[Cryptography] Say 'unguessable' not random

Phillip Hallam-Baker phill at hallambaker.com
Thu Aug 25 10:19:54 EDT 2016


​The bridge problem illustrates some of the problems when we try to achieve
'randomness'. It is impossible to audit processes to see if they are truly
random. And even defining what 'randomness' is tends to lead to circular
definitions.

What we really need is cryptographic secrets that are unguessable.
Unguessability is both the necessary and sufficient criteria.

So for the bridge contest, why not get rid of the random numbers altogether
and instead employ a commitment scheme in which each of the players
introduce as much unguessability into the seed as they like? There would
have to be some form of trusted party that would take in the encrypted
seeds, decrypt them, prepare the hands and then release the seeds at the
end so the process can be audited. But the seeds themselves need not be
'random'.

Thinking in terms of unguessability also helps us when we look at the
reason why passwords fail. The problem with passwords is that the user is
asked to chose a secret that they can remember but an attacker is unlikely
to guess. Of course it is going to fail.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160825/4aa2e379/attachment.html>


More information about the cryptography mailing list