How do you stop users who aren't using the custom browser from accessing these 'tinyweb' HTTP sites? How do you prevent content scrapers and search indexers from accessing them? How do you suppress direct incorporation of 'mainstream' web content into 'tinyweb' content?
If your goal is precisely to create an parallel ecosystem that's "airgapped" from the mainstream web, and you're already going to have to develop custom clients, content formats, and server-side configuration to implement it on top of HTTP, and engage in lots of development work to imperfectly isolate the two ecosystems from each other, why wouldn't you just develop a parallel protocol and start with a clean slate?
If your goal is precisely to create an parallel ecosystem that's "airgapped" from the mainstream web, and you're already going to have to develop custom clients, content formats, and server-side configuration to implement it on top of HTTP, and engage in lots of development work to imperfectly isolate the two ecosystems from each other, why wouldn't you just develop a parallel protocol and start with a clean slate?