On 19th October, CONNECTED BY DATA brought together a group of participants to explore issues related to collective data governance, and particularly, to discuss the concept of ‘community’.
The term ‘community’ can be used in different ways, so the objective was to talk about what we really mean by a ‘community’, and what sort of communities there might be in the context of data governance. This can guide our thinking when we explore ways of getting communities involved in governing data and AI, or when negotiating access to data, or empowering communities to seek redress.
Jeni kicked off the discussion by posing some questions that CONNECTED BY DATA has been exploring:
- How formally constituted does a community need to be?
- When should there be a need for a representative organisation?
- How can organisations that work with data find and engage with affected communities?
- What happens when conflict arises within or between communities?
The Provocations
To set the stage for the discussion, we invited three people that engage with these questions in their work to share some insights:
- Vinay Narayan, Aapti Institute
- Matt Prewitt, RadicalxChange Foundation
- Helena Hollis, Connected by Data, drawing on work from her time at The Institute for Community studies which is powered by The Young Foundation
Vinay Vinay wanted to emphasise three issues. First, that we shouldn’t see community data rights as being at odds with individual rights. We should see how they may work harmoniously.
Second, on the idea of the community, a lot of current literature doesn’t theorise enough about community rights in different legal traditions. Legal rights for communities have been theorised about before and we should draw from this. Additionally, communities can be formed in different ways (e.g. intentionally, from the bottom-up, or algorithmically), or operate on different levels, so it may be challenging to come up with general governing principles. We might need a better typography of different communities, their interests and the rights they might need.
Third, water-tight distinctions between personal and non-personal data may be problematic if rights and protections are entirely hinged on that distinction. We may need to move away from the notion that the processing of non-personal data has no impact, and think of it in terms of a spectrum rather than a binary.
Matt Matt drew from John Dewey’s notion of a ‘public’ as something that forms in response to a problem. For instance, if an entity is polluting 2/3rds of a region, its inhabitants, who may have no prior connection, become a ‘public’ in response to the problem. Responses to many modern challenges like worker protections, environmental regulation, for instance, were brought into being through the formation of new kinds of publics. This conception of the ‘public’ is different from the idea of an organic public, which is a community where people already have some association or connection and may already be (trying to be) a community. This distinction between these new ‘publics’ and ‘organic publics’ is important to keep in mind.
Second, when thinking about whether communities need formal representation, we should approach it as a pragmatic question, rather than a question about the inherent nature of the community. Formal representation can be useful (or even indispensable) when intervening in negotiations around data, even if it may not be the best representation of the community. From the perspective of shifting the balance of power, we may sometimes need to sacrifice perfect representation for having representation in the first place. For instance, there will never be a perfect workers union that fully captures the complex interests of all its members, but it is still useful in offsetting the competing interests on the other side. The powerful interests against which we seek representation definitely don’t represent the community, but do have formal representation. So there may be a need for formal structure in response, even if it’s not a perfect representation of the community.
Helena Helena drew from her work exploring ways of making research funding more equitable, by funding communities (instead of academics) to do research that was useful to them. One of the findings was the issue of external labelling: communities might often reject the label of ‘marginalised’. A more helpful framing would be that of power, and what kinds of groups, and in which situations, have power. Having pathways to redress are also important, particularly when communities feel that they have been unfairly treated. For instance, when their information has been extracted by universities, but not for their benefit. Communities can define themselves in different ways, which can also change over time. For instance, there can be communities of interest or identity or place.
Another insight about engagement was to work with existing networks and relationships, instead of imposing external structure and having a preset formula for what communities need in order to benefit. There is a lot of knowledge work that might already exist, which can become the basis for support - we can focus on supporting that, instead of starting from scratch. So a key takeaway was that communities need support for doing what they were already doing, but bigger and better, and they can benefit from finding diverse pathways for accessing funding.
There may also be a role for intermediaries to make connections between grassroots and national levels. Most importantly, communities want to have a sense of ownership and control over what they’re doing.
The discussions
Jumping off from the provocations, participants highlighted several issues. For instance, when it comes to selecting designated representatives and intermediaries for a community, who decides who those people are, who has access, and who gets left out. Often, people that become intermediaries and representatives come from privilege within their own communities. This complicates the question of representation.
Participants noted that relying on existing structures of representation may make it easier to engage with a community, but this also risks reinforcing and perpetuating existing structural asymmetries. How do we (re)define and engage with communities in ways that make sure to include the underrepresented?
On the question of representation, a participant expressed that perfection shouldn’t be the enemy of the good. There may be value in first creating representation, knowing that it will be flawed, and then working towards addressing those flaws. We may put up with small instances of exclusion, but must find ways to address more systemic forms of exclusion. Addressing issues of representation in these communities may be a more tractable challenge than the complete lack of representation and contestation we currently have against exploitation from a small number of Big Tech companies.
There was also interest in pursuing a plurality of approaches, especially when dealing with such deep rooted structures around data and their pernicious effects. The question of participation in data operates at many levels, so it’s not currently clear what approaches will be successful. As a consequence, there is value in pursuing a variety of methods. Existing institutions can do more to empower people with respect to their data, but there is also space, and interest, in experimentation with new models and institutions - even at the level of a single street. This makes a plurality of approaches, including nascent and imperfect ones, important.
A participant questioned how this plurality approach could be written into policy and law. The challenge is that the looser the policy approach is, or the less guidance people have, the more likely it is for there to be unsuccessful or even harmful outcomes.
One response was that we may need to accept that regulation, though often needed, isn’t the only tool. Given the plurality of problems, we need more participation, experimentation, pilots, and not all of that can always be supported by regulation.
Participants noted finally that drawing from existing regulatory traditions might bring guidance - for instance, consumer protection frameworks echo Dewey’s conception of publics that are called into being in order to resist something. There is also opportunity to retool existing approaches, for instance, the use of consumer law in Brazil to challenge facial recognition. Brazil also has instances of Courts providing remedy to communities on questions of environmental protection, which could also be a model to draw from when thinking about community recognition.
Bringing the discussion to a close, Jeni summarised some key takeaways from the session.
- There can be a wide variety of ways of thinking about communities, particularly those that spring out of responses to specific problems.
- When it comes to representation, perfect shouldn’t be the enemy of the good
- We should explore and encourage a plurality of tactics in pursuit of the ultimate goal.