So let's assume for a moment and this is true and there really were 8-10 million pre-installs of ubuntu on new hardware. If true, this should be detectable and show up in the form of ubuntu boxes running on the network and their users surfing the web. After all 8-10 million is a little under yearly mac sales. Sure maybe not all ubuntu installs will be on the network, but a significant percentage SHOULD be...
But it doesn't add up. Consider:
Apr 2012 Mac 6.5%, Linux 1%
Linux ranges from 0.7% to 2.9% Apple's OSX 6.3% to 14.04%
Certainly this kind of measurement is far less then perfect, but what strikes me is that if as Canonical asserts they are seeing these kind of install numbers for Ubuntu and if one were to add in Debian, RedHat and so on, yet still in the counting of web clients OSX is being seen in the wild several multiples more often than all Linux installs together, it would strongly suggest to me that the Linux numbers quoted are at best incorrect. We know how many new Macs are sold every quarter since Apple shares those numbers.
It's not the year of the Linux Desktop as much as we might all want it to be. Sadly Windows AND OSX are kicking Linux's butt when it comes down to number of users.