Cormac:
What's the margin of error here? Is it significant?
Cormac, I have no exact figures in terms of alphas and standard deviations. The error greatly depends on the day of the month and the size of a wikipedia.
The >average< forecast will become more accurate very fast, as the month progresses, because 200 wikipedias contribute to the figure. But each individual forecast may be wildly inaccurate early in the month.
TV coverage may double the activity in one wikipedia later in the month. Global media attention (e.g. for Wikimania) will make all estimates too low. Small wikipedias will have large differences in activity per month anyway.
I see it as a best guess only so that one can compare figures for the current month with earlier months. Without this forecast people would intuitively use the 31/13 method for Aug 13th, which will be much farther from the truth for columns C and D.
Cheers, Erik