Facing Criticism, Germany Switches to Google/Apple's Decentralized Contact Tracing (reuters.com) 71
"Germany changed course on Sunday over which type of smartphone technology it wanted to use to trace coronavirus infections," reports Reuters, "backing an approach supported by Apple and Google along with a growing number of other European countries."
Chancellery Minister Helge Braun and Health Minister Jens Spahn said in a joint statement that Berlin would adopt a "decentralised" approach to digital contact tracing, thus abandoning a home-grown alternative that would have given health authorities central control over tracing data.
In Europe, most countries have chosen short-range Bluetooth "handshakes" between mobile devices as the best way of registering a potential contact, even though it does not provide location data. But they have disagreed about whether to log such contacts on individual devices or on a central server -- which would be more directly useful to existing contact tracing teams that work phones and knock on doors to warn those who may be at risk. Under the decentralised approach, users could opt to share their phone number or details of their symptoms -- making it easier for health authorities to get in touch and give advice on the best course of action in the event they are found to be at risk. This consent would be given in the app, however, and not be part of the system's central architecture...
Germany as recently as Friday backed a centralised standard called Pan-European Privacy-Preserving Proximity Tracing (PEPP-PT), which would have needed Apple in particular to change the settings on its iPhones.
When Apple refused to budge there was no alternative but to change course, said a senior government source.
The article notes Germany had also received opposition in a recently-published letter signed by hundreds of scientists.
It had warned that Germany's original plan for centralized tracing would allow "unprecedented surveillance of society at large".
In Europe, most countries have chosen short-range Bluetooth "handshakes" between mobile devices as the best way of registering a potential contact, even though it does not provide location data. But they have disagreed about whether to log such contacts on individual devices or on a central server -- which would be more directly useful to existing contact tracing teams that work phones and knock on doors to warn those who may be at risk. Under the decentralised approach, users could opt to share their phone number or details of their symptoms -- making it easier for health authorities to get in touch and give advice on the best course of action in the event they are found to be at risk. This consent would be given in the app, however, and not be part of the system's central architecture...
Germany as recently as Friday backed a centralised standard called Pan-European Privacy-Preserving Proximity Tracing (PEPP-PT), which would have needed Apple in particular to change the settings on its iPhones.
When Apple refused to budge there was no alternative but to change course, said a senior government source.
The article notes Germany had also received opposition in a recently-published letter signed by hundreds of scientists.
It had warned that Germany's original plan for centralized tracing would allow "unprecedented surveillance of society at large".
Re: (Score:1, Funny)
Shut the fuck up and go find a corner to cry in, you weak little beta male bitch.
Re: (Score:1)
Re: (Score:2)
Not really sure why you put rights in quotes. The rights of free movement and free association are enumerated rights in the Constitution.
Are these protesters morons? Of course they are. But these are not "rights" - they are actual rights bestowed by the laws of the land. It just happens that the constitutional amendment that recognizes these rights was written before people modern germ theory was a prevalent idea.
As it turns out, laws written 250 years ago don't always predict circumstances of modern ti
Do they really think nobody's hacking this? (Score:3)
This is gonna be funny, once the first dozen people ruin the entire thing with false data. And probably hack the servers too.
Re: (Score:3)
There won't be any servers since they caved in and it's going to store your data on your own device. You're free to hack the data that's only on your phone however, but that won't achieve much. Just don't get the app then.
Re: (Score:3, Informative)
The only data users submit is a set of personal, randomly-generated codes (which your device would have previously broadcast to nearby devices) and matching timestamps in the case that they get an infection. Those codes and timestamps are forwarded by the service to everyone else’s devices so that they can tell if they saw a match at the specified time, allowing the devices to notify that person of a potential exposure.
Given that submitting your data will in most cases require providing proof of your
Make himself appear lonely (Score:2)
GP can upload random numbers different from the random numbers his phone actually generated. Because no other phones got those numbers, it will appear as if GP hasn't been around any other people. It'll look like he's spent the last two weeks in his mom's basement.
The consequences of this will be - well, nil.
I suppose the bad thing one could do is falsely report that you tested positive. That would cause people who have been around you to get tested unnecessarily. If you're going to do that, it would po
Re: (Score:3)
Even if they did flood the system with random numbers they wouldn't match the random numbers anyone else's phone collected.
Re: (Score:2)
This is gonna be funny, once the first dozen people ruin the entire thing with false data. And probably hack the servers too.
This is going to be funny, if you do that and get a charge for multiple attempted murder. In the USA people have actually been arrested and put into jail for making false claims that they were infected.
PS. I would like to see you trying to hack the servers. There is absolutely no attack surface. The servers accept one specific kind of message from a phone. There is nothing to attack here.
An old strategy, by the way: (Score:2, Insightful)
If you plan to do something evil, propose something even more evil, then roll back to your planned evil, and everyone will either roll with it being "not evil" or be shamed into not being willing to compromise.
Just like those rebates at the supermarket that are just the regular price, but with a fake higher price that is crossed out next to it.
Re: (Score:2)
That's call the door in the face [simplicable.com] sales tactic and it's as old as dirt.
Re:An old strategy, by the way: (Score:5, Insightful)
People keep describing Apple and Google’s approach as an invasion of privacy, evil, or any number of other things, but I have yet to hear anyone actually break down the technical specs that have been released to explain why that’s so, rather than merely spreading FUD.
From what I’ve seen when reading through the specs, participation is opt-in, the service is decentralized with data staying on your device by default, it cannot be used by regional authorities to identify who has been exposed/notified, it cannot be used to identify who is infected, it does no form of location tracking whatsoever, it cannot be used to identify who you meet with or when, and the only form of data submission (which is also opt-in) is untied from your identity. To me, that sounds like exactly the sort of thing we’d want.
Re: An old strategy, by the way: (Score:1)
The Apple and Google approaches are the best thing thus far but by no means are they hacker proof.
Especially on Android any app co-installed can read out the bitstream from hardware and combined with other data on the phone, can be used to identify you at least. Once you have that data, information could be correlated and recorded about the people you meet and exchange data with.
At some point the data from the infected has to be shared, with sufficient data points and a minute amount of compromised devices
Re: (Score:2)
Especially on Android any app co-installed can read out the bitstream from hardware and combined with other data on the phone
This is obviously bollocks. Most phones with BLE also have a secure hardware enclave to protect this data and in any case random Android apps can't access the hardware directly, they can only go through the system APIs and must request permission to do so.
Android is based on Linux. On Linux random apps can't just steal your data by reading out the bitstream from /dev/sda, can they?
with sufficient data points and a minute amount of compromised devices in the loop
Nope. The data is just a per day identifier. When someone shares it they release 14 random IDs, one for each of the past 14 days
Re: (Score:1)
Especially on Android any app co-installed can read out the bitstream from hardware and combined with other data on the phone
This is obviously bollocks. Most phones with BLE also have a secure hardware enclave to protect this data and in any case random Android apps can't access the hardware directly, they can only go through the system APIs and must request permission to do so.
Android is based on Linux. On Linux random apps can't just steal your data by reading out the bitstream from /dev/sda, can they?
Also, why would the IDs ever need to be made accessible to any apps? The API to submit IDs should be "Please send whatever IDs you have to server id xyz", and the OS does the rest. The submission format is well-known, so any OS instance can submit to any server. The OS provider keeps a list of known server ids and only sends to those. Or maybe the apps are self-identified cryptographically and the OS provider knows where to submit IDs implicitly. Either way, the apps never see the IDs; nothing is expos
Re: (Score:1)
Do you know for certain that Apple and Google aren't going to release this particular bit of source code to be independently verified? No.
Even if they do, will that assure you that everything else in the OS that this code interacts with is behaving? No.
Re: (Score:2)
This is a delightfully illogical question to ask.
The technology we're talking about is going to be released as an API via OS updates for iOS and Android, and is being provided by the authors of iOS and Android. If they wanted to abuse your data, they'd have done so years ago in an OS update. If they just now decided to start abusing your data, they wouldn't add it to a short-lived API that will be removed within a few OS updates. If they aren't interesting in abusing your data at all, they certainly won't s
Re: (Score:2)
Especially on Android any app co-installed can read out the bitstream from hardware and combined with other data on the phone, can be used to identify you at least.
Well that is clever. If any app is installed on your phone, they don't need to read any bitstream to identify your phone, because the bloody app is installed on the phone!
Re: (Score:1)
But they can ALSO listen to the BLE stream. It's not enough to identify the phone, you're trying to identify the identifiers being broadcasted by your own device and others.
Re: (Score:1)
I doubt apps have access to the raw BLE stream from the hardware. Maybe they do, but that would surprise me. If not, then apps are only getting a curated stream of events, and the OS should filter out these particular ones, especially the outbound ones.
Re: (Score:3)
Especially on Android any app co-installed can read out the bitstream from hardware and combined with other data on the phone, can be used to identify you at least.
Supposing that first part is true (which I find doubtful, given the security surrounding Android’s equivalent to iOS’ Secure Enclave, which is where this app’s data is getting stored), this API doesn’t contribute anything to someone trying to discover your identity. All it does is generate, observe and record, and submit randomly generated numbers. That’s it. You can’t back out an identity from any of that, unless the “combined with other data on the phone” re
Re: (Score:3)
Especially on Android any app co-installed can read out the bitstream from hardware
No other app can see either the daily keys nor the broadcast RPIs. If you've found a way to do this, you have found a vulnerability worth $100K. I encourage you to submit it to Google and collect your reward.
At some point the data from the infected has to be shared, with sufficient data points and a minute amount of compromised devices in the loop, it should be relatively easy to build a database of movements of these identifiers
Possibly. But how would you collect the data points? You'd need to have your own BLE receivers located everywhere your targets of interest might go. And you'll need a way to distinguish between all of the targets in range; the broadcast IDs don't provide you any help, except for when one of the tar
Re: (Score:2)
COVID-19 is so expensive, if forcing you to join this is needed to save say a trillion dollars in the USA and let things get back to normal, I
Re: (Score:1)
People have a vested interest to spread that FUD.. all that geo data is valuable, and what's the government's track record on protecting the information?
Re: (Score:3)
There is no geolocation data being used in this system, as was already mentioned.
Ask IBM (Score:1)
They have helped the Germans before with great success.
Wie Geil!
Do you trust google? (Score:4, Interesting)
Re: (Score:2)
The only problem with that is that Google isn't the government. If you got rid of actual government, then a bunch of large corporations would step in and become the effective government, a form of feudal oligarchy.
Re: (Score:2)
Big corps that work in cahoots with governments are of course on the same list, it's called corporate fascism.
Anyway, how many people here are thinking about their old other phone they're going to use to game the system?
Re: (Score:3)
Ok, serious question now - why in the fuck would you want to "game" the system? This is an effort that is being done in order to reduce the amount of time we have to have businesses and schools closed by having an increased ability to trace possible infection contact, and inform people that they may need to get checked out before they start spreading this thing.
Literally the entire world is better off if this thing succeeds, and if people don't fuck around. So why do you want to extend the pain of heavy m
Re: (Score:1)
Maybe some don't believe a government has a right to "contact trace." Maybe there are better alternatives, such as if a person has detectable antibodies that means they've already beaten the virus and can't spread it.
But instead you want to a a sheep and roll over for whatever politicians decide.
Re: (Score:2)
The more people refuse to cooperate with this kind of thing the longer the lockdown will last.
Considering your phone is being tracked by the cell towers it connects to anyway it seems silly to forego this tracing system just because you are worried about being tracked.
"decentralized" (Score:1)
Re:"decentralized" (Score:4, Informative)
The reason it's decentralized is that the data is only stored on your own phone, not on a server.
Then, if someone gets diagnosed with coronavirus, they can give health officials access to their phone, and then health officials could e.g. send push notifications to all the other phones with the app that had blue-tooth pinged that one recently.
Re: (Score:2)
The data is stored on people's phones, until the owner uploads the data because they feel a bit sick. At that moment, the central server instantly knows what other BT devices that particular cellphone has been close to for the past x days. If enough people do that, it's easy enough to track exactly who did what where.
Not exactly. Here's how it works (Score:5, Informative)
The server knows nothing about which device has been close to which device. The only thing that the server knows is "somebody who tested positive randomly choosr the number 83683647499473037â. That's it. That's what the server has - a list of numbers that were randomly chosen daily by people who are reporting their infection. The server doesn't know their name or anything, just a randomly generated number.
Before I explain (again) how the rest of it works, consider this:
I've randomly chosen a 1000-digit prime number, which we'll call k. If you raise 11 to the power of k, the last 16 digits of the answer are 84056162056106629. Can you figure out what number I choose? The answer is no, you can't. All of the computing power on the planet can't reverse that math. On the other hand, if I TELL you my secret number, you can easily find the last 16 digits of 11^k. That's the concept of hash - it's a math problem that can be calculated in one direction but not the other. That's handy because you can check whether the number I eventually tell you is correct, but you can't figure out the answer until you're given an answer to check.
(Note this is a trivial example of what a hash does, not the best way to actually calculate a hash for real-world use. Real hashes use Merkle-Damgard.)
Understanding the concept of a hash as a one-way function, which therefore allows verification of a value, we can discuss the protocol.
Every day my phone picks a random number.
Every ten minutes it produces a hash of that random number along with the current time. It sends that hash to anyone who stays near me for many seconds. Note you can't reverse that hash to get my secret random daily number. (But if I TELL you my number you'd be able to check to see if you got a hash of my random daily number). Your phone saves the hashes from people you've been around for many seconds. This number is meaningless at that point, indistinguishable from random digits.
If I test positive, I tell the health department server random daily numbers for the last two weeks. That's all they get - numbers I picked at random. "Random" is important because every mathematically provable security definition includes the phrase "can't distinguish from random". Because my numbers ARE random, they are indistinguishable from random in every possible scenario, so this part has maximum security.
Each day your phone fetches from the health department server a list of the random numbers sent in as infected. Since you're not infected, you don't send anything to the health department, you just get a list of random numbers from them, each random number having been generated by someone who has the virus.
Now here is the magic. Your phone takes the list of numbers that came from infected people and runs the same hash algorithm that my phone ran. Which means it produces the same hash that I produced. If you were within 6-12 feet of me, your phone would have saved my 10-minute hash, so your phone checks the "infected" hashes against your list of "received from people near me hashes". If there is a match, you know that you were within about 10 feet of someone who later tested positive. Therefore you should probably get tested. You don't know WHO it was, just that you were near someone who was infected.
Again, the server knows NOTHING about who was near who, not even which random numbers were near which other random numbers. It knows only that "this random number was randomly picked by someone who tested positive".
It's a pretty clever scheme, providing probably the maximum amount of privacy theoretically possible while allowing you to determine if you've been around someone who has the virus. Nobody else can determine even that much.
Those random numbers would be provided to other usrs.
Re: (Score:1)
Fine. Do you keep your Bluetooth on?
Re: (Score:2)
Fine. Do you keep your Bluetooth on?
I do if it'll tell me I've been around someone infected :)
Who am I kidding, it's on almost all the time. I often use BT for casting sounds to other devices. BT power usage is not exorbitant.
And further, for security aspects, I consider...
At home the wifi is on. I realize that both BT and wifi are attack surfaces, but I think they're both trumped by cell usage. It's my phone and cell is always on and "tracking", if you will.
If I was a Bin Laden I would turn all that stuff off, and more, of course.
If I was a
Re: (Score:3)
Presumably, if you opt in to this programme, you do. As near as I can tell, it's still only opt-in at this point. But even if it weren't and Apple/Google forced it on, it a) wouldn't take much power and b) still be guarding your privacy pretty well.
I have Tile, so yes (Bluetooth Low Energy) (Score:3)
Once upon a time, I used to turn off my Bluetooth to save power and reduce the attack surface since I didn't use it. Now, I routinely use it in the car and I have Tile. Tile uses Bluetooth Low Energy to make it very difficult for me to ever lose my wallet, keys, dog, or phone. So now my BLE is on. The app proposed does need for it to be on - for the next month or so.
Re: (Score:3)
Yes, because the whole point of bluetooth is to be able to connect to devices around you and use them. And modern bluetooth hardware and drivers don't bleed your battery dry like it used to.
Get a phone that was designed in the last 5 years, and stop caring about bluetooth.
Re: (Score:3)
Fine. Do you keep your Bluetooth on?
Yes. If you're thinking that poses an inherent privacy risk, it doesn't. The Google/Apple system changes our Bluetooth MAC address randomly every 10 minutes along with the contact tracing code that is being broadcast.
It will drain your battery a bit but BLE uses very little power. You're unlikely to notice the difference.
Re: (Score:2)
It *is* clever, and you explain it well. Thanks.
Re: (Score:1)
Re: (Score:1)
Re: (Score:2)
Yes, that's right. Centrally, one could, at least theoretically, see that of the X number of people who sign up, Y% report they are positive. Perhaps more importantly, centrally they could see if this % infected is going up or down. Because the transmission means there is an exponential increase or decrease of cases, staying in the decrease zone is critical.
Re: (Score:1)
Re: (Score:3)
Unless the phones routinely rotate their signature tokens, which under the published scheme, they do.
Go ahead and wardrive. Within time period n all the information you've grabbed has rotated, so you only know what anonymized devices were in range at that particular point in time, and there's no way to relate one token to another.
Re: (Score:2)
It should be clear that there's a difference between storing all the data on a server vs each phone storing the data, hence the "decentralized" part.
Fear Not! (Score:4, Funny)
Just my 2 cents
you could do it just by cell towers. (Score:2)
and send sms's to the people who were in the same cells as confirmed cases. of course false positive is large with it, but it's not like it wouldn't work. of course you couldn't use it in nyc, italy or spain because you would just flag everyone with that.
I think the app-religion approach to this is dumb. and I'm an app developer. you can't fix everything with apps.
you could however have done very aggressive quarantining of potentially infected people in the early phase and tracking people who the infected
Re: (Score:2)
and send sms's to the people who were in the same cells as confirmed cases. Of course false positive is large with it, but it's not like it wouldn't work.
I believe you're being over-optimistic. The number of false positives would be way, way too high virtually anywhere but in rural areas. I live in suburban San Jose and highly doubt there's any cell you could go to which wouldn't register at least one infected person. Even using phone location data is too coarse: I believe that's in the 10 to 100 meter range and for useful contact tracing, you need meter-level accuracy.
you could however have done very aggressive quarantining of potentially infected people in the early phase and tracking people who the infected came into contact with. with just man power.
Yes, in theory we could have tried that, had we known there were infected asymptomatic peo
Scylla vs. charybdis (Score:3)
German here. What happened now does not constitute a precedent or something, like "in case of doubt, people will choose corporate data gathering and tracking over governmental data garthering and tracking". While, of course, most people who do not conscisously choose, already have been choosing coporate data gathering and tracking since the smartphone has become a thing, regardless how much they consciously would choose one over the other.
If someone has to consciously choose, or to publicly oppose one or the other, it still depends from case to case. But there may be a tendency here.
This time, even though the centralized, i.e. governmental solution would have been technically "better" for the purpose, and even though there's a widespread anti-coporate and indeed anti-American disposition in Germany, it seems it was opposed more strongly than the corporate solution, "giving our data to greedy American corporations", as some will see it here (as if German businesses would be any better than American ones in any way; they all work for the same purpose, which is profit).
Why?
Because our goverment, like so many governments around the world, even in our great beautiful western democracies, have a really bad track record of excessive data gathering and retention for survceillance purposes, and not out of a motivation that would widely be considered beneficial towards the populace. More than once the highest German court had to stop legislation that ordered to implement even more of what they already got, thus clearly identifying the government's surveillance ideas as exuberant and unconstitutional. Add to this the feeling of helplessness some feel under what they perceive as 'dictatorial' measures in countering Corona, and the outcome is what it is.
With a government the people could and would trust, this would have turned out differently, with a better solution. But in this world neither coporations nor governments can be trusted, and to decide between the two often means to decide between two evils.
Re: (Score:1)
Re:Scylla vs. charybdis (Score:4, Informative)
German here. What happened now does not constitute a precedent or something, like "in case of doubt, people will choose corporate data gathering and tracking over governmental data garthering and tracking".
That's not the case here. The corps get no tracking data whatsoever from this system. All data is only on your phone until you test positive, at which point you allow a set of random numbers to be uploaded to a government server, which distributes it to all of the other phones. No corporate-owned servers are involved, at all. The only privacy risk is from the government, and the design of the system ensures that they don't get any information about where you've been or who you've been in contact with.
Not Google / Apple but DP^3T! (Score:2)
NOT tracking infections (Score:2)
Comment removed (Score:3)
Re: (Score:2)
2) Willingness to use it. If you can not enforce it because step 1, not enough people will use it, making it useless
I'm with you on that. I think you missed #4. By making the database decentralized, you eliminated any possibility of a health service tracking down potentially infected people and getting them tested. IMHO, to make contact tracing work, it can't be voluntary and the contact data must be available to some health authority group.
Here's what might work: unmarked white vans go door to door, rounding up every human. Everyone gets a RFID tag implanted, "gives" a blood sample, and is has a contact tracer ankle bra
Sounds a lot like HAARP (Score:2)
Still not clear on the goal (Score:2)
Is it just me or are others unclear what the point of this is?
OK, some percent of the population has smart phones and downloads the app. It's not even close to 100%. If we get over 50%, I'll be pretty impressed. Then you have to depend on that 50% actually pushing the "I'm infectious" button. Then you have to depend on the other 95% looking at their results and voluntarily doing something about it, such as getting tested and/or voluntarily quarantining themselves (a real quarantine, not just sheltering in p