[zackbrown]

This can already be done today through canvas & webgl, or through clever combinations of DOM + JS + CSS. You can control every pixel that's rendered to the screen through the browser.

So why doesn't everyone roll their own? Some issues:

  - footprint - that whole runtime has to come across the wire just to render "hello world"
  - javascript's single-threaded tendencies make perf a big challenge
  - text rendering [this happens to be the thing the document-focused Web really excels at]
  - text semantics [SEO]
  - gigantic implementation effort [custom engines RARELY make sense even for the demands/budgets/returns of AAA games; how many websites justify this kind of investment?]

[TomMarius]

Yeah, that's because it's not really supposed to be used for that. Every single one of these is solvable if they wanted to support this use case. Ad footprint and implementation effort - of course there'd be just a few major ones that'd be cached at client side. Ad2 implementation effort, we can simply compile Qt or Gtk to WASM; it even works today, it's just slow because of platform limitations. This'd be used for applications primarily, document-oriented pages could stay HTML/CSS and HTML/CSS could be simplified thanks to complex web apps turning to other means.