The problem is that everything in JS exploded during just few years (Node was published in 2009), and it took few years to build tooling around Node to start utilising all this power.
And then everyone started to build "rich" web applications (it is called differently every year), with the "best" tool existing. I personally see two major problems – usually cool startups who can afford using the bleeding edge will die soon (or will get enough money to completely rewrite it), so 'maintanence' in idiomatic way doesn't work here – life cycle is extremely small, and it reflects in the way libraries are written in JavaScript. Also, the barrier is low, and therefore we have a lot of complex websites built by novices, running in development mode in the actual production.
Community also is very "poisoned" with the idea that you _have_ to work in your evenings (contributing to OSS, writing your own or just playing with hot technologies), and therefore new stuff is baked extremely quickly. It is usually very low quality, and you have to cope a lot in order to make it work, and often you have to dive into source code, and it is considered as normal, which is quite sick from my point of view.
Also, it is truly "hype" community, which is super excited about what is hot now – and it means that today kings will be overthrown very soon (right now stuff like Cycle.js and Vue.js is gaining traction). Nobody wants to maintain their libraries – and very often you can see projects with a lot of stars completely abandoned.
Recently, Babel.js was defined as _defacto_ standard, and it basically started transpiling era, which is not going to end in the foreseeable future. It shakes the stability, and makes extremely easy to add new features (like stage-0), and people just used to unrealiable tools.
So, all of this works as a snowball, which only speeds up. People create more and more complex web applications, and also js is going into other fields (like microcontrollers, Node.js grows in popularity, React Native is super hot now). When it will stop? Nobody knows. Maybe WebAssembly after implementing will stop it (people will spread over competing technologies), maybe no.