With all the turmoil in the design of operating systems and applications lately, you’d think there would be more actual improvement.
But there’s not, and there can’t be, and I’ll tell you why. 
First, though, let’s take a short look at the history of books.
When books first started being printed en masse in the late 1400s and 1500s, they were produced in an extremely wide variety of formats and sizes, with widely-varying designs, with many attempting to imitate high-quality manuscript books.
It wasn’t until the 1600s or so that books really standardized as a format. A book from 1650 wouldn’t be that distinguishable from one today, other than the archaic language and font. The format – margin, one line per page, no columns, etc. – would not be any different.
With operating systems and applications something similar occurred, just in a shorter time frame. The first interfaces were clumsy due to both resource limitations and because the fields of UI and UX were new. Everyone then was in uncharted waters.
After about 20 years of use by the general public, a few very-similar standardized interfaces developed that best melded productivity and approachability. ![[AOL%2520vs%2520Windows%25208%255B4%255D.jpg]](http://lh6.ggpht.com/-CzNeiteWOio/T8jPxpuLcwI/AAAAAAAAJ4s/dvmv4tDFqIs/s1600/AOL%252520vs%252520Windows%2525208%25255B4%25255D.jpg)
Note that these might not have been the absolute best approaches – though how one can measure that is questionable – but they worked well enough for the majority of people. Furthermore, once enough people become accustomed to a use paradigm changing it reduces productivity greatly, and often for a very long time, for only extremely minor benefit.
My point is then that GUI design standardized at a near-optimum with current human and technology limitations around a decade ago, and that designers now attempting to introduce useless “innovations” like Metro and Australis even if their designs are better by some amount serve to actually reduce net productivity.
If something like 3-D interfaces or neural interfaces ever occur, then it would certainly make sense to re-examine and overturn many if not all widely-accepted UI design conventions.
However now all designers really can do is make things worse since interfaces (at least prior to the latest design manias) are already fairly near optimal. Ergo, even slightly net positive changes are harmful as it takes most people many years to completely acclimatize to such major changes with minor benefits.
I keep having the paranoid idea that there is an ongoing campaign to polarize digital media into two discrete camps:
programming elite (who can circumvent the massive inconveniences that are being built into systems)
facebook zombies (more to be pitied than scorned)
People in the mid-range, like me, who need computers to get work done (but who have little idea about the internal workings of the critters) are being slowly but surely squeezed out. I can’t think of a single feature of Word after 1997-2003 that is an unambiguous improvement.
And this is not a case of being old and out of date. My students have almost no idea how to use computers to get work done so this is definitely not a question of young people being in the forefront of technological progress.
I’ve written elsewhere on this blog about how people younger than about 28-30 or so will probably not be able to use standard keyboards, as they are only accustomed to phones and tablets, and so will not be able to work in a regular office environment with any competence.
So I think you are right about that, about being polarized into two discrete camps — producers (which likely will one day require licensing to access a general purpose computer) and consumers, with little to no room in the middle.
It was only for a short time, and a complete historical anomaly, that young people were in the forefront of tech progress. That would be roughly those kids who were brought up with unrestricted general purpose computers in the home, roughly between 1995 (when they became common) and roughly 2005 or so. Perhaps a little later, but that’s roughly the largest portion of them. Outliers like me were using GP computers as early as 1979, but that’s not typical. Probably 80% of the computer literates will be in that range.
Before that and after that, there won’t be much general computer literacy, so anyone after the 2005 or so cohort will be just as tech-illiterate as someone born in 1935.
Good observation, by the way. Not many people realize what’s really going on as they are blinded by the myth of eternal progress, when in reality 90% of people without training in 2030 won’t even know how to use a real keyboard, which will still be required to get any actual work done.