I haven't done any iPhone development, so I may be wrong here, but doesn't his version not do what the Google one does? How does his code prove that Google's feature could be done with documented API calls when his version uses touch instead of proximity?
My version uses both touch and proximity. The point was to prove that you can unexpected things even within the limits of the data you have available.
I don't really know if you could remove the touch component and use only proximity without using a private API, but neither does John Gruber. Some really smart people work at Google, and it wouldn't surprise me at all if they had figured out a way.
