[OSM-talk] Rendering barangays for the Philippines

[Scott Atwood]

On Tue, Nov 25, 2008 at 5:41 PM, Erik Johansson <erjohan at gmail.com> wrote:

> On Tue, Nov 25, 2008 at 6:13 PM, Scott Atwood
> <scott.roy.atwood at gmail.com> wrote:
> > [..]  Ideally, there should be a single, consistent representation[...]in
> UK English
> [..]
> > Then, in the UI, the canonical value could be translated into an
> appropriate
> > local expression.
>
> Hi again!
>
> Since there is little consistency in English tags, I don't see why you
> first canonize a list of one-true-way tags then translate it to other
> languages.  There are many things that aren't mentioned in our tags,
> even more isn't mentioned on the wiki, how will you supply
> translations for these things?
>
> And why is it important for foreign tags to be
> 1. documented in the wiki
> 2. translated in the editor interface
> 3. having a good English translation
>
> But you it's ok to use shitty english tags.
>
>
> Consistency is the stigma of OSM, that's what's sweet.
>
> /Erik
> PS. sorry for misquoting you, but I feel that is what you say..
>

I'm coming at this as someone with a hobbyist interest in linguistics, and
as a software engineer with both personal and professional interest and
experience with internationalization and localization.
FWIW, I do feel that you have misunderstood me and unfairly misquoted me.
 What I said, and what I mean, is that there should be a single consistent
internal representation for any given abstract concept.  That makes things
much easier for editors, validators, renderers, and any other automated
software that needs to work with the data in OSM.

The fact that most tags today are in UK English is an accident of history.
 If you ever look at the details of an HTTP transaction, or an SMTP mail
transaction,  you will similarly see that the internal representations in
these systems are in US English for similar historical reasons.  But it
isn't necessary to speak English to use a web browser or an email client,
because these details are abstracted away from the end users.

At least for common, standardized, and/or well-established tags, the OSM
editing clients should abstract away the underlying representation (which
happens to be UK English, but could just as easily have been French or
Swedish, or even an abstract number), and allow the user to add tags in his
or her preferred language.

Since the tagging system is completely open-ended, things get a little
fuzzier for new or ad-hoc tags.  For what it is worth, I'm OK with a little
fuzz around the edges.

There is an inherent struggle between freedom and constraint here.  Part of
the power of OSM is that people are free to invent new tags on the fly to
meet new needs and situations as they arise.  But for OSM to be universally
useful and to avoid balkanizing into mutually unintelligible sub-maps we
need to strive towards a common representation.  Over time, new tags that
"lost" to an alternative representations should be re-normalized to the
standard representation, something that could be aided by appropriate
assistance from editors and validators.

-Scott


-- 
Scott Atwood

Cycle tracks will abound in Utopia.  ~H.G. Wells
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20081125/612603b7/attachment.html>