It’s time to kill the web (2017)

I think another important consideration with respect to modern web development that this article does not touch on is the sheer quantity of dependencies, and their often ephemeral nature. Even a relatively simple web application might have hundreds or thousands of transient dependencies via npm (or similar), and on top of that may well depend on other live services running elsewhere on the web (such as SaaS, APIs into other sites, or other services built by the same author as the webapp). I feel that for many web applications today, a sizeable chunk of the modern internet is effectively a dependency of that application to run as it should.

It would seem that the house of cards must collapse eventually. At some point, either these problems must be addressed, or the ecosystem will begin to lose developer mindshare in favor of easier and more performant solutions.

On the other hand, I think that HTML+JS has become so popular for applications development because targeting a single codebase to many platforms is quite difficult. There is little in the way of "write once run anywhere" UI libraries, and HTML+JS fills that niche well enough that people are willing to use it. I think that most people would prefer tools that are faster, smaller, and easier to develop for than Electron (for example), but if you want to run properly on Windows, Mac, and Linux (let alone mobile), there aren't a lot of really compelling options at hand. At the end of the day, if you want to build a end-user focused GUI program accessible to a layperson which will run on Win+Mac+Linux+Android+iOS, web technologies, whatever their flaws might be, have little in the way of competition.