track download counts in a privacy preserving way
Webservers automatically see who is downloading what from them, so this data can be used without worsing privacy, provided that it is not stored without stripping out the sensitive bits. While download counts are not infallible, they are a very useful source of data about activity and usage. So we should again keep the download counts as an extra set of data to mix in with the Popularity Contest.
Here's Tor Project's method for sanitizing Apache logs: