Carcharoth wrote:
And I
shudder to think of the duplicated effort in checking references. It
would be great if you could look through an article and see that 5
people you trusted had ticked off most of the references as
"verified".
Hmm, in my experience the majority of finds of inaccuracy in
articles
come from correlation with another article, or at least from some
outside view raising a "that's odd" response. This discussion does raise
a suspicion that we still operate a somewhat naive generic fact-checking
approach: any page that is thoughtful about what we mean when we say
"checking facts"? Of course there is one aspect relating to the way a
cited reference may not support a fact as stated. But we do want
something a bit smarter than "make-work" solutions for a site with many
millions of references. I was discussing "over-checking" at the
Cambridge meetup, where you don't so much check a single fact as
surround it with other related facts, from other sources, and assess for
consistency, as a way of bearing down on unreferenced claims, and that
of course goes for things where you don't have the exact reference handy.
Charles