Hi,
On Thu, Jan 04, 2018 at 11:03:02AM -1000, Christian Huitema wrote:
> match, and be stuck. So the resolution system needs to answer "there are
> multiple entities who claim to be example.com, here is a list", and the
> client would then have to check which one has the right key.
I know you know this, but just for observers who maybe don't: the
"client" there may not be a human, and indeed the client that is
presenting the request might not be any longer in touch with the
originating node for the query. The usual example of this is SMTP,
because it is a hop-by-hop protocol and so an SMTP server attempting
to deliver a mail might not be in touch with a human who started the
mail going. But we can easily imagine additional cases --
particularly with the current mania for hooking up new devices
programmed by people who have never worked in an Internet context
before ("Internet of Broken Things").
Now, in such cases, the clients need to know how to handle the
disambiguation, which means that the indirection layer can actually
only ever be useful for human-computer interaction. That's maybe a
nice constraint on development, but it causes trouble when the human
who configured the system walks away and then the naming universe
changes, creating an ambiguity (that now needs resolution) when before
there was no such problem.
The FCFS solution (cf. namecoin) is one answer, but of course that's
really the answer that the DNS used in its early period. As soon as
people with money and lawyers got involved, the arbitrary identifiers
of the DNS became trademarks of WIPO and ICANN's UDRP. FCFS just
doesn't work in that context, and attempting to build it in as a
technical limitation does not make for a realisticly useful system.
I am interested in these problems, but I also remain completely
confounded about how to make the indirection layer work without
creating a new class of design problem that we don't know how to fix.
Best regards,
A
--
Andrew Sullivan
ajs@xxxxxxxxxxxxxxxxxx