Data-care-learning as a three-body problem

Theorising the data-care-learning nexus

In September last year, Abby and I attended 4S Seattle to lead the team’s panel ‘Data, Care, and Learning in Datafied Worlds’ (read our roundup of the panel here). The opening talk that we gave, ‘Theorising the data-care-learning nexus’, was our first attempt to articulate as a group our commitment to thinking about data, care, and learning, with and through each other, extending and troubling existing scholarship to provide new registers for interpretating data-driven healthcare. To prepare the talk, we spent much of the summer returning to readings we had discussed as a group and poring over our notes from those meetings. Developing our emerging understandings of the data-care-learning nexus led us to a series of reflections, three of which felt central to our project.

Firstly, we all felt a firmer grasp on what we each meant and valued about the concepts of ‘data’ and ‘care’ than we did with regard to ‘learning’. It was easy for us to agree that we could never be satisfied with any conception of learning as simply the transmission of knowledge; it was much more difficult to articulate a conception of what we do believe it to consist of. Theories of learning proliferate and contradict each other across and beyond STS. How can we tell when learning is taking place, or whether something has been learned? Should we focus on cognitive theories, pedagogical methods, or look for material traces of changes in knowledge and practice? For now, we take refuge in Tim Ingold’s account of learning as processual, embodied, and environed – a practice of attention, not of transmission [1].

Secondly, our engagement with the literature around data, care, and learning led us to conclude that it was vanishingly rare for the three concepts to have been explored together with equal billing: many of the texts we read were dual investigations (of data and care; data and learning; or care and learning), though in most the remaining concept from our nexus had a nascent or spectral presence, lying underexamined under the surface and suggesting additional questions.

And thirdly, we believed that many investigations employed their conceptual pairings with an explicit or implicit directionality: people wrote about caring for data; learning to care; data about care. Indeed, such framings are often present in the titles of research projects, articles, and books. Yet to frame one’s work in such a way is to preordain one’s findings by defining in advance part of the nature of the relationship.

By positing, with the data-care-learning nexus, that these three things cannot and should not be thought apart, we address the first two of these points. A further commitment in the form of a resistance to presupposing directionality seemed to address the second. However, I still felt frustrated. In order to deploy the nexus as an analytical tool across different sites I felt a deep need to characterise it somehow – even as we agreed that doing so could risk closing down our field of inquiry in some way.

Struggling to articulate my thoughts, I stepped back from close readings of texts in STS and drew on other interests to explain how I see the data-care-learning nexus.

A poetic reading of celestial mechanics

You may or may not be familiar with the three-body problem, a centuries-old problem in physics whose first articulation was in Isaac Newton’s Principia Mathematica (1687). Using Newton’s laws of motion and of gravitation, it is possible to calculate the movement of two bodies orbiting each other given their initial positions and velocities. Introduce a third orbiting body, however, and the problem becomes incredibly complex – so complex that there is no general mathematical solution. Spatially, the bodies in a two-body problem will always describe an ellipse. In a three-body problem, most initial conditions produce chaotic movements.

(Fans of science fiction will recognise this as the problem faced by the Trisolarans in the first book of Liu Cixin’s Remembrance of Earth's Past trilogy: originating on a planet orbiting three stars, Trisolaran civilisation is hampered (and frequently destroyed) by the irregular and unpredictable climate governed by their homeworld’s chaotic movement around and between its three suns.)

Over time, it has been shown that there are some specific solutions to the three-body problem; that is, there are certain initial conditions that produce patterns of movement that are repeating, and therefore predictable. When expressed visually, these periodic solutions are hypnotic and even beautiful [2], recalling the name given by Carles Simó to special cases where the three bodies all follow the same path: gravitational choreographies [3].  Perhaps the most beautiful visualisation is this one from Pere Rosselló. (Only the solutions on the right of the second row down and in the middle of the third row are true choreographies, where all bodies follow the same path).

Begging forgiveness rather than permission from our colleagues in mathematics and physics, we can express the data-care-learning nexus through the language and spatiality of the three-body problem. If the nexus is a three-body problem, then data, care, and learning are locked in a dance within each other’s gravitational pull. Their interactions may appear more or less chaotic, and at times their movements will resolve into regular and recognisable patterns (perhaps these differences can account for the fragility or stability of the worlds that data-care-learning sustain). Even a stable pattern will appear different at different points in time. The patterns and movements that data-care-learning describe will vary by shape, speed, and complexity. Sometimes one concept will take a straightforward path while the others swirl frantically around it; sometimes one will cover a greater area while the others follow a smaller orbit; sometimes two will be distant from each other while the third moves between them. While choreography is reserved for a specific set of solutions to the three-body problem, the term reminds us that in all configurations of data-care-learning there is work being done to produce, manage, and organise the movement that takes place.

Resonances

I find thinking about the data-care-learning nexus as a three-body problem to be a rich source of metaphors that resonate with STS thinking; the three-body problem was (and is), for me, ‘good to think with’ for the sharpening of this analytical tool. Crucially, thinking in this way allows me to articulate that we expect data-care-learning to relate to each other but that we do not assume the shape of those relations.

Bringing data-care-learning into dynamic relation to each other serves to extend and trouble existing scholarship that explores simpler dyadic relations, and to resist the assumptions of directionality that sit within accounts of caring for data; learning to care; or data about care. The determining effect of the initial conditions remind us, too, that if we weight our analysis differently – for example by undertheorising learning as opposed to data and care – the gravitational pattern of the nexus will change and the outcome of our analysis will be very different.

Thinking in this way clarifies, for me, that our intention in DARE is to trace how data, care, and learning move around together, in hope of capturing patterns of interrelation that have yet to be attended to. 

References

[1] See Ingold, T., (2018), Anthropology and/as Education, Routledge, and Ingold, T., (2022) ‘Evolution without Inheritance: Steps to an Ecology of Learning’, Current Anthropology 2022 Vol. 63 Issue 25

[2] See e.g.: https://analyticphysics.com/Gravitational%20Choreographies/Sim%C3%B3's%20Three-Body%20Choreographies%20in%20Action.htm (check the ‘show orbit’ box); http://three-body.ipb.ac.rs/

[3] https://www.maia.ub.es/dsg/3body.html 

Image credit: Twenty examples of special periodic solutions to the three-body problem
By Perosello - Uploaded by Author, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=133294338