Maybe something similar to what we have now, but exposes much lower level aspects of the browser and gives you more fine grained control over it. WASM is a good starting point in that it lets you use whatever language you want. Now, we need a browser API that will let us draw directly to the window if we want to. I think a lot could be accomplished by creating the UI portions of a site programatically then dumping the content into a "renderMarkdown()" function.

That's just my idea, though. Hopefully, the current sad state of the web and the coming browser monoculture will spur others to experiment too. I know some gaming companies have experimented with the distributing and executing of x64 binary executables using the browser. I know others are looking into ways to chop up and do progressive loading on binary executables.

The appification of the Internet qualifies here, with Google Fuchsia running on Chromebooks and ARM-based desktops being a good opportunity for them to move most Internet based activities in apps on a platform they control.