I think for its flaws, its still significantly useful that a FIPS algorithm
or crypto library certificate certifies that an implementation passes its
test vectors, startup tests etc. It gives some reasonable assurance that
the algorithm is implemented according to the spec, and typically some
thought and cryptographic review went into the spec, and it is specified at
least as opposed to "what ever this chunk of code does".
Some-random-stuff that a guy hacked up one evening - does it do what he
thinks it does? Is what he thinks it does secure? If he attempted to
implement to a spec and there is no certification, no test vector - is what
is implemented even the same algorithm? (The degree of stupid mistakes
anyone who does much implementation and crypto review comes across suggests
any attempt to check things is a GOOD THING).
For example in the past I came across some code based on one of my
size-hacked perl codes that had been hacked into a library, they used no
test vectors, while it could decrypt what it encrypted, IT wasnt compatible
with my code. So it was in fact broken - it clearly no longer implemented
the algorithm correctly. Was it secure? Who knows- quite probably not.
My 2c
Adam
On Wed, Jan 26, 2011 at 07:55:53AM +1000, James A. Donald wrote:
On 2011-01-26 12:10 AM, Jack Lloyd wrote:
The problem with this approach is the people doing the certifying do
not actually understand cryptography or security engineering in any
meaningful way, so have no real ability to make such judgements.
An attempt to provide certification by idiots is unlikely to have the
intended results. As government certification always winds up being
done by idiots, government certification is unlikely to have the
intended results.
_______________________________________________
cryptography mailing list
[email protected]
http://lists.randombit.net/mailman/listinfo/cryptography
_______________________________________________
cryptography mailing list
[email protected]
http://lists.randombit.net/mailman/listinfo/cryptography
