More transparency into government requests

About two years ago, we launched our interactive Transparency Report. We started by disclosing data about government requests. Since then, we’ve been steadily adding new features, like graphs showing traffic patterns and disruptions to Google services from different countries. And just a couple weeks ago, we launched a new section showing the requests we get from copyright holders to remove search results.

The traffic and copyright sections of the Transparency Report are refreshed in near-real-time, but government request data is updated in six-month increments because it’s a people-driven, manual process. Today we’re releasing data showing government requests to remove blog posts or videos or hand over user information made from July to December 2011.

Unfortunately, what we’ve seen over the past couple years has been troubling, and today is no different. When we started releasing this data in 2010, we also added annotations with some of the more interesting stories behind the numbers. We noticed that government agencies from different countries would sometimes ask us to remove political content that our users had posted on our services. We hoped this was an aberration. But now we know it’s not.

This is the fifth data set that we’ve released. And just like every other time before, we’ve been asked to take down political speech. It’s alarming not only because free expression is at risk, but because some of these requests come from countries you might not suspect—Western democracies not typically associated with censorship.

For example, in the second half of last year, Spanish regulators asked us to remove 270 search results that linked to blogs and articles in newspapers referencing individuals and public figures, including mayors and public prosecutors. In Poland, we received a request from a public institution to remove links to a site that criticized it. We didn’t comply with either of these requests.

In addition to releasing new data today, we’re also adding a feature update which makes it easier to see in aggregate across countries how many removals we performed in response to court orders, as opposed to other types of requests from government agencies. For the six months of data we’re releasing today, we complied with an average of 65 percent of court orders, as opposed to 47 percent of more informal requests.

We’ve rounded up some additional interesting facts in the annotations section of the Transparency Report. We realize that the numbers we share can only provide a small window into what’s happening on the web at large. But we do hope that by being transparent about these government requests, we can continue to contribute to the public debate about how government behaviors are shaping our web.

Posted by Dorothy Chou, Senior Policy Analyst