on earth would want such data and what insights could be gained from it. It can be challenging to build models with tons of really good data.
1) Set up a pihole or ad guard or similar and block the requests the device makes. You can probably find someones list or it may already be in the default one.
2) Put all the IOT's devices into a virtual wifi lan that doesn't by default doesn't allow internet access. Then only add in the few places you want them to be able to get to. In general putting IOT devices on a network separate from your real computers is a good idea for isolation anyway since they are likely to have poor security.
For that to happen, the value of the data it extracts would have to be greater than the cost of installing and paying for a cellular modem and the data fees over its lifetime.
I think most of us aren't worth it, as it would take away from the upfront profit of the machine to add features people aren't told about just to get a few years of "Subject opened left door. Temp 39 degrees. Subject used ice maker." or whatever.
I think you're underestimating how cheaply a manufacturer could negotiate a one time fee low bandwidth 5G connection for. If it's setup to only upload in off peak hours it's going to just be free money for wireless carriers. Maybe they'll double dip and buy the data for themselves as well.
I even saw the newer broadcast TV standard was going to have 2 way communication for ad tech.
Not necessary. When not connected for more than a month, a device gets declared insecure and stops its core function or gets annoying until you allow it on the network again. Downloading new firmware is only possible by uploading metrics, and provides all kinds of new 'experiences' of course.
In 2 years, a standard fridge will start blinking its inside lights and show 'error' on the temperature display if you don't let it connect.
If it runs on the their network network and not mine, then any activity will be tied to their IP and the incorrectly correlated with their usage.
So - my privacy stays intact as long as I don't link the device ID with my name or personal details (review that warranty card before sending in, does it have unique ID? watermark/non-visible tag) - hmmm, this sounds like a bonus - I get the "smart" features and don't have to worry...
Pi-Hole with a good collection of block lists is a great option. My samsung TV and a few android devices are the top offenders on my pi-hole admin panel.
I remember there was something similar once for web analytics. The extension would obfuscate stuff by changing values, esp. e-commerce values like price and quantity so that the data becomes quite tainted.
I would hate to see smart stuff taken off the market.
A DDoS could cause the company to drop support faster(Like they already always do), and hurt the people who can no longer use the features on their expensive device.
Besides, if DDoSing got popular with average consumers it would never stop, and they'd go after everything that has any privacy risk(AirTag/Tile comes to mind), no matter how critical it is to some people's lives.
Admittedly a bit of a slippery slope argument, but less so in an age where there is a significant minority that would love to undo all tech from the last 70 years.
Instead we could be fighting for laws requiring that that all smart devices use an open and app-capable OS, or that all features exposed via proprietary connection to their server also be exposed via local API.
We spent several hours a week dealing with, worrying about, and developing mitigation strategies for losing out keys, quite possibly even shortening our lives with stress.
If you aren't one of those people that can just walk by something and passively remember it's location, IoT is great.
Tile doesn't completely replace constant vigilance and planning, and carefully thinking about every step, but it does help.
>1. Someone reverse engineer what does the device send to which address. 2. Block the particular device to access internet (and make it easy for others too). 3. Constantly send bogus data to the manufacturer so the personal data they get overall loses value or is unusable. Make it easy for a lot of people to do it as well, or even just rent a bot farm.
requests probably need to be send with valid serials, in which you can't effectively anonymously flood the telemetry by yourself.
given that there is absolutely no way even a small percentage of Samsung fridge (why even buy one?) users will care about this, all it does is reveal the participant's identities and motivations.
for 3. I wonder if you can go a step further and pummel them with extra data. like insane amounts of (bogus) data. At some point even plain s3 storage costs will become problematic for them.
I’m trying to do this now, but dumb high-quality appliances are hard to come by and much more expensive. I don’t think there’s any dumb TVs at the state of the art.
Dumb TVs are sold as digital signage. They are the state of the art for durability/reliability but they typically have a screen from the "previous" generation, so for example they might not have HDR.
Some kind of legislative protection would be nice too. eg any mechanism that collects or transmits telemetry must be able to operate totally separate from any other feature of the device and have a hardware kill switch.
Why would you want to go out of your way to send bogus data to the manufacturer?
I'm all for being able to choose whether or not to disclose that data, but then we'll also have to accept different choices than ours. There's no point in sabotaging others.
If someone's buying the data they will care sooner or later. After all, if they wanted bogus data, they can generate it themselves instead of buying it.
I went through a serious attempt to remove all of my resumes on the web. Paid a firm, whole nine yards.
Several data brokers still have very, very old copies.. and they still sell them.. and recruiters still buy them.. and still contact me.. and still get met with an email politely telling them off.
But that years old shitty linkedin dataset still gets sold to thousands of people for thousands of dollars a year and nobody bats an eye. The recruiters are too stupid to spot the bad data and the brokers too lazy to care.
Usage patterns? Fridge being opened means someone is home. That data point is meaningless in isolation, but can be valuable if you want to use it to confirm/deny other data points - let's say another data broker is trying to get an accurate ad targeting profile but only has breadcrumbs here and there such as IP addresses, user-agents that by themselves don't mean much, but they can use other data points (such as fridge activity data) to link your otherwise-anonymous IP-based profile if they see that the only times this IP lights up is when the fridge was also used recently.
Whether that's currently done is up for debate - maybe there are other lower-hanging fruits that are easier to do, but if you've exhausted all your other options and still want even more accurate profiles, I don't see why you wouldn't do it.
Unless there's cameras everywhere in the fridge (plus advanced object recognition that knows that that bundled up pack of goat cheese inside a ziplock bag is in fact goat cheese) or people are scanning items as they take them out I don't see how that would work and both things seem kinda unlikely so it's weird they'd start with the easiest part which is hook up the thing to the internet.
We're already being sabotaged, by manufacturers - what else would you call this sometimes hidden, non-disablable connectivity/"telemetry", and the disappearance of dumb options? The only question is if we let them get away with it scot-free.
I'm thinking something similar to what https://adnauseam.io/ does, but but amplified:
1. Someone reverse engineer what does the device send to which address. 2. Block the particular device to access internet (and make it easy for others too). 3. Constantly send bogus data to the manufacturer so the personal data they get overall loses value or is unusable. Make it easy for a lot of people to do it as well, or even just rent a bot farm.
There's too many legit and good services that end up being turned down due to abuse and DDos, and they don't even bring anything good to the attackers. Why not using these techniques for something actually good to consumers privacy?