Nice in theory. I still refuse to believe that it works well or anywhere that easy in reality. While this *might* work out for a couple of programs that are lovingly designed for both ways, with each way of operating well-thought through, the majority of freeware/opensource programs will be either/or, forcing the user in different usage patterns for every app.
The finger/stylus recognition left a lot to be desired in OS2007, and as soon as that doesn't work 110% reliable, all the rest becomes awkward and frankly pointless, as you'll just end up using the stylus alone so that you can avoid having to tap every input field a couple of times until the OS recognizes your "big greasy finger" and displays the appropriate keyboard.
Another example are scrollbars - with a dual approach, you have to waste a lot of space for finger-friendly ones, because the device can hardly guess in advance whether or not you're going to use the stylus...