>Desktop was simple earlier in the cycle because Microsoft rapidly reached a quasi-monopoly.

I would go one step further: the MS straight jacket stifled some needed diversiveness and moved a lot of the troubles evoked to the QM - remember dll hell and OS incompatibilities ?
>
>Apart from .Net, Web dev tools are made by non-software companies which seemingly use this as a lever to attract talent.
>
>The last 10 years have seen an ongoing battle between Google, Facebook, Twitter, Yahoo, some large startups and FOSS-driven groups of developers.
>Google: GWT, Angular, Chrome, and more
>Facebook: Firebug, React among others
>Twitter: Boostrap
>Yahoo: YUI
>Ruby on Rails: Prototype/Scriptaculous
>and many, many others
>
>Most of these products require a complete redesign / rewrite of applications to accommodate.

strong disagree: if COMPLETE rewrite is needed, the previous implementation was not nimble / uncoupled / flexible enough.
Did enough ports (sometimes after refactoring unflexible implementations) to be certain of that.
It does sport the skylla of possible perf problems by call overhead unneccessary in single layer designs battling the charybdis of brittleness, but with a "chunky, not chatty" mind set seldom a real problem.

>
>Unless some of these large actors suddenly disappear, it'll remain an open, ever changing market until some sort of standard design will be able to link all elements (HTML, CSS, JS, server, user state) in an harmonious and productive fashion.
>
>As these elements are separated by design (viva la separation of concerns) - as opposed to VFP or others where they're all tied together (boo, no good) - it'll probably take another 10 years until web app dev. reach the same level of RADiness.
>

IMO you need better glue routines: something much easier in Python, JS or vfp compared to C# or Java. But simple minded glue implementations will create MORE work ;-))