All right, finally got a nonempty archive. Looks like it's probably complete. Seems fairly legit. However, there is one thing that confuses me so far. Looks like a lot of links get a link_target attribute that is different from the actual link:

  {
   "type" : "LINK",
   "text" : "http://en.wikipedia.org/wiki/American_Letter_Mail_Company",
   "link_data" : {
     "link_target" : "http://www.google.com/url?q=http%3A%2%2F\
                      en.wikipedia.org%2Fwiki%2F\
                      American_Letter_Mail_Company\
                      &sa=D&sntz=1&usg=[alphanumeric token scrubbed]",
     "display_url" : "http://en.wikipedia.org\
                      /wiki/American_Letter_Mail_Company"
    }
  }

However, links to Youtube, or Google, or any subdomain of either, have a link_target attribute identical to their display_url. In-teresting. I wonder what that alphanumeric token is. It's like how Google search results also have, rather than normal links, links to google.com/url?[a ton of URL parameters]. I assume the latter is so they can gather data about what URLs are clicked in the search results, or possibly pasted elsewhere, and conceivably to discourage scraping. And as for this?

Perhaps it's a kludge for Google chat clients, which need to parse URLs (they do something special with Youtube links) and might be thus freed to do it stupidly. Perhaps Google wants to know what people do with their downloaded Hangout archives. Perhaps Google wants to know what people do with Hangout history in the browser, and they've changed the links in that archive, and then they just leave it that way in the exported format. --Turns out Hangout history in the browser has exactly those links... I'm guessing it's the last one. Well, at any rate, at least it's easy to ignore that field.