## Monday, January 23, 2017

### TRAI's consultation towards a net neutrality framework in India

by Amba Kak, Mayank Mishra and Smriti Parsheera.

### The context

The Telecom Regulatory Authority of India (TRAI) has issued a Consultation Paper (CP) on Net Neutrality seeking inputs for the formulation of final views on the subject. This comes almost a year after TRAI's regulation prohibiting discriminatory tariffs for data services based on content, framed using its power to determine the rates at which telecommunication services are to be provided. The present exercise covers a broader canvas of trying to identify the acceptable limits of interference in the provision of Internet access services. This includes practices like blocking, degradation or prioritisation of specific traffic, which often form the focus of the net neutrality debate. In TRAI's words, it is an attempt to "rethink the first principles of traffic management by telecom service providers (TSPs)".

While issuing the discriminatory tariff regulation, TRAI had highlighted the importance of "keeping the Internet open and non-discriminatory". This idea also flows through the CP and the pre-consultation paper that preceded it in May, 2016. In fact, TRAI acknowledges in the CP that the term "net neutrality" is being used in its commonly understood sense of equal or nondiscriminatory treatment of content while providing access to the Internet. The word "equal", however, does not appear in the ultimate question posed by TRAI on what should be the "principles for ensuring nondiscriminatory access to content". The CP does not clearly spell out the reason for this. It could be due to the difficulties of monitoring equality in a best efforts delivery system; or perhaps because the term non-discriminatory already captures the concept of equality.

### Key issues raised by TRAI

The CP is reasonably comprehensive in its coverage of all major aspects that countries have considered while formulating their positions on net neutrality. Several of these, like reasonable traffic management, scope of prohibited activities and need for transparency were also discussed by TRAI at the pre-consultation stage. The difference, however, is that this CP takes a deeper dive into those issues, identifying the different approaches that could be considered and, in some cases, also weighing their pros and cons. On some issues, TRAI explores new areas, not covered in its earlier papers on the subject. The following are some key points discussed in the CP.

• First, TRAI notes that each country's approach on net neutrality is defined by its local context. Accordingly, it refers to some India-specific factors that may influence its approach. These include, the predominantly wireless character of access services -- 97 percent of Internet subscribers are on wireless networks. The CP later explains that traffic management on wireless networks may pose certain unique challenges due to spectrum constraints and variable usage patterns. It also refers to India's circle-wise licensing regime which often results in the usage of third party networks outside one's home service area. TRAI queries, who will be responsible for any neutrality violations in such situations?
• Second, the CP raises pertinent questions about the appropriate footprint of regulation, in terms of the services that are covered and the persons rendering them. In particular, it refers to the potential exclusion of "specialised services", which could be defined in several different ways. The manner in which India eventually chooses to answer this question will have far reaching effects on the adoption of future technologies in the country, particularly in the context of the Internet of Things revolution.
• Third, beyond trying to identify the reasonable limits of traffic management, this CP gets into more practical aspects of detection and monitoring of violations. It also suggests a collaborative approach for implementing the operational aspects of net neutrality, which, if adopted, would be a novel approach for the Indian telecom sector.
• Fourth, TRAI discusses the role of disclosures and transparency in guarding against discriminatory traffic management practices, an issue that was also raised in earlier consultation papers. However, in this case, it goes on to suggest two approaches on how this can be achieved -- a "direct approach" that would require a TSP to make specific disclosures only its own users; and an "indirect" one would also involve transparency towards third parties like content providers, consumers groups, research organisations and users of other TSPs.

### Scope of acceptable traffic management

The chapter on "Traffic Management" in the CP sets out the crux of the net neutrality policy debate. First, it explains why traffic management is an integral function of access providers -- to address congestion, security and the integrity of the network. Next, it notes that while such motivations for traffic management practices (TMPs) could be considered "reasonable", others might be "non-reasonable" due to their potentially discriminatory and anti-competitive effects. As such, it explains why regulating traffic management is being considered as a policy option and then mulls over the varied regulatory approaches that could be adopted.

Congestion management, which is explained in terms of the variability of demand or "peak-load", particularly on wireless networks, is explained to be one of the reasons for which access providers might legitimately engage in TMPs. TRAI recognises that the real solution to such issues lies in enhancing the overall network capacity but goes on to note that "even in a situation of enhanced capacity, some degree of scarcity might persist", hence creating a role for traffic management. The key takeaway from this discussion is that "differences in network architecture and technology" will play a role in assessing the reasonableness of any TMPs.

Next, the CP highlights that the same commercial considerations that prompt the use of traffic management tools to improve network performance could also become the cause of exclusionary or discriminatory practices. It notes that there have been global examples of TSP interference in networks for patently anti-competitive purposes, namely "service blocking, prioritising affiliated content provider services or throttling competing ones". While this explains the calls for regulation, traffic interference driven by commercial arrangements is not the whole scope of TRAI's enquiry. Instead, the CP provides two conceptual frames that might be used for such regulation -- the "broad approach" and the "narrow approach".

The "broad approach" appears to be a principles-based approach to identifying which practices would be considered reasonable. Drawing from the European Union's regulations, it refers to guiding principles like proportionality, non-discrimination, transparency and absence of commercial considerations that can be used to define the bounds of reasonableness. Practices like application-specific discrimination, throttling encrypted content, deep packet inspections, etc. can then be tested against these standards.

In contrast, the "narrow approach" will confine itself to formulating a "negative list" of practices that will not be permitted, without going into the contours of reasonableness. TRAI leaves the content of the negative list open for discussion, but gives the specific example of practices motivated by commercial/strategic partnerships with content companies as a potentially proscribed activity.

The distinction between the two approaches does not appear to be one of mere semantics. The paper acknowledges that a negative list is, by nature, likely to be restricted to situations that we are aware of today -- "This may motivate providers to develop other types of business practices that are not explicitly covered in the narrow restrictions although they may have similar harmful effects". It also notes that there could be difficulties in establishing a commercial motive. One way to address this could be to treat the lack of any objective/ technical reasoning for a traffic shaping practice as being indicative of "commercial motive", even where this may not be explicit.

By weighing these different approaches, TRAI seems to acknowledge that regulating TMPs is a tricky exercise, with high likelihood of false positives and negatives. Narrowly defined ex-ante rules may therefore be an uneasy fit with the complex and technical nature of traffic management. The controversy over AT&T restricting Apple's FaceTime application, its high-quality video-calling service, on its cellular network, presents an interesting example. In 2012, when Apple first launched FaceTime for use on mobile networks, AT&T declared that the service would be available only on select pricing plans. On the face of it, this constitutes an application-specific discrimination that violates net neutrality principles, as pointed out by many neutrality advocates. The case, however, also throws up several complex issues, which were discussed in the case study prepared by a Working Group of FCC's Open Internet Advisory Committee.

First, pre-loaded applications, like Facetime, are likely to enjoy more large-scale adoption, thus more likelihood of creating pressure on the network -- only about 10 percent iPhone users voluntarily downloaded Skype, while all had Facetime preloaded on their phones. Second, high-bandwidth video calling applications put a particular strain on mobile data networks, in both the upstream and downstream direction. FaceTime, in particular, was found to consume "on average 2-4 times more bandwidth than a similar Skype video call" at that point of time. Third, limited trial deployment of a new application, for instance, by limiting it to particular pricing plans, could be useful for gathering measurement data to assist in developing better TMPs. Fourth, application management can take place on the device, as was happening in this case, or on the network -- "whether it matters where an application-management decision is enforced, and which organization decides on it". TRAI has also touched upon some of these aspects in the CP by raising questions on "application-specific discrimination", "duty-bearers" in a net neutrality regime and impact of traffic management at the level of the device or operating system being used by a person.

### Conclusion

The public discourse that preceded the discriminatory tariff regulation took place in the shadow of Facebook's Free Basics offering. It led to heated debates and sharply polarised views, many of which were focused on the specifics of Free Basics. In contrast, this current consultation is taking place in a relatively less charged environment, with no poster child violation. This presents an opportunity for the regulator and stakeholders to proactively engage with one another for developing a suitable framework for India that can be tested against a range of potential practices. The real test will be to ensure that whatever principles India chooses to adopt at this stage convey a strong regulatory message on non-discrimination, but also have the flexibility to adapt to the dynamic environment of this industry.

The authors are researchers at the National Institute of Public Finance and Policy.

Please note: LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.