On 5/1/06, Matt Brown morven@gmail.com wrote:
My personal opinion is that /currently/ I think nofollow has merit; I think we're extremely bad at handling linkspamming and website promotion in Wikipedia, and thus reducing the benefit might help.
However, if we became better at keeping track of what Wikipedia links to and getting rid of spam, I'd support turning nofollow off.
What would be useful steps to take to keep better track of these? Is there anything that can be done in software? Or procedure/process? External applications?
I've ductaped curl to spamassassin and can now rate the spammyness of websites. :) Now I just need to replace all the email header centric SA rules with web centric ones.
Because there are ~2,800,000 external links from the main namespace alone on enwiki (there is now a patch to mediawiki to set nofollow per namespace, I think no matter what we should nofollow outside the main ns), I see automated scanning as the only reasonable step right now to improve our link quality.
So I'm working on it..