I am loathe to polyfill modern browsers adding completely needless overhead trying to chase some panacea. At least Chrome and Firefox need to agree on (and be in the process of implementing) a feature for me to consider using it and polyfilling it.
The Web Components thing seemed overengineered and too much work for component developers to get wide adoption outside of widget toolkits. All that behind-the-scenes offsetTop confusion tells me that the way Google envisioned makes it a bad fit for the DOM/CSS/JavaScript hodgepodge cruft-pile that has built up over the years. The fact that there are so many open proposals out there for the various features should be a signal to architects to run the other way on Web Components.
So, into this vendor mess strides React, (and to a much lesser extent Angular and Ember due to ocean-boiling), which gives us a fancy component model atop standard DOM and JS. And at just the right time -- web apps are far too big this decade to be coded monolithically. I mean, everyone knows that, but what could you do about it before a decent UI component framework appeared?
Being able to compose your app in terms of nested components is the way build large, high-quality, well-tested web apps because it lets us reason about parts of the UI rather than the whole. And the DOM performance with React ain't too bad neither. And being able to render the same React component statically on the server and dynamically on the client solves a lot more problems too.