Not gonna disagree with that assessment, either. It's not like industry has never been wrong before. I mean, really, the web design community is heading toward vertical-parallax, perpetual-scrolling monstrosities that give 10% of users headaches (literally) and can't be properly bookmarked or navigated. The industry is following them down that path!
There's a reason I'm not following suit: I prefer to make things that are actually usable! Yes, it's cool to add this or that new feature and make it look nice. Everyone wants to feel like what they're building or using is cutting edge, and that's just fine, right up to the point where you start breaking usability in the name of design.
Go ahead, re-implement the select box so you can ensure that it looks the same in all browsers, across all platforms. Add type-to-search and autocomplete to it. I'm all for that, you're improving the UX... unless you don't also handle switching between selections with the up and down arrows, entering and exiting the field via the tab key (trivial if you use a *real* select box and follow progressive-enhancement, replacing the select *for display* with your new control, this is not only possible, but trivial), and confirming the highlighted selection via the enter key. If you don't implement the functionality people are expecting *before* extending, then you're actually *breaking* things. And if you can't implement the native behavior it in a handful of lines of code, you probably shouldn't implement the control at all; most likely, any extensions you make will be inefficient and clunky, as well. But, I digress...
Creating GUI apps on the desktop isn't much better, in reality. You still have a multitude of languages; the application language and GUI framework, which is typically a language of its own (WYSIWYG form builders only carry you so far), if your application interfaces with a database you still have some form of SQL, and that API you need to interface with to access that bit of data from that 3rd party service? If you care about performance, you're proxying API calls through a local server that caches the results; if you need to intelligently clear or refresh bits of that cache based on some bit of application logic, you may need something more sophisticated than a simple HTTP cache, so there's your server-side application language. Don't get me started on version headaches with libraries...
I may have been born in the 80's, but I agree... they'd have laughed and taken away your beer if you had told somebody that programming the web was going to be so much worse than writing native applications. Hell, that's a good way to lose your beer today!