[cryptography] Unbreakable crypto?

Michael Kjörling michael at kjorling.se
Fri Mar 20 05:42:55 EDT 2015


On 19 Mar 2015 16:10 -0400, from kevinsisco61784 at gmail.com (Kevin):
> On 3/19/2015 4:00 PM, Ben Lincoln (F70C92E3 - Cryptography ML) wrote:
>> There's no demo version available, but based on the screenshots and the
>> information on the site, it looks like it's using a "one-time pad"
>> generated using a PRNG or other sequence-generator seeded with a key
>> generated by the application, and it's the key that's exchanged.
> 
> I assume it uses your hardware to generate the random value.

It doesn't matter how the PRNG works. If it's seeded by a key, and the
pad is regenerable given the key (which it sounds like given the
description), then it's not an OTP, and you get _at the very best_ 2^k
bits security (where k is the number of entropy bits in the key)
rather than a proper OTP's 2^n bits security (where n is the length of
the message, in bits).

The reason why a OTP is provably secure in theory (in the
confidentiality sense of secure) is that the key _is_ as long as the
message and completely random; thus you cannot tell whether a key you
just tried is valid without already knowing the plaintext message, in
which case there really is no point to the exercise to begin with.

Combining a seeded PRNG with a simple operation on the PRNG output and
the plaintext or ciphertext is how stream ciphers work.

Now, something like a decent KDF feeding a key into AES running in
counter mode to generate a ciphertext stream which is then used as a
key for encryption in a stream cipher-like construct is _probably
reasonably_ (_very heavy_ emphasis on "probably") secure, and not too
dissimilar from what is hypothesized above. But at that point, you
might just as well use the fairly well-proven AES directly; it will be
marginally faster given identical hardware and otherwise identical
software (since you avoid your extra algorithm, and AES is often
hardware-accellerated on modern CPUs) and it will be at least equally
secure (because the security of the ciphertext in the combined scheme
will be totally dependent on the security of the stronger algorithm
pass anyway -- see Kerckhoffs' principle -- and most people are
unlikely to come up with something that is _more_ secure than AES
against any type of attack, much less all types of attacks). And it
absolutely is _not_ a one-time pad.

-- 
Michael Kjörling • https://michael.kjorling.semichael at kjorling.se
OpenPGP B501AC6429EF4514 https://michael.kjorling.se/public-keys/pgp
                 “People who think they know everything really annoy
                 those of us who know we don’t.” (Bjarne Stroustrup)


More information about the cryptography mailing list