Re: upstream source code authenticity checking

[Russ Allbery <rra () stanford edu>]

Alan Coopersmith <alan.coopersmith () oracle com> writes:
> On 05/ 2/13 11:10 AM, Russ Allbery wrote:

> > I routinely do this.  It's called a key-signing party.  The only trust
> > that I am expressing with that signature is that I have seen and
> > verified, to the best of my ability, some form of reliable
> > identification for that person (ideally a passport I can verify, or a
> > social environment in which it would be very difficult to impersonate
> > someone you are not) in combination with a proof that the key I signed
> > belongs to the person whose identification I checked.

> Though for many open source projects, having a passport or other
> government id is not the sort of identity we care about - knowing that
> you're the person who does git/hg commits under that e-mail address is
> what we care about - if it's a pseudonym that doesn't match your
> passport, that doesn't affect whether we accept code from you or not.
> (The lawyers might care, when it comes to verifying who owns copyright
> and agreed to release code under a given license, but that's a whole
> separate mess to unravel.)

Right.  And that's part of the problem with using existing PGP key
signatures.  They don't convey the piece of information that the project
probably actually cares about.  Open source projects rarely care that I'm
*actually* Russ Allbery, rather than just using that name on-line while
actually legally being named Roger McDowell.  Lawyers may care if they
want to be able to sue me, but that's an edge case.  What projects
actually care about is that I'm the same person, by whatever name, who has
an established track record in multiple other projects and an established
trust basis in the broader community.

You can kind of get there by tying together multiple different pieces of
data, but it's certainly not directly conveyed by PGP key signatures.

-- 
Russ Allbery (rra () stanford edu)             <http://www.eyrie.org/~eagle/>