[Secdispatch] Re: Y2K29 (was Re: DISPATCH/SECDISPATCH - Call for topics for Vienna)

Nico Williams <nico@cryptonector.com> Tue, 09 June 2026 17:40 UTC

Return-Path: <nico@cryptonector.com>
X-Original-To: secdispatch@mail2.ietf.org
Delivered-To: secdispatch@mail2.ietf.org
Received: from localhost (localhost [127.0.0.1]) by mail2.ietf.org (Postfix) with ESMTP id DB07CFE3151E; Tue, 9 Jun 2026 10:40:42 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=ietf.org; s=ietf1; t=1781026842; bh=euLff0Z8y3pGU8TLkaGsj4aloL5Wjia+AS44l+GM0Yg=; h=Date:From:To:Cc:Subject:References:In-Reply-To; b=EPzBkm9nxYQwTdxYZYwcV7aWEyn/FAO+s/JNczzQRmIUG459H197U7t0pYFfNGfA/ q4Gcl90+v7IOklm1QCNalE4JQfw9qYzbCOH8zgqAteDvpwR5Whdyl9cIZu/xytxeCa H/rmzeiv8Iy8v47eVSjTgR4df1YxNxx/YusQuGSY=
X-Virus-Scanned: amavisd-new at ietf.org
X-Spam-Flag: NO
X-Spam-Score: -2.097
X-Spam-Level:
X-Spam-Status: No, score=-2.097 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_NONE=-0.0001, RCVD_IN_MSPIKE_H2=0.001, RCVD_IN_VALIDITY_CERTIFIED_BLOCKED=0.001, RCVD_IN_VALIDITY_RPBL_BLOCKED=0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: mail2.ietf.org (amavisd-new); dkim=pass (2048-bit key) header.d=cryptonector.com
Received: from mail2.ietf.org ([166.84.6.31]) by localhost (mail2.ietf.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 8ZqmBfmo6nWT; Tue, 9 Jun 2026 10:40:42 -0700 (PDT)
Received: from black.elm.relay.mailchannels.net (black.elm.relay.mailchannels.net [23.83.212.19]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (P-256) server-digest SHA256) (No client certificate requested) by mail2.ietf.org (Postfix) with ESMTPS id 67A2BFE314BA; Tue, 9 Jun 2026 10:40:40 -0700 (PDT)
X-Sender-Id: dreamhost|x-authsender|nico@cryptonector.com
Received: from relay.mailchannels.net (localhost [127.0.0.1]) by relay.mailchannels.net (Postfix) with ESMTP id 5E41F462999; Tue, 09 Jun 2026 17:40:39 +0000 (UTC)
Received: from pdx1-sub0-mail-a229.dreamhost.com (100-106-41-195.trex-nlb.outbound.svc.cluster.local [100.106.41.195]) (Authenticated sender: dreamhost) by relay.mailchannels.net (Postfix) with ESMTPA id 63738462988; Tue, 09 Jun 2026 17:40:37 +0000 (UTC)
X-Sender-Id: dreamhost|x-authsender|nico@cryptonector.com
X-MC-Relay: Neutral
X-MailChannels-SenderId: dreamhost|x-authsender|nico@cryptonector.com
X-MailChannels-Auth-Id: dreamhost
X-Lonely-Juvenile: 102a8c755b8c43df_1781026839153_402321136
X-MC-Loop-Signature: 1781026839153:3919315116
X-MC-Ingress-Time: 1781026839153
Received: from pdx1-sub0-mail-a229.dreamhost.com (pop.dreamhost.com [64.90.62.162]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384) by 100.106.41.195 (trex/7.1.5); Tue, 09 Jun 2026 17:40:39 +0000
Received: from ubby (unknown [24.28.102.31]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange ECDHE (P-256) server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) (Authenticated sender: nico@cryptonector.com) by pdx1-sub0-mail-a229.dreamhost.com (Postfix) with ESMTPSA id 4gZbm44Vv5z2G; Tue, 9 Jun 2026 10:40:36 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cryptonector.com; s=dreamhost; t=1781026837; bh=LU/CsH+7segXMrdNffuOIPE/qfoaNCDXKGHYpIA/zdg=; h=Date:From:To:Cc:Subject:Content-Type:Content-Transfer-Encoding; b=r7t71UZt6mVqgtavEQX7HDb62G1Z8YQElytvKex8QHe8C+ROMoLvBLbXEBpTkq/jR Zlwe1QWoNhdeAm9jKXJpgDBhoAALW+G9pbABNrFW6bjsadjYc0rMYNaAh4ROwKPMh1 oDNkrLj4+aq9YjINaTraDzqhE2iBJH/mdAZLZNETXuiAuej0VA7B7ITFg3rFcltNhS lTJ+MsRtYBGiE6fPDj/EuvfJMrNcPEgBgNh5TBz1sgslw7W4pCKt3tUlSIUjdmdPUt Nvi2D7g8PCADdhXbOGmU7mtjJ0FNHMgEs+KfIcK0jb50g5VaX2OUm75xxjwcjEMzPK Coqi7Sq4zlxRA==
Date: Tue, 09 Jun 2026 12:40:34 -0500
From: Nico Williams <nico@cryptonector.com>
To: David Benjamin <davidben@chromium.org>
Message-ID: <aihQEgpa2oq13ndI@ubby>
References: <CADNypP_zNn_AQKCxodeStDPnJvVqNyXqzEZhUXGr8Otgr7xMYA@mail.gmail.com> <aiOZvQO9xA958azm@ubby> <aiT1rDQF9rT3oajf@ubby> <5AC11ABD-CF83-4F68-A53E-4304E4DF803A@proper.com> <CABcZeBPccub39EJqpBh7BGrVcg5X5sN5n3d=AsPEozKkVM_Ztw@mail.gmail.com> <CAF8qwaB7WpO49_dc9KX2TbQ__gzucXX1OdvpHYpdeFp+6yOi=Q@mail.gmail.com> <aicx928q9ynDcMeK@ubby> <CAF8qwaDVxiHazTtoO6x8jXz-nKPaxLucy=Fx9jaSO2DqKjXrpQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CAF8qwaDVxiHazTtoO6x8jXz-nKPaxLucy=Fx9jaSO2DqKjXrpQ@mail.gmail.com>
Message-ID-Hash: IFOSG23KRGVJ5B2KIAASWVUBIRW567LT
X-Message-ID-Hash: IFOSG23KRGVJ5B2KIAASWVUBIRW567LT
X-MailFrom: nico@cryptonector.com
X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; header-match-secdispatch.ietf.org-0; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header
CC: Eric Rescorla <ekr@rtfm.com>, Paul Hoffman <phoffman@proper.com>, dispatch@ietf.org, secdispatch <secdispatch@ietf.org>
X-Mailman-Version: 3.3.9rc6
Precedence: list
Subject: [Secdispatch] Re: Y2K29 (was Re: DISPATCH/SECDISPATCH - Call for topics for Vienna)
List-Id: Security Dispatch <secdispatch.ietf.org>
Archived-At: <https://mailarchive.ietf.org/arch/msg/secdispatch/_pj395bkyaxu0JcUryuM_EWV6fQ>
List-Archive: <https://mailarchive.ietf.org/arch/browse/secdispatch>
List-Help: <mailto:secdispatch-request@ietf.org?subject=help>
List-Owner: <mailto:secdispatch-owner@ietf.org>
List-Post: <mailto:secdispatch@ietf.org>
List-Subscribe: <mailto:secdispatch-join@ietf.org>
List-Unsubscribe: <mailto:secdispatch-leave@ietf.org>

On Mon, Jun 08, 2026 at 06:04:20PM -0400, David Benjamin wrote:
> On Mon, Jun 8, 2026 at 5:19 PM Nico Williams <nico@cryptonector.com> wrote:
> > What about trust anchors?
> >
> 
> When you say trust anchors, I assume you mean what the client trusts? To

Yes.

> get to the point that post-quantum authentication means anything at all,
> it's indeed necessary for clients to stop accepting classical trust
> anchors, so
> https://www.chromium.org/Home/chromium-security/post-quantum-auth-roadmap/
> talks about how to get there.

Right.  I want to be able to amortize the cost of MCT certificate
validation.  That requires caching, for one, and caching that can work
across processes, and also across security boundaries as long as all are
using the same PKI trust anchors / profiles, because that will maximize
the opportunities to amortize that cost.

> That said, if a browser were to only accept certificates chaining to a
> post-quantum trust anchor, that would have pretty severe compatibility
> impacts when talking to HTTPS servers that don't have such certificates! So

Or if a cloud/service were to only have EE certs chaining to PQ TAs.

> the document talks both about:
> - An HSTS-like mechanism to try to apply it only to sites that opt-in.
> This, however, has a laundry list of downsides discussed there. (See Stage
> 2.)
> - How to use our existing automation points to get PQ-*incapable* servers
> to a point where they're *at least* able to authenticate their classical
> key with a PQ anchor, so that the PQ-*capable* servers get downgrade
> protection. (See Stage 3.)

That, I think, is the easy part.

> Of course, throughout all that you have to also expect the ecosystem to
> contain PQ-incapable clients, so all these server states need to still be
> able to serve older clients. It is not practical[0] to expect every client
> to migrate to PQC at the same time as every other client. Broadly, any kind
> of PKI-level transition means the servers have to be able to configure
> multiple certificates, sending the right one to each client.

Certainly.

> Once we get there, these kinds of PKI transitions are much simpler, because
> you can still serve older non-PQC clients without impacting the PQC
> clients. It also means that just because one set of clients are targeting
> one timeline, doesn't mean every other client needs to match that timeline.

Ok, that makes me feel much better: no flag day, though perhaps we'll
have short-lived, fast rotating classical CA certs, both intermediates
and even roots.

> [0] Indeed the common refrain throughout the Internet has been that old
> stuff lasts longer than you think.

Much much longer than one would ever wish.

> > What about KSATs and GSATs?
> 
> You may need to expand those acronyms for me. I'm not familiar with them.

Kubernetes Service Account Tokens and Google Service Account Tokens.
These are JWTs (JSON Web Tokens) -- signed JWTs, typically with Ed25519.

These are short-lived enough, and the issuer signing keys too, that
perhaps the CRQC concerns are low.  But there's what I think, and what
the cloud operators think, so it's worth asking.

> > What do you think of specifying an IPC protocol not unlike the Google
> > metadata service?  Such a service would generally offload cacheable
> > things from in-app TLS and IAM codebases to the service:
> >
> >  - requesting PSKs (if we go that route)
> >  - validating certificates (especially MCT ones)
> >  - requesting tokens (think GSATs)
> >
> 
> Google metadata service? [...]

It's a Google cloud thing.  An IPC service that fronts a Google security
token (JWT) service (STS).

>                   [...]  Is MCT a typo for MTC (Merkle Tree Certificates)?

Yes, sorry, my typo.  Probably due to similarities to CT.

> Are you talking about how MTC relying parties pick up optional information
> about "landmarks" to validate the smaller certificates?
> 
> If so, this may stem from a misunderstanding. MTC does *not* require that
> relying parties have out-of-band data. Standalone MTC certificates are
> analogous to regular X.509 certificates, [...]

My impression is that a) chains will get large, b) validation will get
slow.  Thus my interest in more compression, and more caching /
commonalizing validation into an IPC service.

> Whether it makes sense to standardize that service, I don't know. To
> [...]

It's easy to know: if the chains get huge and validation slow, then it
does make sense to have such a[n IPC] service.

> [...]
> 
> Regarding generalizing between PSKs and MTC metadata, *that* I would
> caution against. PSK distribution has very, *very* different requirements
> from MTC metadata. MTC metadata is client-independent, public, verifiable
> data. You could get it from anywhere. A PSK is the polar opposite. Go too
> far down this road and you've probably re-designed Kerberos. ;-)

Yes, a PSK distribution protocol would very much be Kerberos-like, but
it wouldn't be Kerberos.  (No ASN.1 needed, not Kerberos cryptosystem
needed, many lessons can be incorporated, no client authentication need
be implied, etc.)  The point is symmetric key cryptography being largely
immune to CRQCs, Needham-Schröder is relevant again.  But a PSK
distribution system is only a part of what I listed.

Nico
--