I disagree. Browsers should adhere to standards and not enhance without changing the standard first. Enhancing standards has always been caustic to the web and is how we end up with nonsense like sites that only work with specific browsers.
That's not how the web has ever worked. HTTP/2 is a standardised version of SPDY. Google put it out, people with a compatible browser (Chrome) got SPDY, people with a non-compatible browser got HTTP/1.1. HTTP/3 is being built significantly on QUIC.
Nothing wrong with that, provided that the developer is ensuring cross-compatibility before they bolt on experimental features.
The risk comes in a browser/engine monoculture where the incumbent (Google) is in a position to say "Chrome now uses n by default. Too bad if your server doesn't support it". This is what we had when IE6 ruled the web, Netscape was dead and MS could bolt on proprietary features.
In my examples, Google is the origin of both SPDY and QUIC, but they didn't have the power to simply say "this is what you're using now" (like MS did) because other significant browsers and browser engines existed (Gecko, Trident, EdgeHTML). They had to sell the protocols to the community and convince them to adopt and support them, eventually being written up as standards.
Enhancing standards is fine. So long as it's done in a graceful, open and backwards-compatible fashion. If it's a bad idea, it won't matter and eventually will die. If it's a good idea, someone will eventually write a standard for it.
Hell - HTTP wasn't a standard. TBL knocked together this new protocol, people preferred it to Gopher, and eventually it became a standard because everyone was using it.