Re: [openpgp] Modelling an abuse-resistant OpenPGP keyserver

[Daniel Kahn Gillmor <[e-mail]>]

On Tue 2019-04-16 16:45:45 -0400, Daniel Kahn Gillmor wrote:
>    I think the next draft will use "certificate discovery" to refer only
>    to this third case, and will rename lookup-by-user-id to "certificate
>    lookup".  Does that make sense?

I've just released version 03, which makes this change, and also adds
more nuance related to the ecosystem around keystores and how that might
have to change in the face of abuse-resistant keystores.  Identifying
and naming this distinct interface helped to surface another attack,
fingerprint flooding, which is also now documented.

The new version is here:

   https://tools.ietf.org/html/draft-dkg-openpgp-abuse-resistant-keystore-03

From the document's changelog:

substantive changes between -02 and -03:

 * new sections:
   * Keystore Interfaces
   * Keystore Client Best Practices
   * Certificate Generation and Management Best Practices
 * rename "certificate discovery" to "certificate lookup"
 * redefine "certificate discovery" to refer to lookup by signing (sub)key
 * new attack: fingerprint flooding
 * new retrieval-time mitigations -- tighter filters on discovery and update
 * recommend in-band certificates where possible to avoid discovery and lookup
 * new privacy considerations:
   * distinct keystore interfaces
   * certificate update
   * certificate discovery
   * certificate validation
 * more nuance about unhashed subpacket filtering
 
I really appreciate the feedback i've gotten in the course of this
writeup, and welcome more.

In particular, if someone has any pointers to how to think about e-mail
address canonicalization within an OpenPGP User ID (which is a UTF-8
string), especially in the context of IDN and non-ASCII local-parts
(does this mean RFC2047-decoding or encoding?) i'd love to see some
pointers (or even better, proposed text).

The full text of the markdown source for -03 is attached below.

    --dkg