Beginning in the 1980s through today, CAD software followed the same trajectory as other desktop software: it began with a modest set of functions, and then over the decades bolted on more. What once fit a 360KB diskette now requires a multi-gigabyte download.
Until 2012, feature-bulge was considered the norm in software development: add more functions to software despite having customers who used only a small percentage. Because software companies cannot predict at the per-customer level the functions wanted by each one, the only solution was to give everyone everything, inefficient as it is to present users with software that does too much (cf: the ribbon).
Until now, there were only two solutions to the everything-plus-the-kitchen-sink approach: verticals (write plugins to personalize the software for specific workers) and subscriptions (price the software as if it were smaller). Both solutions, however, required the ongoing distribution and training of multi-gigabyte and giga-functional software packages.
The smartphone and tablet present a third way: granularity. In this approach, customers buy small apps from a huge marketplace, apps that perform specific tasks, yet communicate with one another. The two phrases keys in this sentence are huge marketplace and communicate with one another.
The huge marketplace (hurrah for capitalism!) ensures customers find apps that suit them best. It also ensures small programmers have sufficient customers. This environment creates diversity, where every programmer has a different approach and sets of features, and so I come to prefer Plume over Twitter; Opera Mobile over Chrome; K9 over Gmail; and so on.
By communicating amongst one another, apps are freed from the need to be monolithic. I read an article in gReaderPro, pass it along to my son through K9 email, post it to Twitter through Plume, and save it on ReadItLater (Pocket). Android and iOS are written to make this possible; Windows, Linux, and OS X are not. When a CAD or Office package wants to do similar pass-along tasks, it does them all on its own, by design.
Granularity allows app developers concentrate what they do best, instead of trying to do everything less well.
I don't see the granularity revolution changing today's monolithic-oriented CAD vendors, frankly, and perhaps it isn't a suitable model for how designers work. Can you imagine Autodesk passing an Inventor model to Abaqus for analysis by Dassault Systemes? Me neither, but that's the path happening today on mobile devices. I can, however, see the likes of Autodesk and Dassault wanting to put a stop to granularity on tablets and smartphones, and making their mobile apps monolithic.
(Heh: The biggest monolith of them all, Microsoft, began in late 2012 to suffer from the granularity revolution, as its attempt to extend through Windows 8 its desktop monoply to tablets and smartphones shows signs of failure on all three fronts.)
Nevertheless, I am envious of how well granularity works on my Android devices, and so I wonder if it might be a solution to the now-mature CAD market.
PS: I understand that granularity goes against natural market forces. As a company becomes large, it seeks to monopolize its customers and the rest of the marketplace (cf: Facebook and Autodesk). Indeed, the likes of Facebook and Autodesk have no choice but to attempt to create monopolies of themselves. Which I understand, because I'd love to have the monopoly on CAD books!
Granularity works well only when software companies are forced to remain small (due to overwhelming numbers of competitors, as found on the Android and iOS online stores) and so cooperate with each other.