Re: [therightkey] Draft charter for a Transparency Working Group

[Ben Laurie <[e-mail]>]

On 11 December 2013 17:35, Daniel Kahn Gillmor <[e-mail]> wrote:
> I'm interested.  I think this has strong potential for improved
> authenticity on the 'net (and improved confidentiality follows from that).
>
> However, I'm also concerned that cryptographically-verifiable global
> logs create an enumeration concern for the space that they log.  This is
> similar in some ways to the issues raised around DNSSEC's NSEC (and not
> particularly effectively addressed by NSEC3).  Enumerability like this
> is potentially a major table of metadata that could potentially be abused.

Agreed, at least for some protocols. I think this area has already
been well thrashed out for HTTPS certificates, the initial target,
which are _mostly_ public. It seems we have good mechanisms in place
for those who want public CAs to issue private certificates, too (that
is, name-constrained intermediates).

> I'd appreciate it if any Transparency Working Group explicitly tries to
> address concerns around enumerability.

This is a _hard problem_, and verifiable logs suffer from the same
limitations that NSEC3 does, at least as currently envisaged.
Mechanisms akin to name constraints can perhaps go some way towards
addressing this issue in general, but the general problem I have no
idea how to solve. But it would seem to be on-topic for the general
case.

Possibly we have to admit we have no such mechanism for namespaces we
want to be non-enumerable.