This argument is interesting because it both a) makes the law worthless and b) removes the "scary slippery slope" argument.

After all, if you can provide a key to anything, then all you have to do (whether it's encrypted financial documents or random noise) is say, "Yep, it's encrypted, here's the key, it's the text of the Wikipedia page for 'kittens.'"

Prove that's not the correct key. If the onus for producing a key (whether one exists or not) is on the defendant, isn't the onus for proving the validity of the decrypted file on the prosecution?

All this comes back to cases like the one from CA (I think) where the guy who refused to decrypt the evidence that would prove his guilt.

And from another perspective, if you're Bernie Madoff and the evidence that will convict you is encrypted, won't you refuse to decrypt with a smile and take the 2-year punishment (with $500M in the bank) over life for financial fraud (and bankruptcy)?

This is why OTP encryption is unbreakable.

http://en.wikipedia.org/wiki/One_time_pad

Assuming the encryption method can create the bit sequence that is the random data. It very well might not. There may be gaps in the encrypted data's number space.

For any non-trivial encryption method, you'd be brute forcing your way through a bunch of them to find the key that can decrypt the random noise to that message. Typical "20 times longer than the existence of the universe" warnings apply. :)

Only if you use a key the size of your data, which is not commonly done.