Then take into account any changes to the DOM, innerHTML calls or document.write()s? Seems simple and worthwhile. Sure, google couldn't link to anything created by AJAX, but it would know the content was there, and maybe take that new content (like my news comment) into account...
It wouldn't surprise me if they already have started working on something like this, but it's a pretty hefty project. Despite the W3C standards and Javascript RFC, think about how many different ways Javascript is handled by browsers these days. HTML code is pretty much WYSIWYG (even with CSS), but Javascript can change the rendering of the page in many many different ways. There are also countless libraries that they would have to debate supporting and the amount of "improper" javascript out there could throw alot of developers into a headspin when their results change. I would love to see them do it, but I think the scope of the project is a little insane at this point.
They could certainly use any number of available parses out there. I just don't think it would be as "simple" as you suggested. Security and other issues aside, I think the resource-intensive nature of Javascript (on a large scale) might be the big hindrance. Javascript is an actual scripting language, not just a presentation/markup language. For example, HTML doesn't have infinite-loops.