Re: [Cfrg] ECC mod 8^91+5

[Hanno Böck <[e-mail]>]

On Mon, 16 Oct 2017 15:08:16 +0000
Dan Brown <[e-mail]> wrote:

> The main point of this curve is to use it in a system of
> multiply-applied diverse crypto, where its security features (special
> CM curve, minimal room for trapdoor) could complement those of other
> crypto algorithms (including PQC and other ECC algorithms).  Using
> this variant of ECC as the sole (PK) crypto would be risky (due to
> lack of track-record/aegis/scrutiny/etc.).

People rarely do multi-crypto these days, and if they do they usually
have good reasons.
I don't see those reasons here.

As we all remember we had a very lengthy discussion about new ECC
curves not so long ago. New curves were standardized with what many
people consider desirable security properties.

As you indicate some people want to combine that with pqc. In this case
multi-crypto makes sense, because it can give you something that you
can't have in a single algorithm: well-tested crypto and postquantum
crypto.

But standardizing yet another new ecc method seems a rather obscure
threat scenario. You'd assume that you use multiple ECC methods and
there's a security flaw that affects all others, but not your new
curve.

If you feel you want multiple different standardized ECC methods you can
already use nist ecc + c25519 in combination. I'm not aware that anyone
is doing this.

In general I feel it's not desirable to have as many standardized
algorithms as possible, as it only complicates algorithm choices. A new
algorithm should always have good reasoning for its existence.

Therefore I'm against standardizing this unless further arguments are
brought up why it's advantageous compared to existing standardized ecc
schemes.


-- 
Hanno Böck
https://hboeck.de/

mail/jabber: [e-mail]
GPG: FE73757FA60E4E21B937579FA5880072BBB51E42