3 ms·

Even better, they should encrypt with a method that is reasonably, large-scale crackable around the window of their planned release. That would bring a lot of a

by ergo98 16y ago

Even better, they should encrypt with a method that is reasonably, large-scale crackable around the window of their planned release. That would bring a lot of attention to cryptography (and the vulnerability of the same) as every media organization rents out massive AWS GPU clusters to try to crack it. That would be pretty entertaining. And if something happened to them...well it's going to be discovered anyways.

JoachimSchipper 16y ago
> the vulnerability of [cryptography] Well, no. Competently done crypto is uncrackable for all practical purposes without side-channel attacks. Or, maybe, quantum computers (RSA; but why would you use RSA for this?)
- ergo98 16y ago
  Well, yes, of course. Competently done. Competently done crypotgraphy is rare, and most common cryptography is not competently done. Which is exactly the point of my post.
  - JoachimSchipper 16y ago
    Most crypto implementations are vulnerable, yes, but the percentage of vulnerable deployed crypto implementations as a fraction of all deployed implementations looks much better. Note that I'm talking crypto: the OpenSSL people seem quite able to make all sorts of mistakes, but it's almost never the crypto itself that is at fault. (Today's news: a workaround for Netscape's stupidities apparently allows attackers to downgrade to a worse protocol suite in certain cases. Stupid bug, but not crypto per se.) So yes, there are errors, but publishing something AES-encrypted and giving away, say, half the key doesn't tell us anything about those.
- arethuza 16y ago
  Are there any forms of encryption which take a predictable and precisely defined amount of computing power to decrypt? e.g. Some asymmetric so that I could encrypt something easily but then require someone who wants to decrypt the message to perform 100 Trillion operations to decrypt it?
  - JoachimSchipper 16y ago
    It's almost always probabilistic - someone guessing the right key the first time could break it very quickly. It may be possible to reduce variance in cracking times by requiring multiple cracks (use a secret sharing scheme to break up the "real" key in chunks; encrypt each chunk in a crackable way.)

by ergo98 16y ago

JoachimSchipper 16y ago

> the vulnerability of [cryptography] Well, no. Competently done crypto is uncrackable for all practical purposes without side-channel attacks. Or, maybe, quantum computers (RSA; but why would you use RSA for this?)

ergo98 16y ago
Well, yes, of course. Competently done. Competently done crypotgraphy is rare, and most common cryptography is not competently done. Which is exactly the point of my post.
- JoachimSchipper 16y ago
  Most crypto implementations are vulnerable, yes, but the percentage of vulnerable deployed crypto implementations as a fraction of all deployed implementations looks much better. Note that I'm talking crypto: the OpenSSL people seem quite able to make all sorts of mistakes, but it's almost never the crypto itself that is at fault. (Today's news: a workaround for Netscape's stupidities apparently allows attackers to downgrade to a worse protocol suite in certain cases. Stupid bug, but not crypto per se.) So yes, there are errors, but publishing something AES-encrypted and giving away, say, half the key doesn't tell us anything about those.
arethuza 16y ago
Are there any forms of encryption which take a predictable and precisely defined amount of computing power to decrypt? e.g. Some asymmetric so that I could encrypt something easily but then require someone who wants to decrypt the message to perform 100 Trillion operations to decrypt it?
- JoachimSchipper 16y ago
  It's almost always probabilistic - someone guessing the right key the first time could break it very quickly. It may be possible to reduce variance in cracking times by requiring multiple cracks (use a secret sharing scheme to break up the "real" key in chunks; encrypt each chunk in a crackable way.)