[Wikipedia-l] Categories considered harmful

[ruimu uestc]

From: "Delirium" <delirium at hackish.org>

> I do think ad-hoc categorization is going to end up mostly useless.
> Either it's going to end up hierarchical, but with a rather
> idiosyncratic and arbitrary hierarchy, or it's going to be very
> inclusive and overlapping, with most articles fitting under a large
> number of categories.  Nearly all medical articles can go under
> [[Category:Alternative medicine]] as well, for example; [[The Bible]]
> can go under about a million categories relating to Christian sects, or
> even non-Christian religions, sects, groups, cults, and all manner of
> other organizations that have something to say about the Bible; etc.

I don't know but some simple rules could help to avoid too many categories,
like avoiding grand-parent and cousinage, when possible (ie if [[The Bible]]
links to Religious Texts, it shouldn't link to Texts nor to Mythical Texts)

The problem of hierarchy and thesaurus (or ontology) is that structuring the
world is always "ruling" it in a way. If you place Religion under Culture
under Human, you affirm that God is invented by human beings, in a way.
That's why categorising is easily going againt NPOV. But any list or "see
also" links can share the same problems. This "structuring is ruling" (or
"describing is deciding") problem is only more precisely uncovered with
categories (compared to other projects or lists), thus the problem is easier
to adress with them.

Imho categories are also usefull for technical and usability reasons:

a) If I click on a "see also" link, I may get lost because I'm not shure to
find a backward link on next page, and I may forget (after few jumps) where
I was starting from. Nothing is more annoying that having found somewhere an
info and not being able to find it again (that why I often open linked page
in blank explorers). jumping from branches to branches is easier than
swimming in a marsh, even if the tree (not a real one indeed, more like a
nest) cannot be proved to have the best shape.

b) As already said, categories can be helpfull to watch a subject one like.
I never used "related changes" to [[List of China-related topics]], wich is
now split in two parts, because there are too many articles there. I would
watch category Chinese Thought, because it is a lot narrower and is my
prefered topic.

c) Categories tools can be developed to go slowly to something more or less
like those beautiful project boxes, but with the standardisation that will
ensure usability.

d) ...

(gbog)