It would definitely need to be more elaborate than how I described it in my original comment. But I imagine, for example, that could be solved with a list of "

It would definitely need to be more elaborate than how I described it in my original comment. But I imagine, for example, that could be solved with a list of "preferred engines", which of course would probably fallback to a default "user or system preferred" engine if none were available. It's true it would become a small problem of its own (to try to figure out how to build for just enough engines to make it possible to render correctly on all computers) but I think that makes it possible for the entire ecosystem to have many more browsers and for a lot more experimentation to take place. Because the browser developer no longer has to worry about "Well, if I can't get enough market share to matter to developers, then there's no real point for me to build a rendering engine, since it will probably never be used". Instead if there is even a small number of people interested in their work, there can be a small community of developers who build for that engine, and it will be used (or at least recommended) when a user visits their websites.