The Future of Internet Access
Facebook's Free Basics, a repackaging of Internet.org, has sparked off a worldwide debate on Net Neutrality. But what is the viability of alternate internet models in the atmosphere created by this very debate?
Free Basics, Facebook’s free ‘internet’ platform, has been a topic of fierce debate over the past few months. It is now available in roughly 35 countries through Facebook’s partnerships with mobile carriers who see it as a way to persuade people to buy data plans. Facebook insists that it is a utility that will connect hundreds of millions of new internet users to a few, essential services and show them the power of the entire internet. Free Basics is open to all app developers that meet its technical standards and Facebook claims that it is consistent with the tenets of net neutrality.
Those opposed to Free Basics, including internet activists, IIT professors and comedians, state that it violates net neutrality and turns the internet into a ‘walled garden’. Just recently, the Telecom Regulatory Authority of India (TRAI) called for comments to its Net Neutrality consultation paper to determine what the appropriate framework for governing net neutrality and differential pricing should be in the country. Facebook and the TRAI have since clashed over the nature of responses to the consultation papers.
There is widespread consensus on the importance of getting the people of India connected to the internet at scale and the ensuing socio-economic benefits. The public consciousness is heavily in favour of doing this as effectively, fairly and quickly as possible. The country’s online population soared past 400 million in 2015 and is predicted to touch ~600 million by 2020. But in order to sustain this, there is a vital need for better networking infrastructure to support business models that facilitate low cost access.
As with most of you, I’ve been reading various perspectives on the issue, to educate myself on the numerous variables and come to an informed opinion. SavetheInternet.in has done a terrific job aggregating countercomments to the arguments for differential pricing for data services.
I thought I’d evaluate the relative merits of a few popular internet model alternatives along the dimensions of internet conversion effectiveness, business model sustainability and net neutrality, as evidenced by whatever unbiased data is out there. So here goes:
1) Zero Rating of Limited Services by Telcos and Private Partnerships
This includes services such as Facebook Free Basics, Airtel Zero and Wikipedia Zero.
Facebook claims that 40% of new internet users on Free Basics platform subscribe to the full internet within 30 days of using the application. One of their reasons for limiting the number of accessible services as opposed to applying a data/rate limit to full internet, is less potential consumer confusion for new internet users. Facebook believes high-bandwidth actions such as watching video on data/rate limited services would be a terrible experience and turn users away from the full power of the internet. They pitch Free Basics as an ideal ‘bridge’ to the internet as it gives users access to some essential services on the internet and a glimpse of its potential, which in time will let them ‘level up’ to the full internet.
The country’s online population soared past 400 million in 2015 and is predicted to touch ~600 million by 2020.
Apart from Free Basics, telecom operators have been offering their own zero rated services including Facebook/Whatsapp data plans/packs and services like Airtel Zero and unfortunately there is very little open, empirical data on their effectiveness. Facebook suggests that Free Basics is a more sustainable model than data subsidies as it “supports diverse, dynamic internet markets with thriving developer communities and an abundance of local content so that people have an opportunity to consume a vast array of content and services.” However this creates an artificial barrier for consumers to create their own local content for other users by subjecting them to an approval process and having them compete with other zero-rated offerings.
There is a need to understand actual end-user preferences in specific economic and cultural contexts before deciding on the best access model. Some preliminary qualitative research by Amba Kak at Oxford shows that new, lower income users in India would prefer a constraint on how much they can use the internet, not how much of the internet they get to use. The study shows that while WhatsApp and Facebook were undoubtedly the most popular communication platforms, internet use was far from restricted. Fear and confusion associated with data billing has led to zero-rated plans being rejected, confirmed by executives from telecom companies as well as recharge shop retailers themselves. Short duration, unlimited access plans appeared to be the most popular option.
They (Facebook) pitch Free Basics as an ideal ‘bridge’ to the internet as it gives users access to some essential services on the internet and a glimpse of its potential, which in time will let them ‘level up’ to the full internet.
How has Free Basics performed so far? Mahesh Murthy, venture capitalist, lashed out against the metrics reported by Facebook stating that they were misleading. He says that the majority of Free Basics users (~80%) were already users of the full internet. Reading closer into the data, though he admits this is his own personal interpretation, it seems that there are effectively 10K new internet users that are locked into Free Basics. One wishes the data was more transparent.
The cost of Free Basics is borne by the telecom operators with the strategy that the platform will establish a funnel that acquires a larger number of new internet users that will convert to paying customers for the full access internet plans. Zero rated offerings like Airtel Zero (that offer music streaming services like Wynk) and social media packs attempt to lure new users by assuming the costs of popular services or having content providers (such as Flipkart) do so. Telecom operators continuing to zero rate services is not sustainable if users do not convert to recurring customers at scale, which is necessary to build, maintain and grow network infrastructure.
Net Neutrality Costs
Telecom operators should provide access to the network that is the Internet, and allow consumers to choose where they wish to go. They should not turn ‘access’ into an ‘audience’ business where they become the gatekeepers between an internet business and its customers and provide preferential treatment to those that they partner with. Free Basics, Airtel Zero and similar services by other telcos like Idea Cellular and Vodafone could lead to the carving out of private Internet bubbles that distorts access to information and knowledge.
How real is the lock-in syndrome with a platform like Free Basics? In Indonesia and Nigeria, a limited Geopoll survey that Quartz commissioned, showed that there were a significant number of users (about 10% of FB users in the surveys) for whom the boundaries between Facebook and the Internet were blurred. In fact more than half of those who don’t know they’re using the internet, say they “never” follow links out of Facebook, compared with a quarter or less of respondents who say they use both Facebook and the internet.
It’s premature to extrapolate lock-in dangers without country-specific research. It depends on the ability and awareness of users around the internet. There is a very real danger of competition being stifled but without enough transparent data on specific user behaviour it is hard to predict the extent of these effects.
Zero rating schemes have been completely banned in The Netherlands, Chile and Slovenia. There is the non-absolutist argument which says that non-profit, non-competitive services such as Wikipedia Zero (with local language support) and government information websites can be zero-rated. This is a slippery slope and we will need to decide whether there is any room for zero rating at all.
2) Free Data/Rate-limited Access to the Full Internet
Broadly speaking, this category include telcos offering free rate limited services (Aircel’s free 64 kbps service) or free data packs to the unrestricted internet or the government providing data plans as a direct benefit transfer (DBT).
Aircel is offering free 64 kbps service working on the 2G spectrum. The first three months of service will be free after which users will have to recharge for at least Rs 150 every month. They plan to roll out the service later this year, starting with Tamil Nadu. This model has brought up questions about the poor experience of surfing mobile websites that are not optimized for 2G speeds and confusing billing to customers that wish to upgrade to faster speeds or 3G or 4G. We need usage data to measure effectiveness.
Telcos like Idea Cellular, Vodafone and Airtel offered full internet data plans at free or discounted rates or with time restrictions on special occasions like Diwali. The effectiveness of seasonal promotions needs to be studied - but the fact that users prefer shorter duration, high data cap plans indicates that such plans may see successful adoption.
Telcos can provide free rate-limited 2G service when they are able utilize existing spectrum more efficiently and have recouped initial investment costs.
Direct Benefit Transfer, authenticated by Aadhaar (~950 million Indians now have an Aadhaar number) and covered by government funds like the Department of Telecom’s Universal Service Obligation Fund (USOF) are a good alternative as well. As Nandan Nilekani and Viral Shah say, Facebook can choose to contribute to this fund to bring India online. He suggests the government allocate at least ~120 MB of data annually to every user, provisioning the first 10 MB of free data usage for each user every month. With the Aadhaar infrastructure, he claims that Data Pack DBTs can be deployed within 3 months.
Telcos like Airtel have opposed DBTs claiming that they are not different from zero rating by TSPs but actually less inclusive as zero-rated plans allow users to access that content without being Internet customers. This is untrue, as phones capable of internet access through zero rating will work with DBTs too, once a SIM has been activated. In fact, DBTs have all the advantages of zero-rating minus the restrictive and anti-competitive effects. Of course, not all Indians are on Aadhaar yet and some manual override would be required for everyone to obtain the DBT.
Some are skeptical about the quality/speed of execution and the impact on low income, new internet users. Will such a nominal amount of data per month be enough to experience value and what type of awareness/education is required to aid discovery and exploration? LPG DBTs provide LPG subsidies to the bank accounts of a 100 million Indians, which is a good sign that that such a system can be installed successfully, but it is not without criticism.
Telcos can provide free rate-limited 2G service when they are able utilize existing spectrum more efficiently and have recouped initial investment costs. In Aircel’s case, particularly, they were unable to secure new spectrum in February 2015, so this makes good business sense for them. For data packs, telcos will expect high conversion rates to 3G in order to continue discounted or free handouts - which will happen as consumers see the benefits of 3G over 2G speeds. For major ISPs, data usage and revenues, especially on 3G, is growing rapidly. The incentives are high for TSPs to invest in data or spectrum for provisioning data services and since data is no longer cannibalizing voice revenues.
For DBT, the USOF has a pool of about 40,000 crores to deploy towards universal access. Based on Nilekani’s proposed scheme, the costs of providing 120 MB per year for a potential ~800 million users at 25 paisa/MB on 3G would cost 2,400 crores per year. And at the government’s scale, this could be at lower tariffs - which is promising.
Net Neutrality Costs
All these schemes align with a broadly accepted definition of net neutrality. They do not involve differential pricing of content services, zero-rating and provide access to the full, unrestricted internet.
3) Ad-supported Free Data Plans
This model is essentially having a company pay for unrestricted internet services (usually a data cap) with a “brought to you” attribution or third party advertisements. Mozilla has been experimenting along this model with Orange in Africa and with Grameenphone in Bangladesh. There are apps in India that use a sponsored data model as well.
Free data plans on the purchase of handsets are a good way to promote sales and convert users into paid data users.
Orange, in African countries like Senegal and Madagascar, launched plans to give users 500MB of free data to access the entire internet and 6 months of free voice and text, on purchasing a $37 Alcatel Klif Firefox OS handset. Mozilla grants 20MB of free unrestricted data to Grameenphone users in Bangladesh for watching ads on their ‘WowBox’ marketplace. However, Mozilla just recently announced that it was not continuing to offer Firefox OS phones through its carrier partnerships due to user experience issues. It is still early to comment on the success of the ad-based model but it could be successful in India, especially if cheap Android OEMs can partner with ISPs to offer deals.
There are also apps like mCent from Jana and Gigato, that provide data credits to use for unrestricted internet access, upon using particular partner services. mCent has over 30 million registered users, is making healthy revenue and was the fifth most popular app in India in 2015. It has partners like Twitter, Amazon and Flipkart subsidizing data costs for full internet access.
This costs of internet access are carried by the sponsoring companies but is also transparent to all parties. Companies paying for equal rating access are compensated with brand value and advertising revenue. The sustainability of the model will come from how profitable ads shown to low-income users can be, and in the cases of Gigato and Jana, whether app impressions are meaningful in the long-term (app developers pay based on performance). Free data plans on the purchase of handsets are a good way to promote sales and convert users into paid data users.
Net Neutrality Costs
Ad-to-access models do not support differential pricing, zero rating and provide access to the full, unrestricted internet. They are also transparent about who is paying for services and brand compensation. That said, Jana and Gigato do offer a competitive, discovery advantage for their partner services that are sponsoring data. App data usage on these platforms is tracked in order to estimate data rebates and billing. Both these platforms work with developers of all sizes, from local shops to big multinational brands who pay the same rates. It remains to be seen if this has any anti-competitive effects but being non-discriminatory, it is a much more neutral alternative to walled gardens like Free Basics.
So I’ve come away with three major conclusions:
1) We need a lot more unbiased evidence on what kinds of internet access models align with new internet user needs. Models should be net neutral and access should not be controlled exclusively by private interests. We need systematic, empirical studies looking at the effectiveness and sustainability of existing models and pilots/trials for new models.
2) The need of the hour is to build robust networking infrastructure at scale to for better connectivity and capacity. For low cost internet access to be sustainable, networks need to be beefed up, more internet fibre needs to be laid down, especially in rural areas, and the government also needs to invest and partner with innovative delivery networks (hotspots, mesh, drones etc.) while preserving net neutrality.
3) We need transparency around cost structures and efficient implementation. Direct Benefit Transfer, on the surface, seems like the best long-term approach thus far to connect India and ad-based models would act as a great catalyst until we make large scale infrastructural improvements. Given these options, we don’t need to choose between universal access and Net Neutrality when we can have both.