Hacker News Reader

> 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.

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

>>> 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.

While you are right, note that the typewriters and certificates were just off-the-cuff examples to show that you cannot trust a lot of things any more, certainly not the ¨end-to-end encryption¨ promises but also many other things.

I am not advocating for typewriters as such, but there was the story about India ordering thousands after Snowden's revelations, and I just wanted to say why that is silly. If I want to find out something really badly, I will, so there is no reason to believe that people who have more resources cannot do the same.

> 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.

> 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.

Ya, kind of like the genius idea of getting a custom license plate with alternating L's and ones: l1l111l1

Might be hard to read or remember but it sure does stand out.

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.

> Do you trust your bios?

Yes, it's Coreboot with disabled and neutralized Intel ME.

> Your usb stack?

No, it's isolated in a dedicated, hardware-virtualized VM on Qubes OS.

> Your network card firmware?

Same as above. And it's FLOSS.

> surrounded by cameras and microphones

For those, I have hardware kill switches. On the phone, too.

> surrounded by cameras and microphones

I'm using https://github.com/qubesos/qubes-video-companion

Not even running a CPU with open source microcode implementing an open source ISA? Pfff, casual! /s

Breadboard processor and custom ISA would be wise here.

> disabled and neutralized Intel ME

Pinky swear.

https://puri.sm/posts/deep-dive-into-intel-me-disablement/

Your devices are not only one around you.

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.

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

The upside of this is that it just created a business opportunity for Faraday paint :)

Being the one guy in the neighborhood using faraday paint does not sound suspicious at all. ;)

Not at all unusual, especially for modern buildings. Thermic film on windows are a huge problem for relevant frequencies, wifi or broadband.

If it's indoors, would anyone outside even know?

Yes if they're trying to monitor you.

If your threat model is people who you already know are trying to monitor you, isn't it implied that they think you're suspicious with or without the paint?

I disagree. It is still quite trivial to reasonably exchange information without anyone being able to eavesdrop on it.

Authenticity isn't important if the message content can be verified. It can even pose a danger. People like to attack the messenger, especially if he has a point.

Don't know about your devices, but mine do not put me under surveillance. Smartphones are critical, but even those are controllable. If you are in public you won't know, but otherwise you can be reasonably safe.

I agree. And moving to e2ee for everything and away from big services like gmail and office 365 is part of the solution.

> Do you trust your bios? Your usb stack? Your network card firmware?

Sure, because I have the complete source code for all three of those things:

https://www.raptorcs.com/TALOSII/

But do you inspect all the source code?

"You need to be able to increase the costs of getting caught" -- Edward Snowden

Deliberately inserted malware, laying in plain sight in source code, is an existential risk for a company with a reputation like IBM's. Of course, there isn't a 100% guarantee of it being noticed, but a 1% chance of that happening is enough for them to tell the spooks to fuck off.

Special Access Programs only work because God-mode coprocessors like the Intel ME ensure zero risk of blowback for the manufacturer. Force them to have skin in the game and they aren't pushovers anymore.

> 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.

At some arbitrary point you just decide that the risk is small enough for you to be comfortable with it.

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.

> 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?

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/.

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".

> 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?

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.

There are quantum-resistant encryption schemes.

See: https://en.wikipedia.org/wiki/Post-quantum_cryptography

You’ll probably be dead by then anyway.

Some people guard secrets that they think need to stay hidden for more than a lifetime.

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