http://foo.com/#!/some/page

But when a spider crawls the site use this:
http://foo.com/?_escaped_fragment=/some/page

and have the page serve up a straight up html implementation optimized view showing whatever SEO code you want.

The end user should -never-see the “escaped_fragment” url, so the idea its messy is not really relievant. It could be _sussage_ for all it matters, its just a string to identify that what follows is a…umm…escaped fragment.

History tokens/fragments by themselfs, meanwhile, are a perfectly fine way to exchange and store states without needing cookies or servers to know what your looking at.
If you really need universal links in your app that work even for people without javascript, then you can do it….but your app would need to supply a “?=” style link, which could be auto-converted to a # when people have javascript on.

Honestly though, the idea of somehow the net was “better” before apps like GoogleMaps or Gmail is just nuts.

Streaming data makes a lot of sense for many different use’s, and HTML as a layout/presentation layer can work just dandy with data being streamed too it.

“Which means I have to program code in order to get on the effing web?”

No, it doesnt. I dont even know what you mean by that statement.

Im in the middle of making a internal Html5 ipad app, that HAS to have that app feel. using hashbang urls with PathJS allows us to dynamically load content. without moving away from the page, giving that app feel our client wants.

The links also preserve state too – yes its two requests- but the main page is loaded then the hash handled. it being internal, we don’t worry too much as the multiple requests …