[tz] Proposal for new rules

[Zefram]

Stephen Colebourne wrote:
>I appeal for others to comment on the proposal, as it really is pretty
>darn simple and effective. #1 ensures that every region considered to
>be important enough to have an ISO-3166 code is given a full historic
>time-zone rule.

We used to have that rule.  I'm OK either way on this.

>                #2 ensures that no political faux pas occurs by
>linking across boundaries.

This actually sounds like a source of *more* political tension.  The whole
concept of inter-country links (which you're saying we should avoid)
depends on us associating each city with exactly one ISO 3166 region
that "contains" it.  There will on occasion be political disagreements
about which one that should be.  (Even apart from the ISO 3166 regions
that uncontroversially overlap.)  But if we do allow linking a city to
a region that it is not in, we can sidestep these arguments.  We're not
claiming that the city is in your country; we're just claiming that it's
the biggest city that uses the same timezone rules as you.  People will
still argue about which country a multiply-linked city "really" belongs
to, but the argument is irrelevant to us.

I think we could do with explicitly adopting an inclusive policy for
zone.tab entries.  Say: we include a city-country link iff there is
a significant population that self-identifies with that country and
uses the clock behaviour of that city.  This is a relatively objective
criterion that doesn't opine on the matters that are subject to big
political disagreements.

This approach applies regardless of whether we're attempting to provide
every country with a named zone that is exclusively its own (your rule
#1).  It also applies regardless of the inclusiveness of our approach
to pre-1970 data.  Of course, a more inclusive view would lead to using
fewer cross links.

In the above, I'm using the term "country" as a synonym for "ISO 3166
region", by which I mean a region to which ISO 3166 assigns a regular
alpha-2 code.  I think this is uncontroversial ground on this list,
but worth explicating in this context: the biggest political question
we're ducking is what constitutes a country.  Delegating that decision
to ISO 3166 is an excellent policy, and was expressed well by Jon Postel
in RFC 1591 "Domain Name System Structure and Delegation" (1994-03):

#     The IANA is not in the business of deciding what is and what is
#     not a country.
#
#     The selection of the ISO 3166 list as a basis for country code
#     top-level domain names was made with the knowledge that ISO has a
#     procedure for determining which entities should be and should not
#     be on that list.

This policy worked very well for the DNS until the Internet became big
enough to wag the dog.  I think it'll work for us for a while longer.

-zefram