Can understanding infrastructure help us reach data equity?

Digest #70

Data has become a fundamental resource in modern society. The embedding of digital platforms and tools in most aspects of life has generated vast amounts of data that provide granular insights on human behavior and most facets of the world around us. This has catalyzed data driven decision-making for everything from guiding business strategy to informing public policy – placing a centrality to data’s role in the functioning of society as a whole. Yet despite the embedding of data into society, it has raised many tricky governance questions that are not being grappled with effectively. Control over data, and access to it, is very skewed in the modern digital economy with already marginalized groups often feeling further negative effects. There are unfortunately an abundance of inequities when it comes to the collection, processing and sharing of data.

There have been efforts over the last decade to address these inequities in various forms. Data protection legislation, data governance mechanisms and bottom-up initiatives that seek to provide communities with greater control over their data have all emerged. These efforts have had varying degrees of success, but have ultimately been unable to fundamentally alter the concentration of power in the digital economy, and unlock the full potential of data for larger societal benefit. If anything, the last few years have seen us moving in the opposite direction, with a deepening of inequality and the few who have access to data and the ability to use it growing even more powerful.

Deciding how governments and society as a whole govern data for equity is incredibly difficult, but we think that applying an infrastructural lens to the digital ecosystem can be helpful in deciding a way forward. Infrastructure is typically understood as the basic systems or facilities that help a society function, and the governance of infrastructure typically accounts for things like access, resilience and equity. If we can identify the ‘infrastructures’ of the digital economy - those systems that are essential to the functioning of the digital ecosystem - we can also examine how the governance of traditional infrastructures can be applied to the digital ecosystem, and ultimately help us achieve better data equity.

The ‘infrastructure’ of the digital world and why it is important

Understanding digital infrastructure helps uncover opportunities and harms, as well as providing insights on how to design better governance. For example, even if control over and access to data was completely fair and open, this on its own is insufficient to achieve data equity. Data needs to be understood alongside the other core components of the digital ecosystem: hardware, the cloud, and standards and protocols. Without these, extracting insights and value from data is impossible. These too face inequities and while regulations for these infrastructures exist, research has shown that their governance is often blinkered, with gaps that miss both the harms and opportunities stemming from these systems. For instance, there has been a focus on regulating the cloud infrastructure market from the lens of privacy and security, but there is a gap on deepening market oligopolies and threats to environmental sustainability.

These infrastructures play an important role in the data value chain by enabling the movement, collection and use of data, as well as in the larger digital ecosystem, for example the AI value chain. AI models are trained using hardware in the form of microprocessors (or chips). To meet the extremely high volume of computer power required to train AI systems, developers often use cloud infrastructure such as storage, processing, and even services and tools. This hardware and cloud infrastructure is used in tandem to train AI models using vast amounts of data. Standards and protocols determined by global institutions provide an underlying basis that allows for communication between the various systems at play. All of these different infrastructural elements are harnessed by developer communities to build AI models. However, the governance approach around AI models seems to focus more on the models themselves and the datasets used to train them rather than the enabling inputs.

How understanding the ‘infrastructure’ can help us govern data more fairly 

Ensuring equity and justice requires adapting our governance approach and we believe this should also allow room for dynamic innovation with digital ecosystems, with flexibility for the growth of individual components without affecting the entire environment. Again, thinking about the different infrastructures of data and the digital economy can help us allow for this. 

Terminology around ‘infrastructure’ is not new in the digital ecosystem, it is however a scattered notion. The term ‘infrastructure’ itself doesn’t have a clearly agreed upon definition but there are a few commonly recognized characteristics: 

• It is critical - The basic services, without which primary, secondary and tertiary types of production activities cannot function

• It facilitates further production - A system of interaction of economic agents, ensuring a link between phases of production and consumption

• It is modular - elements can be combined and adapted for customized solutions that are independently developed but seamlessly integrate into the rest of the ecosystem

• It is dynamic and has interrelated components - They encompass dynamic networks and assemblages that enable and control flows of goods, people, and information over space

• It addresses societal needs - The networked assets must be designed to address societal needs, which may be most clearly evidenced in the aftermath of service disruption

• It generates externalities - Their operation leads to positive and negative externalities, which are the indirect benefits or harms of a project that are complex and difficult to predict

Looking at these characteristics of infrastructure, we have identified four basic ‘infrastructures’ for the digital ecosystem, namely: data, hardware, the Cloud, and standards and protocols. 

We must not forget that these infrastructural elements are incomplete and unable to function without one other critical component: people. While communities may not fit within the technical notion of infrastructure, there is something to be said for “social infrastructure” that encompasses those involved in the creation and continued functioning of these more technical infrastructures. Taking communities into account when considering governance mechanisms for the aforementioned infrastructures can enable the centering of equity and broader social value in data creation and use. 

Having mapped these as infrastructure, we at Aapti are currently working on identifying the buckets of harms and opportunities associated with these infrastructures to inform recommendations on how best to govern them for equity. Broadly we have identified:

• Sustainability and resilience

• Competition and innovation

• Individual protections and collective rights

• Privacy and system security

• National security and sovereignty

• Access and participation

These are common to all four of the digital infrastructures, but play out in specific ways for each. We are currently working on a classification of various governance mechanisms by analyzing the way existing digital infrastructures are governed to account for these harms and opportunities. For example, we have examined how executive action (in the form of a government mandate) and legislative action (in the form of the Electricity Act, 2003) in India in the energy sector illustrate different mechanisms for ensuring sustainability (by ensuring compulsory action towards incorporation of environmentally sustainable practices to reduce long-term harm​) and resilience (providing need-based measures that can bolster domestic markets with global support). We hope this research can serve as a guiding framework for specific data governance mechanisms that account for the harms and opportunities presented by the infrastructures that support them and enable a more equitable and just digital ecosystem.

We conclude this post with a meta provocation: 

If “infrastructure” is taken as one metaphor for elements or systems within the digital ecosystem, are there other metaphors that can help in guiding us towards a more equitable digital world? If you have any, we would love to hear them!