This is a new series of articles titled “Open Ideas for India”, where I hypothesise on technology interventions that can make an impact in India.
Recent times have been challenging for the relationship between society, government, and social networks. There have been alleged issues with meddling in democratic elections, appropriation of power to decide who can speak, unfair compensation of news agencies, banning of social media run out of specific countries, amongst many continuing cases of violations of privacy.
One may not be surprised at this turn of events. Digital technology has become so pervasive that it is but natural that these social networks reflect the good and evil already in our society, if only more acutely. The role and regulation of such social networks is perhaps one of the most defining debates of our times.
My concern with this article is not to propose ways to regulate social networks. Instead, I want to try to understand how the (many good) ideas from social networks can be put to service in creating social impact. More specifically, I am concerned with the question: How can social networks become social nets that work for the next billion Indians?
I go about this in three steps:
First, I try to make sense of the trajectory of evolution of social networks across data modalities.
Second, I outline the broad features of social nets that could work for the next billions Indians.
Third, I enumerate the open challenges towards building such social nets.
Part 1: Trajectory of Evolution of Social Networks
I like to think of the evolution of social networks in a simplified framework as shown in the following diagram. The diagram shows the evolution of social media along the four modalities of text, images, video, and audio. Note that the evolution along these modalities is not a strict temporal one; instead it is the evolution of ideas in distinct organisations at different points of time which in hindsight I collate along an axis of evolution. In this simplified view - across each modality, social networks have evolved in two distinct stages. I chronicle these two stages with some companies which seem to reflect that stage of the evolution the best.
The first stage addressed the challenges of democratising tools of production and building platforms for easier discovery. The former ensured a large number of producers1 created content cheaply (such as YouTube simplifying video sharing) and the latter ensured that consumers knew about these producers and their work (such as Apple podcasts showcasing many indie podcasters).
The second stage of the evolution further accelerated both content production and consumption. On the production side, the barrier to content creation was reduced by allowing amateurs to be producers (such as tweeting by individuals who may not have earlier fancied blogging). On the consumption side, these networks were optimised to have content go viral (such as TikTok showing a stream of videos from people one may not even know). These innovations definitely led to further increase in popularity of these networks, but with unclear overall outcomes.
Thus, the first stage was motivated by the honourable goals of democratising content production and building platforms at scale for discovery. The second stage, on the other hand, was driven by the objective of harvesting the maximum human attention. This was done perhaps at the expense of a lower signal-to-noise ratio of the content on these platforms and also primed its users to have addictive and impulsive behaviours. Many of the challenges mentioned at the start of this article seem to be related to this improper choice of objectives.
This decomposition of the evolution of social networks into the two stages helps clarify my intention: The innovations in the first stage over two decades have much to offer to us as a society if framed in better objectives than the ones chosen in the second stage. With this intention I ask the question - What does a social network designed for the next billion Indians look like?
Part 2: Social Nets that Work for Next Billion Indians
In this part, I outline the contours of social nets that work next billion Indians. Before that a side-comment: Technology is not culture-neutral. Building for the next billion Indians requires a deep understanding and appreciation of India’s cultural contexts and use-cases. This combination of technological and cultural know-how is particularly missing. More capacity must be built amongst those who are culturally rooted to innovate with technology.
The Modality of the Future
On to the social net that works for the next billion Indians. The first choice concerns data modality. The diagram on the evolution of social networks above is drawn out for different modalities - text, images, video, audio. Which one amongst these is most promising?
As illustrated by the wild success of Clubhouse that launched in a pandemic world, audio as a modality has many advantages:
There seems to be more of a human in one’s voice than in one’s written word.
Audio enables real-time distributed creation.
Consuming audio does not require one’s full attention (as opposed to reading Twitter timelines or watching TikTok videos).
Further to these advantages, audio has few India-specific advantages:
Text input tools in Indian languages still remain a major, if not the most important, barrier to content creation in the long list of Indian languages.
Audio democratises participation amongst people of lower literacy - which as a group represents a large fraction in India.
Audio is cheap to produce (microphones are available a plenty) and cheap to transport (audio can be compressed and piped along cellular channels if broadband is unavailable).
Those are strong reasons to consider audio as the exclusive modality for the social nets that work. Audio as a modality also suggests a name for our social net - Vaak2, which is Sanskrit and many other Indian languages means speech.
The Dynamics between Producers and Consumers
The second stage of the evolution of social networks blurred the lines between producers and consumers. As we noted, this attempt quickly runs against challenges of misinformation and consequent challenge with content moderation. For Vaak, which aims to maximise social impact rather than engagement, parity between producers and consumers is neither required nor desired. Thus, Vaak plays very different roles for producers and consumers.
A high bar must be set to allow production of content on the platform, i.e., Vaak must ensure that a root of trust can be established in each user who is allowed to produce content. This trust must entail from existing and verifiable social credentials. Such dependence on credentials may put those outside the establishment at a disadvantage, but only until algorithms infer credibility based on participation on Vaak. On the flip side, the value of existing credentials would decay if not matched by the high standards of quality produced on Vaak.
While the bar is high for producers, Vaak must simplify on-boarding of consumers as much as possible. Vaak should support consumers with or without an app, with a smartphone or even with a feature phone. Indeed, the ability to host a range of consumers with varying degree of digital access and literacy would be a central piece to the success of Vaak. Further, Vaak must make it easy to discover content that brings meaningful change to the consumers’ lives.
Thus, the primary responsibilities of Vaak are to verify and maintain a high level of trust on on-boarded producers and to simplify access to and discovery of meaningful content for a wide range of consumers.
Components of the Audio-based Social Net
So, what exactly is Vaak? I discuss below some contours of the key components.
Production: At its core Vaak is a platform where a selected set of producers are allowed to record conversations on identified topics and release them to the net. These conversations are recorded on an app that allows the producer to moderate and edit the conversation.
Consumption: Producers can invite individuals to listen in by sharing a link or a toll-free number. Consumers can listen to conversations either through an app, a website, or by dialling a toll-free number guided by an IVRS in their local language. The listening process should be hassle-free: For instance clicking on a link in a message should directly lead the consumer to an interface (app or website) and start listening without any further authentication.
Discovery: Consumers can discover conversations on the app based on proximity on several features including geographic, linguistic, professional, cultural interests. Algorithms would learn over time to recommend more meaningful content to consumers. Social sharing with deep links will enable wider dissemination.
Dialog: Some of the conversations would only be available live while others would be archived and available to listen for specific durations. Live conversations may allow consumers to participate as enabled and moderated by the producer of the conversation. Some of the archived conversations may also allow consumers to post private or public questions which can be followed up with responses.
Credibility: Producers would have socially verified credentials, which will continue to be updated based on algorithmic and manual analysis of quality of produced content and subsequent dialog.
Authentication: All authentication would be based on only one’s primary phone number. Specific roles (such as producing content, or participating in a conversation) would require further authentication performed on an as-is-needed basis.
Thus, Vaak should be designed to provide trusted producers with the tools to create conversations and a wide range of consumers to seamlessly listen into and at times join in on the conversations. While this lays out the structure of Vaak, it would be more insightful to discuss some of the use-cases that Vaak can support.
Creating specific Social Impacts
Social networks are designed to be the pipes. Indeed this is not just metaphoric, but a deep feature that allows them to abdicate legal responsibility for what flows through their networks. On the other hand, Vaak which is built for creating social impact, must clearly visualise, advertise, and facilitate specific use-cases. Indeed, Vaak must measure itself against the success of these use-cases. I list some of these use-cases below.
1. For knowledge sharing between professional groups
Digital Green is a pioneering organisation that trains farmers by disseminating offline videos where farmers with small landholdings chronicle their problems, solutions, and success stories. This is a good example of how knowledge sharing can empower a professional group to collectively succeed. In this standard use-case, Vaak would host organisations like Digital Green as producers to share their content for wider dissemination amongst farmers. Farmers on Vaak would automatically be prompted when content relevant to them (similar crop, similar geography, same language) is available. Vaak here would provide a larger listener base to magnify Digital Green’s existing social impact. Vaak would also enable farmers to post follow-up questions and comments to Digital Green.
Another professional group, quite different in nature, are the lakhs of teachers using the Diksha app. The app facilitates sharing of knowledge, imparting of professional skills, and discussions amongst teacher. An embedded version of Vaak could be part of the Diksha app and facilitate both creation of conversations by identified moderators and listening by consumers. For instance, a daily evening show could be hosted to discuss specific issues of benefit to teachers. Vaak here would provide the technological plumbing to have a convenient, live conversation amongst teachers. Alternatives such as a Zoom call or YouTube event would require more planning from producers and more setup for consumers.
In both these use cases, Vaak provides an easy-to-use and louder speaker phone to scale-up the social impact in existing efforts to create communities in professional groups.
2. Anonymous support groups
An important use-case for Vaak is to enable anonymous participation in the many safe sharing spaces for discussing vital issues such as mental health. Identified moderators of these spaces would host live conversations which can be joined by consumers without revealing their identity. These conversations will maintain full privacy and disallow archival of the audio.
Sometimes, individuals may not recognise their alignment with a specific support group and may be in need of general mentorship. This is particularly true for college students who face new and challenging situations and are in need of a steady stream of motivation and support. Vaak can support such conversations anonymously.
In both these cases, apart from anonymity, an important component is discovery. Based on location of the consumers, Vaak can recommend nearby groups to consumers. For instance, for a college student, Vaak can recommend a mentorship program in her college that is hosted on Vaak.
3. Sabhas
Panchayats are a great example of working grass-root democracy in India. To magnify their impact, meetings of the panchayat can be Vaak conversations to which people from the nearby villages could have a live access to. This would bring transparency and also allow participative democracy. As of today, no tool exists that enables panchayats to cheaply and effectively enable such dissemination. A no-hassle simple-to-use Vaak app tested in the ground zero of Indian villages can contribute to making democracy work at its roots.
To ready more students to become participative citizens, parliamentary debates are held in schools and colleges. These debates provide vital opportunity to learn, understand, and shape opinion on important issues. Currently, there is no single platform for such debates around the country. Vaak can on-board moderators from certified education institutions and broadcast their debates to interested students.
In both these use-cases, Vaak provides the technical plumbing for a greater voice to democratic practices. In addition, the clear and sharp focus on such specific use-cases can help create guides on conduct of these democratic practices.
4. Government’s speed-dial to citizens
Governments in democracy are heavily invested in communicating their performance to citizens. Advertisements in TV and newspapers, radio programs, active Twitter accounts, etc. are now standard staple of government’s public relations. However, these methods are either too broad-based (eg. an advertisement in a top national daily) or do not address the next billion Indians (eg. a Twitter feed on a ministry’s new policy). Vaak would provide an effective mechanism for each government’s department to have a direct line to citizens who are currently served by policies of the department. For instance, when a citizen opens a Jan Dhan bank account, the Finance ministry may invite the citizen to a regular stream of Vaak updates on financial literacy. Such updates can be personalised to the language and background knowledge of the citizen.
99 dots is an influential program in a similar space. It is being widely deployed in the nation’s battle against Tuberculosis (TB). When a person is known to be infected with TB, he/she is put on a medication plan for 99 days. Proper follow-through of this medication for this period of time is crucial for completely defeating the disease and avoiding further spread. A patient enrolled in the program can be on-boarded on a Vaak conversation and receive periodic updates directly on the phone. These updates would provide the required support and information on each phase of the recovery.
A similar case can be made for the contemporary vaccination program for CoVid-19. For instance, an embedded version of Vaak could be implemented in the COWIN app. This would allow the citizens who have been vaccinated to receive constant inputs from the medical body and also have the chance to report any adverse symptoms. Such efforts are particularly important in the light of the increasing hesitancy in vaccination.
In these use-cases, Vaak can support a “drip schedule” wherein each person enrolled on to a Vaak conversation would get a series of pre-scheduled conversations at designated intervals from the date of on-boarding. Engagement of the consumers on such content can also provide vital feedback on improving the programs further. Need for such engagement exists across verticals and apps, and Vaak can provide a common platform to facilitate that.
There may be several other use-cases to consider. An essential goal is to identify a small set of these use-cases with high social impact and build templated experiences in Vaak to support these. As mentioned, Vaak’s success would be based on maximising the impact of these use-cases rather than prioritising proxy metrics such as monthly active users. Vaak’s evolution over time would also be based on learnings from these use-cases.
Part 3: Open Challenges Ahead
In premise, the design of Vaak is relatively simple in today’s software engineering terms. However, there are several challenges to address towards reducing costs, improving access, and sustainability. I note a few of these below.
Technical challenges
Network: India has made significant progress in access of high speed mobile connectivity over the last few years. The data usage in India is amongst the highest in the world, while data costs are amongst the lowest in the world. However, we still need to innovate to make voice-over internet protocol (VoIP) work seamlessly in real-time with low resources.
Audio: At the heart of Vaak is audio quality. Background noise is perhaps the single biggest complaint today of those forced to work through conferencing tools. We need efficient techniques to reduce such noise and improve audio quality in real-time across a range of hardware devices.
Artificial Intelligence (AI): India has lagged the world in adoption of AI technologies. This is particularly visible in the poor state of affairs in AI technologies for Indian languages - both text and speech. Vaak would heavily depend on Automatic Speech Recognition (ASR) working on all major Indian languages. Recent efforts such as the renewed focus on the National Language Translation Mission are welcome, but need to deliver. Our efforts at AI4Bharat, which I co-founded with Mitesh Khapra from IIT Madras, is a small beginning towards this.
Human Computer Interaction (HCI): To address the next billion users, Vaak must have an interface that is efficient for the consummate producer to moderate conversations and is simultaneously easy to learn for a new consumer dropping into a conversation. Achieving this range of features is a major challenge and must be informed by active research in the HCI space.
Monetisation
Monetisation and sustainability of Vaak is a crucial challenge. There exist several options in current social networks. Several social networks (eg. Twitter) try to monetise human attention by serving advertisements. Some others (eg. Spotify podcasts) charge subscription fees which makes them better aligned to users’ needs. Some others (eg. Substack) allow producers to charge consumers a fee. Yet others, do not have explicit monetisation (eg. Clubhouse) affecting public’s trust on them.
Vaak’s aim is to create social impact as a technological platform at scale. It thus cannot serve advertisements, charge a subscription from its users, allow producers to monetise content, or have implicit business relations. Vaak cannot also be government funded, for that would affect its independence and speed of innovation. The only possibility therefore is for Vaak to be funded by publicly known foundations and non-profits whose aims align with Vaak’s.
Data privacy and Open source
Vaak must stand by the highest standards of data privacy. Data should not be shared with any third party, unless it is required by authorities as prescribed in the law of the land. Any usage of the data within Vaak must be clearly specified. Data should also not cross Indian borders and be stored in data-centers hosted within India.
Vaak must be developed in the open-source. This not only creates the highest possible openness for what Vaak does, but also encourages socially-minded technologists to contribute to improving Vaak. It would also allow replication of Vaak in other geographies and communities accelerating both social impact and learnings to improve the core technology.
Summary
Today, we are facing some very tough challenges in the regulation of social networks. However, we should remember that the social network revolution over the last two decades has been incredibly innovative. While we work to fix current social networks, we need to think afresh on what is the social net that would work for the next billion Indians. I make the case for an audio-based Vaak social net. Vaak should make production and discovery of audio content very simple. It should be designed to maximise the impact through specific use-cases. It should be developed in the open-source with funding from publicly trusted foundations. And it should become a platform to sound the deep inner voice of a nation with an ancient civilisation but many new aspirations in the decades to come.
I use the word producers while a more common denotation is creators.
Vaak here is just a placeholder for ‘social net that works for next billion Indians’. There are no copyrights placed or intended.
Awesome explanation about social net, Indian culture and Vaak.