On 04/01/2008, Lars Aronsson <lars(a)aronsson.se> wrote:
Nice thinking, but this is very unlikely to happen.
Infoboxes and
similar templates are created spontaneously by people who are more
interested (and knowledgeable) in 19th century poets or vintage
cars than in data structures. Your kind of structuring would send
them to programming classes before they can start to document
vintage cars and that would kill off their enthusiasm.
I don't think it's that hopeless. What they care about is that the
infobox lets them put in all the useful information they need to. If
the infobox interface doesn't change - or if changes are fixed by bot
on the pages - then the details of the plumbing underneath can be
changed quite a bit to what's sensible and useful. So everyone can
work to their strengths.
(For a not-quite-analogy, I almost never create a stub on en:wp with a
stub type, just with {{stub}}. I'm confident the people who care about
stub-sorting will come by and fix it.)
One day, a template for vintage car infoboxes is
created, having
parameters for make, model, year and picture. Someone starts to
add number of gears and top speed to the articles. In a classic
computer science setting this would be an error and these
"undefined parameters" should be removed. But in Wikipedia it is
a useful extension and the template should be adopted to display
these new values where available.
Yes, the data format people need to be mindful of keeping it all wiki
on the surface. Ideally the persondata stuff is something inside the
outer template visible on the page, which can be hacked as needed.
(If there is a visible template. e.g. the en:wp opera project hate
infoboxes, but persondata would still be fantastically useful to have
on the pages. And the persondata template is invisible. w00t!)
So, what can a computer scientist do to assist this
messy process?
You can extract semi-structured parameter data from template calls
in the database dumps. You can compile statistics on which
parameter names are most commonly used in various templates (e.g.
"year" vs. "age", "name" vs. "title") and give
advice to how
parameters should best be named in new templates. For each
template you can compile statistics on which parameter names
(defined or not) and values are actually used and provide feedback
on the "Template talk:" page. You can work together with
WikiProjects on the proper use of templates and infoboxes.
I think there's lots of room here for people working to their
strengths - content or formatting.
- d.