News for hackers. Stuff that matters.
Re:Inside the Fortified, Nuke-Proof Bunker that's Now Hostin... by JoachimSchipper on Thursday December 02, 2010 @04:08AM (#1961564)
> 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?)

Re:Inside the Fortified, Nuke-Proof Bunker that's Now Hostin... by arethuza on Thursday December 02, 2010 @10:03AM (#1962295)
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?

Re:Inside the Fortified, Nuke-Proof Bunker that's Now Hostin... by JoachimSchipper on Friday December 03, 2010 @03:55AM (#1965225)
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.)

Re:Inside the Fortified, Nuke-Proof Bunker that's Now Hostin... by ergo98 on Thursday December 02, 2010 @09:11AM (#1962163)
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.
Re:Inside the Fortified, Nuke-Proof Bunker that's Now Hostin... by JoachimSchipper on Friday December 03, 2010 @03:53AM (#1965223)
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.