Actually, you are right with regards to the bang pash of hashbang - that’s just a Google defined way of letting them access the content.
Let's say I'm writing a web based word processor, and a user clicks on a document. I want the URL to be a reference to that specific document. The only way to change the URL to be specific without requiring a whole-page refresh is to use the hashbang syntax.
The hash without the bang. It's only been done for about 10 years. You can put whatever you want after the hash. It's up to your application to decide the meaning of it.
For an example, see the OSX SDK documentations: http://developer.apple.com/library/mac/#documentation/Cocoa/...
<a rel="nofollow" href="/#!about/" title="Click here to go to About">About</a>
without the hashbang shit, that's a nice normal link that any normal http client can work with. you can still layer on ajax to bind to that link, and use the fragment for the benefit of browser state and user interaction should you think that is a wise thing.
I also don't see why these web pages (specifically twitter) shouldn't be an ajax application - I think loading a page statically, and then the data via ajax is a Good Thing™.