I wonder how these results compare to the user-agent analytics of hacker news.