Jay JG wrote
"Charles Matthews" wrote
And I think clicking Random Page 100 times supports the impression that things improve. One example is that the category system has come on
well,
over the past three months.
Well, some might consider that "thin anectodal evidence". ;-)
Not really. It's a repeatable experiment, and can produce confidence in its results as it is repeated. For example noting the number of uncategorised pages in 100 random ones gives an estimator of the proportion in the whole. Do it a few times, and the estimated proportion should settle down to the real figure. That does assume that Random Page is uniform over all main space pages (which might be wrong).
Charles