Re: [Cfrg] Fwd: I-D Action: draft-yonezawa-pairing-friendly-curves-01.txt

[Peter Gutmann <[e-mail]>]

Björn Haase <[e-mail]> writes:

>We know that the cost of conventional attacks is low and many applications
>are actually "worth" the effort of an attack.

Another thing about PQC is that all of this is entirely new crypto that we
have no experience in using.  We've had decades of experience with using
PreQC, and have mostly managed to get it right (a lot of the attacks being
performed were known about years ago but were ignored until someone published
an attack paper with accompanying tools and newsworthy name, and even then
there's a huge amount of code in PreQC crypto designed specifically to prevent
entire classes of attacks), while we have zero experience with using PQC.
Which means we're going to see years if not decades of new attacks, or the
same old attacks that were fixed in PreQC implementations, popping up with
PQC.  It's quite possible that PQC will make us a lot *less* secure, if QC
never really happens but the expected vulnerabilities in using PQC do.

In fact I'll make this prediction now:

  Likelihood of successful attacks due to QC: Epsilon.
  Likelihood of successful attacks due to use of PQC over PreQC: 100.0%.

(the second figure should actually be much higher than 100%, because there'll
be many, many of them, not just one).

Peter.