I think one thing that hardware manufacturers, OS publishers, and application developers tend to overlook about touch screens is that for actual work (as in a person sitting down for hours and interacting with them), they can be quite fatiguing to use.
If they're used flat, they cause neck strain from constantly looking down at them. If they're used vertically, they cause arm, shoulder, and wrist fatigue. If they're tilted a bit like some people like their keyboards to be positioned, the displays tend to be hard to read. And in any event, touch-screen keyboards provide no wrist support and have tactile feedback that's mediocre, at best.
None of these matter very much if you're using the machine to find the nearest Wally World or the latest line on the third race at Aqueduct. But for work where you're interacting with the machine for hours on end, touch screens can make your days miserable.
That's my biggest gripe with the idea of a "universal" interface a'la Windows 8. There can be no such thing because there is no such thing as universality in the ways that people use computers. It's very much like saying that driving a tractor-trailer should be just like driving a Ferrari because they're both vehicles, that playing a guitar should be just like playing a glockenspiel because they both make music, or that using a hammer should be just like using a screwdriver because they both drive fastening hardware.
Another thing that can't be ignored is that historically, the biggest rushes to upgrade have been motivated by one (or both) or two things: A crappy predecessor, and increased functionality.
Take Windows 95, for example. The improvements were earth-shattering, from the 32-bit architecture to the user experience and multimedia capabilities. People camped out overnight at stores to buy Windows 95. Windows 98 was a decent seller on the consumer end because of improved capabilities that home users, mainly, would want, but the rush was nothing like the Win95 rush.
On the business side, Windows 2000 was a good seller, but not a "hotcakes" seller. It was marketed mainly to business users as an upgrade to NT, which was already cherished by business users because of its stability. But a fellow I used to know at MS said that a very large percentage of users who upgraded to Win2K were actually home and small-business users who had been using 98 and craved stability. (I actually remember having to talk a few home users out of upgrading to Win2K after ascertaining that it wouldn't run all of their family members' applications, such as most games.)
Companies that had been using NT tended to be neither in a hurry to upgrade to Win2K, nor afraid of doing so. NT was a solid, stable system, so as long as it served business users' needs, they were happy with it. But Win2K preserved user experience, had relatively few hardware or application incompatibility problems, was rock solid stability-wise, and was an easy, painless upgrade from NT, so there was no real fear factor, either.
XP was a very hot seller on the home user and SoHo side. When XP came out, I couldn't keep up with the upgrade demand -- and that was without actually marketing XP upgrades other than with a few lines on my Web site. XP had a familiar interface, but considerable improvements in functionality and vast improvements in stability, memory management, and security. Most of my XP upgrades were consumer-initiated because they were sick of the instability issues with 98 (or even worse, WinMe).
Business users, however, held back from XP for quite some time because of considerable application incompatibility issues and the lack of any functionality improvements that they cared about. Many of the differences between Win2K and WinXP had to do with sorting out problems with VxDs and other issues related to upgrading from legacy 9x systems to a system based on an NT kernel, and business users couldn't care less about any of that. Win2K was stable, solid, fast, and ran their applications nicely. They had no need of XP.
So again, home users who had been stuck with the obsolete and increasingly unstable Win9x (or the bastardized WinMe) flocked in droves to XP because it represented a vast improvement in stability, with little or no functionality loss, and a consistent interface that was only slightly different from what they were used to. But business users, for whom there was little improvement and some headaches, held off upgrading.
And then there was Vista, which in its original release combined some stability issues and application compatibility problems (many related to UAC) with little in the way of improvements that users actually cared about. It also sported more changes to the interface than users felt like accepting, and therefore it rapidly faded away into history, with users actively avoiding it, and in fact paying extra for downgrades.
And then came Win7, which implemented what was very good about Vista -- but perfected -- and also had many under-the-hood improvements that users could feel more than understand. It was very well-accepted by home users (many of whom had tried Vista and paid to get it gone from their machines), and reasonably well-accepted by most business users (except those who were using ancient applications) because they did understand the improvements; and barring any hardware or application compatibility problems, the upgrade and learning curves were shallow.
And now there's Win8 -- an outstanding OS under the hood, but whose improvements, quite frankly, are not as apparent to users as they are to geeks who study such things. Furthermore, it comes with an interface that many people hate, and which will result in some re-learning and temporary productivity loss on the corporate side (which businesses hate). Its most noticeable improvements are that it makes your computer look like a phone -- not something that the corporate world has been loudly clamoring for -- and that it's optimized for touch screens, which people either like or don't, depending mainly upon how they use their computers. It also follows Win7, which is an excellent OS; so there's no rush to be rid of it.
So viewed in the historical context, it's not at all surprising that Win8 sales have been lackluster. It offers few improvements that most people care about (much less understand), and comes at the cost of re-learning and productivity loss for many users. And really, that's a shame, because under the hood, it's probably the best OS Microsoft has ever released. It deserves more respect.
-Rich
