In the interest of full disclosure I'll admit to some personal and painful experience with User Interfaces (UI). They probably colored my perspective to this day.
The first UI we learn as humans is language which allows us to interface with mom and other humans. In the US in particular, most people learn the "language UI" only once - English. As an immigrant from Italy in 1970 (before English had became the world's global language) I suffered the cost of confronting my UI (Italian) abruptly obsoleted and I had to waste a whole semester studying English as second language before I could start full speed with my real courses (not an inconsiderable penalty considering that I still completed a BA and MBA in 4 years).
Looking back, then I chose to almost completely abandon using Italian in favor of English (the former I would not forget while the latter is still a work in progress). I distinctly remember that bargain: Pay a price to get something valuable for the rest of my life. The alternatives were to go back to Italy or to limit my future to an Italian language neighborhood: both were easier, neither had ROI.
The cost of UI changes
I think all UI changes reflect the same ROI calculation I did back then. The fact that, despite all the push for Vista and Win7, XP still has a 70+ % market share, in my view, reflects that same calculus: Why invest effort and incur the cost of disruption (personal or organizational or both) to change to a UI that does not have a compelling, demonstrable unquestionable advantage? 70% so far say NO.
Furthermore, to follow Microsoft into the promised land we'd have to trash perfectly viable PCs that just happen to have an engine insufficient for the new OS. It is like GM offering you a car that carries no more people, goes no faster, saves no money and requires you to learn to drive with a joystick sitting backwards and looking at the road through a mirror - eventually you'll like it.
Arrogance as a strategy
Todays' announcement that the miraculously uncluttered, minimalist new-design Internet Explorer 9 (IE9) will not run on XP is the reflection of an arrogance we would not tolerate from automakers. It's tantamount to GM announcing cars wider than the lanes in our roads, for awesome benefits to be sure, like improved cornering stability, or all passengers sitting in first raw, and replying to our objections with "build new roads" (a "let them eat cake" attitude comes to mind).
In the days when most (low power) cell phones can sport an effective web browser, one has to wonder the necessity of dumping viable PCs just because they cannot run Win7 and IE9 - has anyone heard we are in a recession? Yet the me-too desire for IE9 will prompt many to "eat cake": buy new expensive hardware with a new OS and IE9 to do no more than an iPhone or Android phone would do. Arrogance may pay off after all.
Arrogance however always carries a price to be paid when the American public abruptly switches to alternate suppliers more responsive to their needs: see the experiences of GM and Toyota, Internet Explorer and Firefox/Chrome, Motorola and HTC/Nokia/et al., United/USAir and Sothwest/JetBlue, and so on.
One would be tempted to add to the above list Microsoft and Linux/Ubuntu, but we cannot. It may be a sea change in the making with Android (an Ubuntu skin), but is not yet here. Why after 10 years of Linux has it not happened? It's not the economy, stupid; it's the UI. Yes, the UI.
Linux in all its incarnations can run better than XP on old PCs (as on new cell phones) and can deliver equivalent or the exact same applications (through WINE) but pigheadedly keeps coming out requiring the user to adapt to a new UI. If there were a benefit to that UI change users would do it, but since there isn't they just don't change - 70% market shares says so.
Geeks and Linux fans spend much effort touting the greater elegance and whatever many other strengths, but seem to be totally oblivious to the fact that Toyota can pull customers away from GM and Ford ONLY BECAUSE THE UI IS THE SAME and the technical improvements/advantages require NO learning curve.
In the business parlance of the day the saying goes: OUR (business/product) value proposition is... That's great because it means we understand that a value proposition is key, but we should strive to phrase it in terms of THE BUYER, not in our terms. Perhaps: FOR CUSTOMERS THAT WANT X we offer Y that does.. It would set the right frame of mind for OUR mind.
So, in closing here is a thought for all of you innovators out there looking for ideas: Produce a Linux that looks and feels (the UI) identical to XP, so that "XP migrants" have NOTHING to learn, make it run like Ubuntu on old PCs (those obsoleted by Win7), with a minimalist looking IE9-style web browser that requires no new PC and OS (Firefox/Chrome?).
Beware - The window of opportunity is closing fast. Android, a garden open to all, will soon deliver a uniform UI from cellphone-to-tablet and eventually to PCs (as Apple promises from pad-to-phone). If a new option with an XP-UI appears, it can still capture the transition moment, and you can capture a market, else the UI consistency from phone-to-PC promised by Android will be the David's stone that slays the arrogant Goliath.
Either way it will be the UI, not the power, not the features, not the economy, the UI.
P.S. I loved my birth language UI, I changed it because of the ROI, but cannot forget the price I paid