Lies, Damned Lies and Statistics

A number of tech news outlets have noted the expiry of the two-year period Microsoft gave themselves to stamp out spam.

Step forward Microsoft spokesman Ryan Hamlin, who claims that by stopping 95% of spam reaching inboxes on MSN, he considers it solved. For once, I'm not going to be semantically pedantic and ignore the numerous possible definitions and interpretations of the word "solve" in this context. Instead, I'm going after the maths.

The problem is this: how on earth do they know they're achieving a 95% success rate in catching spam? By definition, in order to calculate such a figure, they'd have to know how many spam messages their customers receive, in addition to the figure for how many they're blocking. Question: if they can count every spam message, why do they let 5% of them through? And if they can't count them, where are they getting the 95% figure from? Do they really expect that their customers report every instance of spam getting through to their inboxes?

But then again, aren't 54.2% of all statistics made up on the spot anyway? So surely they should be able to join in. Or is that 71.56%? I can't quite remember. Possibly both.

[ Comment on this article ]