[openpgp] AS2+OpenPGP protocol extension review request

[Ben McGinnes <[e-mail]>]

Hi all,
	For those of you either not subscribed to gnupg-devel or who
may not have paid as much attention to it over the new year period;
I've been working on a little thing which is ready for some form of
peer review.

Essentially it's a design for extending the W3C's ActivityStream
version 2.0 (AS2) and ActivityPub (AP) protocols for federated social
networks (e.g. Mastodon and Pleroma) with OpenPGP in order to provide
a host of features not inherently built into AS2 and AP.

The AS2 and AP designers considered it, but realised that they didn't
have enough cryptographic knowledge to design it in a way that
wouldn't shoot someone in the foot; and so they did the responsible
thing in not making assumptions.  Instead leaving things open for that
void to be filled later.

I found their work early last year and, upon reading the specs,
realised that I was looking at a transport protocol.  Not only that,
but all the most essential cryptographic functions which needed
filling were already thoroughly addressed by another existing
protocol: OpenPGP.

My post on this thing to the gnupg-devel mailing list is here:

https://lists.gnupg.org/pipermail/gnupg-devel/2019-January/034167.html

The W3C AS2 and AP specifications are here:

https://www.w3.org/TR/activitystreams-core/
https://www.w3.org/TR/activitypub/

My extension proposal is the second draft and the first draft that has
been posted publicly (the actual first draft was sent to the W3C AS2
designers, a couple of GnuPG developers and a handful of others).  My
design document is available here:

https://files.de.adversary.org/crypto/ac/index.html

The supplemental files (including public and private keys used in the
examples) are here:

https://files.de.adversary.org/crypto/ac/supplemental.zip

Note: files.de.adversary.org is an AWS S3 bucket, so it will trigger
an SSL cert error.  Ignore it or drop back to HTTP at your own
preference.

Anyway, there are a number of people on this list in particular who I
think could make sure that I haven't catastrophically cocked the whole
thing up.  I don't think I have, but that's precisely when you should
double-check to be sure.

So if as many of you as can spare the time could please weigh in and
try to pick it apart; that'd be greatly appreciated.  It'll help make
it stronger.  Those of you who've already been down the protocol
design path are amongst those I'm particularly keen on checking this
thing.

I do realise there are still matters to discuss and finalise with
regards to transmission of keys; currently there are multiple options
included and I expect some discussion around that.  There's also two
versions of the Encrypted Note and I'm inclined to favour the second,
slightly more complex version for a number of reasons; including
working with other functions (like nesting a Signed Note inside it in
a similar manner to signed & encrypted PGP/MIME emails).


Regards,
Ben

P.S.  I'm cross-posting this to the IETF OpenPGP WG mailing list on
      the off chance that there might be subscribers to that who are
      not also subscribed here or to gnupg-devel.  That may not be
      very likely, but it is possible and so a copy there doesn't
      hurt.  Apologies to those who will see it twice as result,
      though.  Especially since you're the ones I'm most keen to want
      reviewing this work.