Gah. #1 was sensible. #2-3 are terrible.
People's own sites are
reliable sources for information about them. It's perfectly
reasonable to use a person or company's own site as the primary or
even sole source for a stub or relatively short article. Yes, when
they get to good and featured length they'll need more, but it's
perfectly possible to have an embryonic article that relies entirely
on the subject's own pages. The sole useful effect of #2 and #3 is to
make it possible to do incontestable deletions of articles that some
people have notability problems with. Absolutely not.
And lest anyone think I'm being hysterical here, have a look at
http://en.wikipedia.org/wiki/Carlos_Ferrero. (Found though random
pages. Took me about 20 to find a good example.) This is a prime
example of the sort of article our sourcing fever doesn't really
think through. It's a stub or barely above a stub. It has more or
less purely factual information, and has a link to the subject's
official site. Anyone who wants to know about Carlos Ferrero is well
served by this article - they get a general overview and a link to
his website. Less useful if you don't speak Spanish, but I'm guessing
we'd be hard pressed for comprehensive English-language sources on
him anyway. (We'd get a good number, but most of them would be
incomplete and writing an article out of them would involve a lot of
very messy stitching together.)
The article is firmly in the large class of articles that is good
enough to keep up but not good enough to call done. It should not be
speedied, prodded, or deleted through any other means. It should be
edited. If that takes a while, it takes a while, but that's OK
because the article is serving a useful purpose right now. (Heck, I
just learned something from it!)
-Phil
And it would take you under a min to find a second citation for
that..... so it is a poor example.
Yes, with any policy change, someone is going to be able to dig out the
odd example where it would not really help. If we don't act because of
marginal damage, then we will never change anything. When we have 1.5
million articles we need to think bigger than that - and consider net
impact on the project, not one or two cases.
Whatever we do, the status quo is not an option.