[Cfrg] Maturity vs. Novelty consensus documentation requirement (was:Re: Bad and Rigid Curve (Rigid << NUMS))

[Stephen Farrell <[e-mail]>]

Speaking as an interested non-cryptographer... but with
no hats as well, so a bare headed non-cryptographer:-)

On 07/08/14 18:31, Dan Brown wrote:
> Rather, we should be placing
> much more emphasis on the maturity of cryptanalysis.

As I mentioned in Toronto, I think it'd be great if CFRG
documented its consensus on this point too. That is, I'd
like to pose a requirement that CFRG explicitly attempt
to reach consensus on the value or non-value of selecting
curves that have been known for some time and that that
be documented. (Not in 100's of pages of formulae mind
you, a paragraph or two of consensus text should be all
that's needed, if such a consensus exists.)

For new algorithms, I think being new is a very obvious
downside, but its less clear to me if that also applies
to some or all newish curves. And I think there will be
other IETF folk who could also do with some documentation
of that aspect too.

For example, if CFRG conclude that this is a non-issue,
and decided to generate some brand new curve(s), then
documenting the consensus that that is ok would really be
needed I think, since the concern about novelty would be
raised later on for sure.

Equally, if CFRG think a bit of aging in oak is valuable
then saying that would be good. Or, if there's no consensus
on this point (which I could imagine given how it might put
a finger on the scales) then some documentation of the
arguments would be useful.

Thanks,
S.