The author makes a very valid point, and it would be quite interesting to see what kinds of tactile UI designs might be achieved. But I think there's an important distinction to be made in how we build tools to solve physical problems and how we build tools to solve conceptual ones.

Apart from purely remediative technologies such as Braille, I can't think of any technology from any era of human history in which conceptual information has ever been conveyed via the tactile sense. There have never been tactile clocks, tactile books, or any kind of tactile language. When human minds attempt to import ideas from the outside word, they use the eyes and ears, not the hands.

There's certainly a real problem with the UIs presented in the MS video, but it's not that they're visually-oriented. It's that they're designed to appeal to the eyes themselves, and fail to encode information in a way that's optimally suited to the mind. The aesthetics of the UIs in that video are stunningly beautiful, but I have no idea from looking at them how I would use them as tools; each notification, dialog box, and prompt for input seems fine in isolation, but when I try to conceptually 'zoom out' and understand how each function integrates into a workflow that allows me to apply my capabilities toward fulfilling my needs, I'm completely at a loss.

There seems to be an unfortunate trend toward pure visual aesthetics in the software industry today - perhaps a cargo-cult attempt to emulate some of Apple's successes - and MS seems to be suffering from it almost as badly as the Ubuntu and Gnome folks.