[liberationtech] Berkman's Circumvention Tool Usage Report

[Danny O'Brien]

http://cyber.law.harvard.edu/publications/2010/Circumvention_Tool_Usage

(For those for whom 13 pages is TL;DR, there's also good summary at http://www.technologyreview.com/web/26574/ with comments from many of libtech's most Esteemed And Usual Suspects)

Great read, with actual data and carefully described methodologies for how that data was gathered.  I for one spent slightly too much time after reading the paper exploring https://www.google.com/adplanner/ and  http://www.google.com/insights/  (eg the regional interest stats for http://www.google.com/insights/search/#q=facebook+proxy&cmpt=q and http://www.google.com/insights/search/#q=proxy&cmpt=q are both fascinating.). Of course, the Berkman folk warn everyone to take their estimates with a heavy pinch of salt, but it's nice to find some research that has something other than salt as an ingredient :)

The big quotable conclusion is that circumvention usage is small (Berkman estimated <3%, even in heavily filtered countries). I don't think that would come as much of a surprise to anyone here. The real meat for me was in the relative usage of circumvention tools/web proxies/VPNs and http proxies, with web proxies coming out a bigger winner than I would have guessed, both in terms of their penetration and the lack of blocking by filtering states. 

My hasty, fact-free analysis: while I don't doubt that 3% figure, I wonder if web proxy predominance points to a wider *variability* in circumvention usage. Web proxies are convenient and easy for accessing the occasional blocked site, as opposed to VPNs and circumvention tools which assume continuous use. I can imagine that such knowledge once used once becomes a unexercised but everpresent skill among those who have needed it in the past. That would depress monthly usage numbers, but would allow for large jumps in circumvention if/when a controversial topic is blocked.

It's also not clear to me how widespread circumvention needs to be  before the policy goals of introducing filtering are undermined. It may be that for most topics that would be traditionally censored *and* are of potentially widespread interest, like corruption stories, strikes or protests, or opposition declarations, just a few users obtaining the information and then spreading it through unblocked routes is just as effective as widespread circumvention tool usage. 

Finally, the Berkman folk were very coy, but I'd still love to know what VPN company it is who also runs a filtering blacklist service.. As the paper says, "we leave as an exercise for the reader [the act of] deciphering the considerable value for a filtering company of running its own circumvention system."

d.