Server-side javascript is only relevant here if it has access to a working DOM, since that's what any HTML generation or templating is going to be based on. At

Server-side javascript is only relevant here if it has access to a working DOM, since that's what any HTML generation or templating is going to be based on. At that point you basically do have a headless browser, whether you call it that or not. I find Google's proposal goofy; I think I agree with the poster who suggested feeding google static HTML (probably via a sitemap) which references canonical URLs that might be ajaxy, rather than having it crawl an ajax site and do this magic token manipulation to get the static HTML equivalent.