With articles it is obvious. The subject matter that
will be provided IS what
is advertised. This is NOT the case with re-directs. They point to somewhere
arbitrary and there is no way to ensure that the redirect remains consistent
and fits the subject of the Wikidata item well.
I've seen Wikipedia articles change topics many times as well. This is
particularly the case with the very kind of Wikipedia articles that would
have site-links to redirects in Wikidata.
I will admit though that this is a real problem. We can not guarentee that
a redirect will continue to point to where we expect it to. Redirects do get
broken from time to time as well. I'm sure that a creative solution could
be thought up for this problem. The first step would be marking redirects
when they are used, which I don't think anyone who wants redirects has any
problem with.
Personally I doubt there is value in redirects.
Several others have pointed out the value in redirects. They allow for you
to interwiki link articles together that otherwise would be impossible to
link together. They help create this web of internationalised knowledge.
It helps link concepts and explainations together across language boundries.
I find them very Wikipedia centric.
Isn't the whole concept of site-links in general Wikipedia centric?
Given the examples given, there was no Wikidata in the
first place.
Harvesting redirects is an exceedingly bad idea that will pollute Wikidata
with many items we should not have.
How does allowing site-links to redirects "pollute wikidata with many items
we should not have"? This does not create new Wikidata items, it merely
allows us to efficiently site-link items that we already have. For compound
concepts (does anyone have a better term for these?) we would already have an
item for the compound concept and its individual constituent concepts anyways.
Thank you,
Derric Atzrott