4 ms·

Cryptography is very dead. Cryptography relies on the keys being kept private. The keys must be on computers to be usable. Computers can be compromised, therefo

Cryptography is very dead. Cryptography relies on the keys being kept private. The keys must be on computers to be usable. Computers can be compromised, therefore the keys can be compromised, therefore they encrypted payloads can be decrypted. Any question? And by "Computers can be compromised", this means computers also cannot be trusted to hold any sensitive data.

valarauca1 13y ago
By this logic nothing is secret and nothing is safe, ever. Horrible example. Crypto, like every other secret keeping method comes down to trust. Who, and what do you trust to keep your secret.
- chris_mahan 13y ago
  You are correct: nothing is secret, nothing is safe. The governments can, at will, access any computer in the world, any file, and document. The only thing that stops them is the law, and as Snowden has shown, that's an oft ignored constraint. So, one must assume that given enough interest, the governments will be able to compromise any system, and any data on these systems.
drdaeman 13y ago
One can make a microcontroller (not much space for government trojan horses here, huh) handle crypto. Or build a computer on FPGA. It won't have much processing power, but it will still be sufficiently fast to handle strong crypto on text messages. Or, if they're really really paranoid, there is always an option to use old trusty TTL components.
- chris_mahan 13y ago
  And who is this microcontroller maker and how do I ensure that they have not been "cooperating" with anyone interested in breaking the encryption? Or do I have to build my own chips? And hopefully I didn't make an implementation mistake.
TacticalCoder 13y ago
I don't totally disagree but... What about encrypting your message on a fully offline computer, like what Schneier is now doing if I'm not mistaken? What if both him and his recipient do the same? Then you need physical access (and unnoticed physical access I'd add) to the offline computer to be able to decrypt right? And physical access doesn't scale.
- pyre 13y ago
  Devil's Advocate: What if they swap Scheier's USB key (or drive) with one that will exploit the USB stack on his offline computer (disabling 'auto-mount' won't help you here)? Yes, it's a physical access attack, but physical access to his person might be easier to achieve (while remaining unnoticed) than access to the offline computer.
  - chris_mahan 13y ago
    And he's probably high enough in the list of NSA "persons of interest" to warrant a field agent's time.
  - MichaelGG 13y ago
    I think the idea is to use simple physical media, like CDROM, where hiding a malicious payload is far more difficult since there's no embedded electronics on a disc. If that's too cumbersome, one could always implement a simple teletype-like system with a low bitrate. Like an automatic machine that can physically press keys on a keyboard. Then remove every key except letters, numbers, shift and tab. You could use a similar system or OCR to get data off the gapped machine.
    - pyre 13y ago
      One could, but IIRC Schneier has mentioned using a USB drive (though it could all be a clever ruse to throw off the NSA ;-).
- chris_mahan 13y ago
  How would you decrypt the message? The only safe option would be to print out the encoded message, and retype it into the offline computer. I don't think anyone does that. Also, the offline computer can be physically compromised.
EthanHeilman 13y ago
Its not cryptography which is dead, but the notion of a keeping a key secret forever. Key compromise will happen with some regularity, but there are three things you can do about it: 1. make key compromise harder, for instance require that an attacker compromise both your hardware and software to get your key, 2. limit the damage a key compromise can cause, for instance expire keys more often, randomly reissue keys, 3. and increase the speed with which you can recover from a key compromise. For example, the web PKI is completely compromised if a single CA key is compromised, but chrome can detect valid but forged google certs using key pinning and thereby alert the world. You can only use the stolen CA key to attack Google users once. The security community is advancing on all three fronts.
seagreen 13y ago
You don't have to keep your private key on a general purpose computer. See Cardano -- http://trilema.com/2013/snsa-first-product-the-cardano/ http://trilema.com/2013/snsa-first-product-the-cardano/
- chris_mahan 13y ago
  Cardano, you think, cannot be coerced by Romanian security services to backdoor their products? Also, check out the link at the bottom to polimedia.com, the parent company. I don't think I would entrust my private keys to this outfit, sorry.
AnimalMuppet 13y ago
This is not new at all. Clear back in "Applied Cryptography", Schneier said that there is no security without physical security. If someone can physically get at your box, it's all over. For most people, he said, that level of paranoia is overkill, but for some it is very real. And then there's what happened to Target. Someone got to their machines, not physically, but with a trusted download. That paranoia now has to be carefully thought through by everyone who cares about their security...

valarauca1 13y ago

By this logic nothing is secret and nothing is safe, ever. Horrible example. Crypto, like every other secret keeping method comes down to trust. Who, and what do you trust to keep your secret.

chris_mahan 13y ago
You are correct: nothing is secret, nothing is safe. The governments can, at will, access any computer in the world, any file, and document. The only thing that stops them is the law, and as Snowden has shown, that's an oft ignored constraint. So, one must assume that given enough interest, the governments will be able to compromise any system, and any data on these systems.

drdaeman 13y ago

One can make a microcontroller (not much space for government trojan horses here, huh) handle crypto. Or build a computer on FPGA. It won't have much processing power, but it will still be sufficiently fast to handle strong crypto on text messages. Or, if they're really really paranoid, there is always an option to use old trusty TTL components.

chris_mahan 13y ago
And who is this microcontroller maker and how do I ensure that they have not been "cooperating" with anyone interested in breaking the encryption? Or do I have to build my own chips? And hopefully I didn't make an implementation mistake.

TacticalCoder 13y ago

I don't totally disagree but... What about encrypting your message on a fully offline computer, like what Schneier is now doing if I'm not mistaken? What if both him and his recipient do the same? Then you need physical access (and unnoticed physical access I'd add) to the offline computer to be able to decrypt right? And physical access doesn't scale.

pyre 13y ago
Devil's Advocate: What if they swap Scheier's USB key (or drive) with one that will exploit the USB stack on his offline computer (disabling 'auto-mount' won't help you here)? Yes, it's a physical access attack, but physical access to his person might be easier to achieve (while remaining unnoticed) than access to the offline computer.
- chris_mahan 13y ago
  And he's probably high enough in the list of NSA "persons of interest" to warrant a field agent's time.
- MichaelGG 13y ago
  I think the idea is to use simple physical media, like CDROM, where hiding a malicious payload is far more difficult since there's no embedded electronics on a disc. If that's too cumbersome, one could always implement a simple teletype-like system with a low bitrate. Like an automatic machine that can physically press keys on a keyboard. Then remove every key except letters, numbers, shift and tab. You could use a similar system or OCR to get data off the gapped machine.
  - pyre 13y ago
    One could, but IIRC Schneier has mentioned using a USB drive (though it could all be a clever ruse to throw off the NSA ;-).
chris_mahan 13y ago
How would you decrypt the message? The only safe option would be to print out the encoded message, and retype it into the offline computer. I don't think anyone does that. Also, the offline computer can be physically compromised.

EthanHeilman 13y ago

Its not cryptography which is dead, but the notion of a keeping a key secret forever. Key compromise will happen with some regularity, but there are three things you can do about it: 1. make key compromise harder, for instance require that an attacker compromise both your hardware and software to get your key, 2. limit the damage a key compromise can cause, for instance expire keys more often, randomly reissue keys, 3. and increase the speed with which you can recover from a key compromise. For example, the web PKI is completely compromised if a single CA key is compromised, but chrome can detect valid but forged google certs using key pinning and thereby alert the world. You can only use the stolen CA key to attack Google users once. The security community is advancing on all three fronts.

seagreen 13y ago

You don't have to keep your private key on a general purpose computer. See Cardano -- http://trilema.com/2013/snsa-first-product-the-cardano/ http://trilema.com/2013/snsa-first-product-the-cardano/

chris_mahan 13y ago
Cardano, you think, cannot be coerced by Romanian security services to backdoor their products? Also, check out the link at the bottom to polimedia.com, the parent company. I don't think I would entrust my private keys to this outfit, sorry.

AnimalMuppet 13y ago

This is not new at all. Clear back in "Applied Cryptography", Schneier said that there is no security without physical security. If someone can physically get at your box, it's all over. For most people, he said, that level of paranoia is overkill, but for some it is very real. And then there's what happened to Target. Someone got to their machines, not physically, but with a trusted download. That paranoia now has to be carefully thought through by everyone who cares about their security...