Mapping Poll - REQUEST

Andrew Sullivan ajs at
Tue Feb 9 16:34:14 CET 2010

On Mon, Feb 08, 2010 at 03:56:44AM -0500, Vint Cerf wrote:

> 1. Would the WG like to adopt the current "mapping document" as-is?

I thought we already had adopted it?  Is the question whether it
should be forwarded to the IESG with a request for publication?  

> 2. Would the WG like to engage in further discussion about this
> document, for example in the context of the Unicode TR46 that
> advocates substantially more mapping than the present "mappings"
> document?

No.  In my opinion, Unicode TR46 is harmful to the effort that caused
the chartering of this WG.  If we're going to take that approach, then
the entire approach of the new protocol is just a waste of effort.

> 3. Would the WG propose an alternative path towards dealing with the
> question of mapping and if so, what proposition(s) are offered by the
> WG members?

There are, at base, two ways to look at mapping:

    1.  Mappings need to be done in a context-sensitive way according
    to the local needs of the local linguistic community, based on
    data available in the local context.  

    2.  Mappings need to be done in a globally-consistent way, so that
    the same input always results in exactly the same domain name, in
    order that users cannot be surprised by the effects of what they
    type when, for instance, they change environments.

(1) is a "localization of IDN for IDNA" principle that was, as I
understood it from the beginning, the core of the IDNA2008 effort.
The WG's mapping document tries to make this clear, I think, although
it's fairly noncommittal in its recommendations.  Assuming we are
agreed that this is the right thing to do, I believe the only question
is whether we need to publish an informational document that says,
"You might need to map, or not; if you do, maybe do it this way, or

(2) is a return to the "internationalization" effort that led to
IDNA2003.  There are good arguments in favour of that position,
including that (1) amounts to a wide-open permission for all manner of
abuses and transformations that one would not wish.  Also, in the
absensce of a reliable way to understand the _user's_ "internal
locale" (as opposed to, for instance, the configured locale of the
device with which the user is interacting), there are almost certainly
opportunities for confusion and even possible fraud.

The main problem with (2) is that it amounts to a rejection of the
principle on which all the work was premised.  If we adopt an Official
Global IDNA2008 Mapping (and we haven't even opened the question of
whether it ought to be standards track), then we have abandoned one of
the very principles that led to the chartering of the WG in the first
place: that internationalization was too inflexible to accommodate
competing local needs, and that what we needed was something more like

All of this is to argue that I do not believe there is much middle
ground here.  This really is a matter of two irreconcilable principles
set up one against the other.  I think we need to pick one, and I
think the WG did as part of its foundation.  Therefore, I don't think
there is another path open to us.

Best regards,


Andrew Sullivan
ajs at
Shinkuro, Inc.

More information about the Idna-update mailing list