Protocol, not policy (was: Mississippi Hißes)

Andrew Sullivan ajs at
Mon Dec 14 15:54:16 CET 2009

On Mon, Dec 14, 2009 at 08:10:56AM -0500, John C Klensin wrote:
> That is the obvious consequence of the above... and much more
> nearly aligned with how trademark law actually works.

And anyway, it doesn't _matter_ how trademark law actually works, or
any other such thing.  We already decided, just by chartering the WG,
that the protocol was going to be permissive.  There's no technical
reason at all that these controversial characters not be included _in
the protocol_, even if by policy they're never used anywhere.  This is
exactly the debate we had, for instance, about cuneiform.  There was
an argument, "But nobody's ever going to use these," and we agreed
that that was not the test we were applying.  We also agreed that
actually, probably for every zone ever likely to be instantiated, the
characters ought to be denied by registry policy.  That's a completely
other matter, and we had best not conflate policy and protocol.

The _only_ reason we have for restricting these characters is that it
is harmful not to, because of the way they were previously handled.
That is, we have a protocol-interoperability problem, which
effectively makes a policy matter into a protocol matter because in
the past the line was not drawn so sharply.  That is the sole
justification we can come up with for why some character ought or
ought not to be permitted _by the protocol_.  (We actually have some
exceptions to this principle.  I thought they were wrong then, and I
still do, but in the interests of converging on a consensus I was
willing to support those decisions anyway.  They're still not


Andrew Sullivan
ajs at
Shinkuro, Inc.

More information about the Idna-update mailing list