In that case "web application" is a misnomer. If the current state has no natural URL, it's not a legitimate part of the World-Wide Web. Instead the authors are tunneling a proprietary protocol over AJAX to carry opaque content to a single-purpose GUI app, just like all the terrible client/server apps from the 90s only slower.
If the current state has no natural URL, it's not a legitimate part of the World-Wide Web.
But most of the popular content accessible via the web now fits this description. Look at Google's own homepage, it's a complex Javascript application that's completely opaque. @bruceboughton is right, the problem isn't that people aren't respecting the WWW specification, it's that the specification is no longer adequate to describe what the Web has become.
Google has competent web developers who practice progressive enhancement. Their search form and results have stable (even sensible) URLs and are perfectly usable without trusting their js.
Sure the application has a URL. But once I load it, it constructs what I see on the page on the fly. I can turn Javascript off and load an alternate page, but if I leave it on I'm loading an application that's every bit as opaque as Flash.
