The problem is that this biases the statistics due to the 14-day window: everyone who buys a new iPhone goes to the App Store in order to download some "staple apps", but after the unitial setup a lot of users might only go to the App Store every three weeks (for example), meaning that newly purchased devices will be over-represented in the statics, and they all have iOS 6.

Honestly, though: even correcting for this new-device bias, this still isn't what you care about... normal users don't sit around browsing the App Store or the Android Market... they open the app when they are interested in a specific download because of some external sale having occurred. Your market as a developer are "people with a device", not "people weird enough to try searching on their device".

Regardless, though, yeah: the reason Apple is showing this statistic is because it is the identical metric reported by Google, and they were making a big deal about these numbers during the WWDC keynote, even to the extent of pointing out that Google had changed the metric in this way (to Market visits) was cheating... comparing, though, the like metrics isn't that bad.

> The problem is that this biases the statistics due to the 14-day window: everyone who buys a new iPhone goes to the App Store in order to download some "staple apps"

How do you know that’s how Apple measures OS version share? Why would Apple limit itself to only measuring when users open the App Store?

Any iOS device that ever connects to the Internet checks in with Apple to be activated and registered. After that, iOS regularly checks for OS and app updates. If I own an iOS device that I never update and never visit the App Store on, that device will still check for updates and will transmit version data to Apple. Also, because every iOS device has a unique ID, more active users won‘t be counted double.