With these devices being radically different, attempts to merge the two UIs will always end in failure. Play to the strengths of each and develop accordingly. A phone is not a workstation, and a tablet is not a desktop.
No, a tablet is not a desktop. But many tablets
are notebooks, and many desktops are practically stationary tablets. The hardware is here. Tablets that plug into keyboards or other docks, touch AIOs, touch-enabled notebooks...sure it's all half-assed, right now, but it's all pretty new, as far as mass market users go. But, they're here, and we need software to catch up and make good use of them. MS, meanwhile, needs to do some serious thinking, when their next CEO comes on board, because when they screw up, they still get left supporting that for many years, and force software vendors in the mean time to also support their mistakes for years. Clearly,
how they develop needs to change.
Programs for Real Computing Work(tm) need the information density that a mouse/trackball/tit-centric GUI can provide, plus keyboard controls. None of that is going away, for those users. Hell, you can still
easily buy PCs with serial and parallel ports.
But there's a lot of the interface
where none of that matters, and merging it by making the mobile/touch version as powerful as the older version, even if it means a few more clicks, would result in a product that would be, on the whole, superior, because new tablets, convertibles, touch-enabled notebooks, and touch-enabled AIOs, would be much easier to use that way, for much of their common use.
Pretty much all of Control Panel, that isn't in the MMC, could be well replaced by streamlined mobile-friendly interfaces, instead of keeping all of the older ones around. Doing so would then have the side benefit of making those "toys" more powerful, on the software side of things. If they focused on making that side of things better, effectively empowering it, we would be better off. Programs and features, Devices and Printers, Sound, Mouse, keyboard, game controllers, network options, personalization, power options, and so on, could easily be made with touch-friendly interfaces, without losing one little bit of functionality. Doing that, they could push upon us one big learning curve for version n+1, refined in n+2, and then we can move on, with Windows phones, tablets, notebooks, etc., all basically just being tweaked implementations of the core Windows, but still feature-complete.
IoW, if the new style UI were the only one for that set of options,
and it offered the same functionality,
and it played nice with the same window management (Windows 8 gets a false for all 3), it would be a good excuse to do things like redesign the whole of devices and printers, the network config dialogs, etc., based on a solid well-thought-out HIG. Much of those interfaces are OK, not because they're good, but that they've changed so little over the last 15 years that
we know them well. Empowering tablet users would be as good of an excuse as any to redo them, IMO.
Where the UIs are too different, like gestures, and hiding whatever at screen edges and corners, such features should be configurable, preferably by conditional rules (IE, just like having different AC and DC power options, you could have docked and undocked, or keyboard or keyboardless, GUI options). What Sinofsky and others around him fail to grasp is that what we need is to make these new (in the sense that regular people use them, now) form factors better, so that they become less toy-like, rather than dumb down the main OS and programs to match the competitors' toy device UIs. Thinking that people want only dumb appliances is not seeing the forest for the trees: such interfaces and devices are currently the only economically viable way to have such technology, so it's what we've got. As their RAM and CPU power improves (namely, RAM bandwidth and caches), the OS' UI is gradually is becoming the main limiting factor, behind the human interface itself.