On Thu, May 1, 2008 at 3:35 PM, Charles Matthews charles.r.matthews@ntlworld.com wrote:
Oh, very.
We need a different question, really. If MediaWiki were deliberately designed to have some namespaces that were for search engines to find, and others not, how should that be set up?
Something like the article space, Wikipedia: space and some Development: space might do. All discussion namespaces are inherently dodgy. But one more namespace that was non-article, non-policy might do; and one more namespace on the other side, so that AfDs didn't have to be in the Wikipedia: space - now this is sounding more reasonable to me.
I think the closest we're going to get to a good technical solution would be to have some magic words __INDEX__ and __NOINDEX__ that will force indexing on or off for a document. The default could be set per-namespace, and I'm assuming we'd want articlespace to be the only one indexed by default. Not unlike how tables of contents are handled: <4 sections and there is no TOC, >=4 and there is before the first section, but __TOC__ and __NOTOC__ can be used to override this.
And whatever the magic words are, there should be a way to specify in a template "index what transcludes this, but *don't index this template*" for things like {{policy}}. "<includeonly>__INDEX__</includeonly>" would take care of that nicely. (This assumes the keywords actually have effect when transcluded, which IMO they should.) Same for {{essay}} and the like. That way much of the userspace content that people will want to search for will be in Google. People would be advised to avoid using __INDEX__ on their userpage to uphold [[WP:NOTMYSPACE]], or on userfied articles so substandard articles don't show up in Google, etc.
My $0.02, kind of rambling, but hopefully there's some salvageable ideas in there...