2
$\begingroup$

There's no doubt that third party cryptanalysis is an essential part of the crypto ecosystem. It's how we know that there's some degree of practical "goodness" to a new algorithm or protocol.

Unlike a number of other information products, a robust third party testing, analysis and validation industry doesn't seem to have evolved for crypto. For example, NIST's CAVP and CMVP are very specific and focused in scope while the Common Criteria PP's are focused on modules and implementations rather than the constituent algorithms.

Are there trusted third parties - labs, academic institutions, government agencies, to whom new algorithms and concepts can be submitted for cryptanalysis? If so, who are they and how can one contact them?

Many thanks,

Adam

$\endgroup$
3
  • 3
    $\begingroup$ The best way, IMHO, would be to publish the suggested algorithm. That is, at least if the algorithm is interesting in one way or another. If it's not interesting, it's unlikely someone will do the cryptanalysis for free (but in this case, you might want to consider using an existing algorithm). $\endgroup$
    – Aleph
    Apr 4, 2016 at 15:24
  • $\begingroup$ Alternatively, you could select a public cryptanalyst (someone with published attacks on cryptosystems similar to yours), and pay them to analyze your cipher. Now, if they don't find anything, it wouldn't serve as proof to the rest of the community (your cryptanalyst might have missed something); it would serve as evidence that there's no trivial weaknesses... $\endgroup$
    – poncho
    Apr 4, 2016 at 15:57
  • $\begingroup$ Thank you for your time and your responses. I think I may have been unclear. What I'm looking for is an agency or organization that is in the business of cryptanalyzing algorithms. The organization would have to be impartial, trusted by the community and have clear credentials in the space such that its report would carry weight. Does such an entity exist? $\endgroup$
    – afirestone
    Apr 4, 2016 at 20:19

1 Answer 1

5
$\begingroup$

Does such an entity exist?

No, not really. There isn't any organizations who's in the business of doing public cryptanalysis, and there certainly aren't any organizations that are sufficiently trusted for the cryptographical community to say "we know algorithm X is secure - organization Y said so".

Let's go through the likely suspects:

  • NSA (and other TLAs) - they have the internal resources to do a decent cryptanalysis, and while it's not guaranteed to catch everything, they'll catch most weaknesses. However, they are certainly not in the business of soliciting contributions, they are most certainly not interested in publishing their results, and even if they did, no one would trust that they're revealing all that they know.

  • NIST (and other standards bodies) - you might think otherwise, but they typically don't have the resources to do an exhaustive analysis themselves (NIST, at least, does have the resources to analyze results found by other people, but they don't pretend to have the wherewithal to do it entirely in house).

  • IACR - actually, the IACR is not in the business of doing cryptography; instead, they're there to promote cryptographical research in others (and they do a very fine job at that IMHO). If you want to organize a cryptographical conference, the IACR might be of some help; if you want some cryptanalysis done, well, the best the IACR can do is point you at some relevant researchers.

  • Private organizations - I believe that there a few companies that will accept submissions, and publish analysis on them (for pay); however at present, most of them consist of a single researcher who will do analysis for hire. Not horrid; however they don't have the "clear credentials" that everyone will accept as weighty (assuming, of course, that they don't find anything).

You appear to be wondering "there's CAVP and CMVP, which people accept; why isn't there the analogue for cryptanalysis?". Well, CAVP and CMVP is mostly a list of checkboxes; the reviewer has a list of requirements (things like: encrypt this AES block with this key; did you get that result); meet all those requirements, and you'll get the certification.

Cryptanalysis can't be put into such a checklist - instead, the analyst needs to figure out if there are any clever ways to break the system - it's hard to put cleverness as a checkbox item.

And, ultimately that's why there aren't any 'trusted organization'; even if someone where to say 'looks secure to me', there's always the possibility that someone else would find something. The best we have is have a lot of independent people look at it; it's possible that everyone missed something, but it's less likely.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.