There's probably an easier marketing benefit to hard buttons that would change determined by context than just a flat screen where the haptics in a localized area would be able to move, but that's just something worth playing with over some use cases, and then in real life to see what happens. A case can and should be made to see what happens. Personally, I'd rather not have a d-pad as it can restrict what a developer might want to do. In the same wise, going with haptics+gestures might not always be optimal as well. As for what to come next, I'd like to see maemo take a chance with something not done before, hence the post about a folding-haptic/gesture screen design. Something where navigation and input take an approach that is based more on interfacing with the content rather than interfacing with the device to get to the content.