5 ms·

Don't believe vendors' lies about "end-to-end" encryption. If caught red handed, they will always say it depends on how you define where both "ends" begin. Do

by jll29 4y ago

Don't believe vendors' lies about "end-to-end" encryption. If caught red handed, they will always say it depends on how you define where both "ends" begin. Do not trust a cloud service that you have not developed and deployed yourself. You may trust untrusted hardware with your encrypted content, but only if you have given it your content pre-encrypted by yourself, not trusted a third party to encrypt it on your behalf. Obviously, this excludes mobile devices. Do not trust a tree of certificates if you cannot trust the root certificate because it belongs to an organization that is in a jurisdiction where people may be interested in what you have written and said in your encrypted message. Don't trust old-school typewriters and the postal system either. Letters are routinely opened and typewriters can be matched. For example, the Stasi (secret police of the former GDR - German "Democratic" Republic) had an archive of type samples of all sold models of typewriters for re-identification of political pamphlets. You can trust a few things: You can trust your Linux box with your self-compiled kernel (no 3rd party drivers), at least as long as it is not on a network. To build a safe environment, you could start there, taking a defensive approach. Remember, last time the paranoid turned out to be naive when Snowden revealed the real status quo in 2013 (ten years ago, when I couldn't buy a 1 TB USB stick).

voxic11 4y ago
> You may trust untrusted hardware with your encrypted content Couldn't someone still capture them from the untrusted hardware, wait until quantum computer technology is available, then decrypt them?
- beisner 4y ago
  There are quantum-resistant encryption schemes. See: https://en.wikipedia.org/wiki/Post-quantum_cryptography https://en.wikipedia.org/wiki/Post-quantum_cryptography
- nicoburns 4y ago
  You’ll probably be dead by then anyway.
  - denton-scratch 4y ago
    Some people guard secrets that they think need to stay hidden for more than a lifetime.
- greiskul 4y ago
  Quantum computing is theoretically able to attack many of the public key encryption algorithms we use. Hardware encryption of HDs doesn't use those algorithms, it uses symmetric-key algorithms, that are not vulnerable to quantum attacks.
tablespoon 4y ago
> Don't trust old-school typewriters and the postal system either. Letters are routinely opened and typewriters can be matched. For example, the Stasi (secret police of the former GDR - German "Democratic" Republic) had an archive of type samples of all sold models of typewriters for re-identification of political pamphlets. That's theoretical. I highly doubt anyone extends that much effort to target typewriters anymore. The best they could probably do is match a series of messages to the same typewriter. Though they might not even be able to do that, because the law-enforcement skills to match typewriter documents to each other have also probably nearly completely atrophied. You're probably more likely to be caught by being the weirdo still buying typewriter ribbons.
- bobthepanda 4y ago
  I mean, photocopiers have the exact same technology so the US Secret Service can see who’s trying to photocopy dollar bills. (Which is what they were originally created for, not as bodyguards) https://www.eff.org/press/archives/2005/10/16 https://www.eff.org/press/archives/2005/10/16
  - tablespoon 4y ago
    >>> Don't trust old-school typewriters > I mean, photocopiers have the exact same technology They absolutely do not have the "exact same technology. That's totally, 100% wrong.
- jwalton 4y ago
  > That's theoretical. I would agree, but then I also would have agreed with this sentiment about much of what Snowden revealed. If you want to be paranoid, may as well go all in.
  - tablespoon 4y ago
    > I would agree, but then I also would have agreed with this sentiment about much of what Snowden revealed. If you want to be paranoid, may as well go all in. What Snowden revealed is way more believable than the government continuing to invest tons of effort into profiling typewriters. The whole thing with Snowden was that his leaks confirmed a lot of things reasonable people speculated.
gizmo 4y ago
Do you trust your bios? Your usb stack? Your network card firmware? I get the desire to control the entire chain, but nowadays you’re surrounded by cameras and microphones 24/7, every smart device is a surveillance sleeper agent waiting for the magic word. Certificate chains are the least of your worries.
- unyttigfjelltol 4y ago
  Anything your monitor displays or sound system outputs is subject to ubiquitous surveillance. And, when there's a gap in the surveillance that's also is subject to interpretation. Privacy already is (mostly) lost and the battle now simply is to reliably establish authenticity that can't be impersonated or doubted. That's where encryption remains helpful.
  - gizmo 4y ago
    I agree. And moving to e2ee for everything and away from big services like gmail and office 365 is part of the solution.
  - CapstanRoller 4y ago
    You don't even have to emit signals. Consumer WiFi equipment can already be used as active radar[0] to track individual movement, and ubiquitous deployment of dense high-frequency cellular networking equipment (5G, 6G) will basically render most house walls transparent (especially if you're living in a USAmerican stick-framed drywall house) potentially allowing for things like face tracking at a distance. [0] https://www.youtube.com/watch?v=HgDdaMy8KNE https://www.youtube.com/watch?v=HgDdaMy8KNE
    - anonym29 4y ago
      The upside of this is that it just created a business opportunity for Faraday paint :)
      - geysersam 4y ago
TremendousJudge 4y ago
> You can trust your Linux box with your self-compiled kernel Can you? If we're going this far, you can't trust the compiler either.
- seba_dos1 4y ago
  At some arbitrary point you just decide that the risk is small enough for you to be comfortable with it.
  - Aeolun 4y ago
    And that comes when you self compile your kernel? Seems like an odd point to stop. Like, the chances of someone sneaking something terrible into an open source kernel seem lower than of any of the hardware coming out of China to be poisoned.
    - staunton 4y ago
      > Like, the chances of someone sneaking something terrible into an open source kernel seem lower than of any of the hardware coming out of China to be poisoned. what causes you to think that? For example, how often has there been something terrible inside an OS kernel? How much poisoned Chinese hardware are you aware of?
anonym29 4y ago
This is a very efficient post - the TL;DR is accurately summed up with the first three words!
seanw444 4y ago
I agree with your point of taking security measures into your own hands, but given how pompompurin was only caught through using a personal email address with his name on it, despite using Windows (!), I think you're a lot safer than you think. Most of the time at least. Problem is that it's impossible to know if now is "most of the time".
jupp0r 4y ago
You need to think more about your model of trust. You will quickly discover that all security is built on trusting someone, even in the examples you mention above. You can compile your own kernel, but you trust kernel developers to not have put exploits in the source code. You can't audit the source code itself. You also trust your compiler to accurately compile the source code. You also trust compiler developers to not have put exploits into the compiler itself. You also trust the compiler they used to bootstrap their toolchain to compile the compiler. Fundamentally, all security is based on trusting somebody, understanding their incentives and intentions and weighing risk and reward. An excellent book on the subject is Bruce Schneier's "Liars and Outliers" https://www.schneier.com/books/liars-and-outliers/ https://www.schneier.com/books/liars-and-outliers/.

by jll29 4y ago

voxic11 4y ago

> You may trust untrusted hardware with your encrypted content Couldn't someone still capture them from the untrusted hardware, wait until quantum computer technology is available, then decrypt them?

beisner 4y ago
There are quantum-resistant encryption schemes. See: https://en.wikipedia.org/wiki/Post-quantum_cryptography https://en.wikipedia.org/wiki/Post-quantum_cryptography
nicoburns 4y ago
You’ll probably be dead by then anyway.
- denton-scratch 4y ago
  Some people guard secrets that they think need to stay hidden for more than a lifetime.
greiskul 4y ago
Quantum computing is theoretically able to attack many of the public key encryption algorithms we use. Hardware encryption of HDs doesn't use those algorithms, it uses symmetric-key algorithms, that are not vulnerable to quantum attacks.

tablespoon 4y ago

> Don't trust old-school typewriters and the postal system either. Letters are routinely opened and typewriters can be matched. For example, the Stasi (secret police of the former GDR - German "Democratic" Republic) had an archive of type samples of all sold models of typewriters for re-identification of political pamphlets. That's theoretical. I highly doubt anyone extends that much effort to target typewriters anymore. The best they could probably do is match a series of messages to the same typewriter. Though they might not even be able to do that, because the law-enforcement skills to match typewriter documents to each other have also probably nearly completely atrophied. You're probably more likely to be caught by being the weirdo still buying typewriter ribbons.

bobthepanda 4y ago
I mean, photocopiers have the exact same technology so the US Secret Service can see who’s trying to photocopy dollar bills. (Which is what they were originally created for, not as bodyguards) https://www.eff.org/press/archives/2005/10/16 https://www.eff.org/press/archives/2005/10/16
- tablespoon 4y ago
  >>> Don't trust old-school typewriters > I mean, photocopiers have the exact same technology They absolutely do not have the "exact same technology. That's totally, 100% wrong.
jwalton 4y ago
> That's theoretical. I would agree, but then I also would have agreed with this sentiment about much of what Snowden revealed. If you want to be paranoid, may as well go all in.
- tablespoon 4y ago
  > I would agree, but then I also would have agreed with this sentiment about much of what Snowden revealed. If you want to be paranoid, may as well go all in. What Snowden revealed is way more believable than the government continuing to invest tons of effort into profiling typewriters. The whole thing with Snowden was that his leaks confirmed a lot of things reasonable people speculated.

gizmo 4y ago

Do you trust your bios? Your usb stack? Your network card firmware? I get the desire to control the entire chain, but nowadays you’re surrounded by cameras and microphones 24/7, every smart device is a surveillance sleeper agent waiting for the magic word. Certificate chains are the least of your worries.

unyttigfjelltol 4y ago
Anything your monitor displays or sound system outputs is subject to ubiquitous surveillance. And, when there's a gap in the surveillance that's also is subject to interpretation. Privacy already is (mostly) lost and the battle now simply is to reliably establish authenticity that can't be impersonated or doubted. That's where encryption remains helpful.
- gizmo 4y ago
  I agree. And moving to e2ee for everything and away from big services like gmail and office 365 is part of the solution.
- CapstanRoller 4y ago
  You don't even have to emit signals. Consumer WiFi equipment can already be used as active radar[0] to track individual movement, and ubiquitous deployment of dense high-frequency cellular networking equipment (5G, 6G) will basically render most house walls transparent (especially if you're living in a USAmerican stick-framed drywall house) potentially allowing for things like face tracking at a distance. [0] https://www.youtube.com/watch?v=HgDdaMy8KNE https://www.youtube.com/watch?v=HgDdaMy8KNE
  - anonym29 4y ago
    The upside of this is that it just created a business opportunity for Faraday paint :)
    - geysersam 4y ago

TremendousJudge 4y ago

> You can trust your Linux box with your self-compiled kernel Can you? If we're going this far, you can't trust the compiler either.

seba_dos1 4y ago
At some arbitrary point you just decide that the risk is small enough for you to be comfortable with it.
- Aeolun 4y ago
  And that comes when you self compile your kernel? Seems like an odd point to stop. Like, the chances of someone sneaking something terrible into an open source kernel seem lower than of any of the hardware coming out of China to be poisoned.
  - staunton 4y ago
    > Like, the chances of someone sneaking something terrible into an open source kernel seem lower than of any of the hardware coming out of China to be poisoned. what causes you to think that? For example, how often has there been something terrible inside an OS kernel? How much poisoned Chinese hardware are you aware of?

anonym29 4y ago

This is a very efficient post - the TL;DR is accurately summed up with the first three words!

seanw444 4y ago

I agree with your point of taking security measures into your own hands, but given how pompompurin was only caught through using a personal email address with his name on it, despite using Windows (!), I think you're a lot safer than you think. Most of the time at least. Problem is that it's impossible to know if now is "most of the time".

jupp0r 4y ago

You need to think more about your model of trust. You will quickly discover that all security is built on trusting someone, even in the examples you mention above. You can compile your own kernel, but you trust kernel developers to not have put exploits in the source code. You can't audit the source code itself. You also trust your compiler to accurately compile the source code. You also trust compiler developers to not have put exploits into the compiler itself. You also trust the compiler they used to bootstrap their toolchain to compile the compiler. Fundamentally, all security is based on trusting somebody, understanding their incentives and intentions and weighing risk and reward. An excellent book on the subject is Bruce Schneier's "Liars and Outliers" https://www.schneier.com/books/liars-and-outliers/ https://www.schneier.com/books/liars-and-outliers/.