Statcounter's Windows market-share data is not accurate or reliable, and I can prove it

It's time for another round of Statcounter stories on Windows 'market share,' but none of the data makes any sense, and I have the receipts to prove it.