Through the looking glass: Analysing transparency reports

Over the past decade, a few private online intermediaries, by rapid innovation and integration, have turned into regulators of a substantial amount of online speech. Such concentrated power calls for a high level of responsibility on them to ensure that the rights of the users online, including their rights to free speech and privacy, are maintained. Such responsibility may include appealing or refusing to entertain government requests that are technically or legally flawed, or resisting gag orders on requests. For the purposes of measuring a company’s practices regarding refusing flawed requests and standing up for user rights, transparency reporting becomes useful and relevant.Making information regarding the same public also ensures that researchers can build upon such data and recommend ways to improve accountability and enables the user to understand information about when and how governments are restricting their rights.

For some time in the last decade, Google and Twitter were the only major online platforms that published half-yearly transparency reports documenting the number of content take down and user information requests they received from law enforcement agencies. In 2013 however, that changed, when the Snowden leaks revealed, amongst other things, that these companies were often excessively compliant with requests from US’ intelligence operations, and allowed them backdoor surveillance access to user information. Subsequently, all the major Silicon Valley internet companies have been attempting to publish a variance or other of transparency reports, in hopes of re-building their damaged goodwill, and displaying a measure of accountability to its users.

The number of government requests for user data and content removal has also seen a steady rise. In 2014, for instance Google noted that in the US alone, they observed a 19% rise for the second half of the year, and an overall 250% jump in numbers since Google began providing this information. As per a study done by Comparitech, India sent the maximum number of government requests for content removal and user data in the period of 2009 - 2018.8 This highlights the increasing importance of accessible transparency reporting.

Initiatives analysing the transparency reporting practices of online platforms, like The Electronic Frontier Foundation (EFF)’s Who Has Your Back? reports, for instance, have developed a considerable body of work tracing these reporting practices, but have largely focused at them in the context of the United States (US). In our research, we found that the existing methodology and metrics to assess the transparency reports of online platforms developed by organisations like the EFF are not adequate in the Indian context. We identify two reasons for developing a new methodology:

1. Online platforms make available vastly different information for US and India. For instance, Facebook breaks up the legal requests it receives for US into eight different classes (search warrants, subpoenas, etc.). Such a classification is not present for India. These differences are summarised in Annexure
2. The legal regimes and procedural safeguards under which states can compel platforms to share information or take content down also differ. For instance, in India, an order for content takedown can be issued either under section 79 and its allied rules or under section 69A and its rules, each having their own procedures and relevant authorities. A summary of such provisions for Indian agencies is given in Annexure 3.

These differences may merit differences in the methodology for research into understanding the reporting practices of these platforms, depending on each jurisdiction’s legal context.

In this report, we would be analyzing the transparency reports of online platforms with a large Indian user-base, specifically focusing on data they publish about user information and takedown requests received from Indian governments’ and courts.

First, we detail our methodology for this report, including how we selected platforms whose transparency reports we analyse, and then specific metrics relating to information available in those reports. For the latter, we collate relevant metrics from existing frameworks, and propose a standard that can be applicable for our research.

In the second part, we present company-specific reports. We identify general trends in the data published by the company, and then compare the available data to the best practices of transparency reporting that we proposed.

Download the full report. The report was edited by Elonnai Hickok. Research assistance by Keying Geng and Anjanaa Aravindan.