Can a model care?

In the third episode of the series “Caring matters in water governance”, Rozemarijn ter Horst, Bruno Bonté and Jean-Philippe Venot address critical questions about modelling that are seldom raised in scientific and professional debates in the water sector.

For the workshop “Caring matters in knowing water”, we – Bruno Bonté and Rozemarijn ter Horst – were invited by Margreet Zwarteveen to organise a session on critical modelling. Bruno is a computer modeller working with the French National Research Institute of Science and Technology for Environment and Agriculture – INRAE – in the G-EAU research Unit. He just had an experience of working with a social scientist on his practices of modelling on a project in Algeria, specifically to understand jointly how the model represented humans and nature and to what effect. Rozemarijn ter Horst is a social scientist, currently working with the Water Resources Management Group at Wageningen University, and previously at IHE Delft. She is working on her PhD on how mathematical water models influence and interact with water governance and management, based on case studies in India, Spain and Palestine.

Critical modelling

It took us time to find a suitable approach to the session. Models embody world views and disciplined ways of assessing the world (which are very different for both of us). Margreet connected us specifically for this session, and it took us time to understand how the other saw the world, and what methods both of us used to understand this world. We learnt that interdisciplinary cooperation takes time, curiosity and respect. We decided to ask the participants to explore different case studies in which models played a central role, and to ask them to share elements that they found crucial to be included in a critical and reflexive analysis. We explicitly invited people to share their thoughts about why their suggestion is important, from what kind of perspective or discipline they suggested it, and with what expectations. To facilitate this interdisciplinary process, we decided to start with a simple representation of what a modelling process is, used by Bruno in his PhD. This Minsky Triad represents the relations of the modeller (also named “observer”) to the model and to what is being modeled. We slightly restructured it to the set-up below, introducing the model-users who may be different from the Modeller.

Case studies

We invited four of the workshop participants to present a case study in which a modelling process is central to think with. We then split the participants in four groups (one per case study) and asked them to focus on the relations between the elements of the Minsky triad, and the questions the groups had on those relations. We briefly present here the case studies, together with the outcomes of the discussions during the workshop.

Jean-Philippe Venot (G-EAU research Unit in Montpellier) presented the participatory modeling work he developed with colleagues in the Cambodian floodplains, notably highlighting the fact that such type of modeling could be explorative as opposed to explanatory, granting the modelers allow themselves to be surprised by the participants and account for their inputs. In the participatory modelling process described by Jean-Philippe, the model was a Role Playing Game (RPG) which is a common kind of model in such participatory modelling approach. However, within the RPG, mathematical models were used at some points.

The discussion focused on the limitations of modelling, and on how often the exercise sets clear boundaries to think within. It makes it more challenging to think outside of this box, to maintain a plurality of perspectives, to connect with place and culture, and to imagine what cannot be predicted based on available data. The group compared mathematical modelling and RPG modeling approaches that they called gaming approaches, taking modelling as a more rigid approach, and gaming as a way to play with “reversing, constructing, inverting and switching” to explore and imagine different futures.

Ankita Prayag and Andres Verzijl (IHE Delft) presented the case of modelling surface and groundwater in the Indian Cauvery Delta, with the aim to (re)connect science, policy makers and grassroots initiatives to discuss and promote sustainable delta futures. Science and policy has long focused on surface water thinking and the data and knowledge tools that are used have been incomplete and contested. This is why citizen science and remote sensing that democratise data, and situated science are central in the project: these tools are used to develop a surface-groundwater model, in close relation with farmers perceptions and experiences.

The discussions revolved around two main issues. Firstly, what can we expect of a model, and are our expectations not too high? Secondly, to what extent does a model help to ‘make’ the reality it models? Taking this into account, we should accept a model is never comprehensive, and that models can be a great tool to bridge farmers and policy makers – at least in the case presented – when used with care and humility.

Bruno Bonté (G-EAU research Unit in Montpellier) presented the agent-based model of a social and hydrological system developed for a thesis in geography (by Idda Salem) to understand the dynamics of the oasis traditional agriculture in front of social and environmental changes that occurred in the modern Algerian state. He shared experiences of analysing the modelling process from both from a computer and a social science perspective, showing how this dialogue lead to the identification of a common research object (hybrid water: social and physical), of similar research questions (what contributions do agent-based models make to the understanding of the relationship between water and society?) and of the meaning given to research (producing alternative narratives to dominant discourses).

The discussion focused on what models represent, and for who, and emphasised the political aspects of models. Models do influence the reality they model and by presenting a certain vision, as well in this case, as a vision of the past to the model users. The model and the modelling process are truth making, world making, and heuristic support, and as they are political tools they can also be used in an inclusive way, as a boundary object that can serve different people with different needs. In this perspective of a model as a political tool, we note two original points that raised out from the discussions about the things that could (or should?!) be observed during the modelling process: First that the human or non-human that can be affected by the model should be at least listed if not consulted, second that the message conveyed by the model should at least be made explicit and, if explicit, could even lead the modelling process.

Jonatan Godinez Madrigal presented the case study that was central in his PhD research: the Zapotillo dam and the related water transfer project in Mexico (Godinez Madrigal et al., 2020; 2022). The Zapotillo project aims at augmenting urban water supply to Guadalajara and León, two large cities in western Mexico, but a social and legal conflict has stalled the project until today and is projected to be retrofitted to decrease its negative effects to three villages within the reservoir area. To examine the role of modelling in the conflict, he mobilized the Greek concept of Phronesis as the most important intellectual virtue dealing with context, practice, experience, common sense, intuition, and practical wisdom,  and analyzed the design and use of water resources models produced by key actors aiming to resolve the conflict. While the use of models is a proven method to construct future scenarios and test different strategies, the parameterization of scenarios and their results are influenced by the knowledge and/or interests of actors behind the model. He found that in the Zapotillo case, scenarios reflected the interests and strategies of governmental actors, resulting in increased distrust of the grassroots movement. Aware of this pitfall, Jonatan repurposed the same model through a participatory modelling process to be able to develop scenarios that also reflected the interests and perspective of the grassroots movement, which allowed to discuss the politics, uncertainties and ambiguities of all scenarios. This model was discussed in the workshop.

The discussion focused on the modelling process, questioning who decides what is being modelled and what not, and for what purpose. What is the role of the modeller? Is it to answer questions posed by decision makers, and/or is it to raise questions? What allows the modeller to work in isolation, outside of politics, while the work is political? And what are the ways for non-modellers to understand and challenge a model? A main question that was raised is ‘who is accountable’, especially as modeling exercises are presented as neutral scientific exercises.

Reflections

In the three hours we had for the workshop, the Minksy Triad provided a helpful framework to guide discussions on hydrological models, their influence, and on what elements to focus to promote critical and reflexive modelling. Based on the Triad, we asked participants to reflect on the natural and social elements that were being modelled, how this is modelled, who is modelling, and who is using the model, grounded in the four different case studies.

A limitation of the exercise was that we excluded a clear category for those who are affected by the development and the use of the model. In scientific analyses of models, this is largely excluded. When working on models in a critical way, including those that may be – or are – affected is crucial. This issue was raised in all groups during the discussion. We could also relate the work on critical modeling to the workshop we had on caring as an approach to water governance. There are many conceptualisations of caring, and here we want to highlight one. When someone cares and wants to make something better, or ‘well’, it is based on a certain understanding of what better and well entail. It is often also the aim of a model, whereby the modeller, commissioner or funder has an idea of what is “well” for the object modelled (a river for instance, or the aquatic life, but also water users). Making these ideas explicit and discussed in the process of modelling social and hydrological systems is essential for a critical and caring way of modelling.

Elements that were very similar across the four discussions are the question on the influence of models. In what ways do they affect the object they are modelling? And in what ways does it affect the model users and the modeller him/herself? Does it ‘make’ a certain reality of an object? And if a model is never comprehensive and shows a certain view on reality, what does that mean for the decision making process? To what extent could the modeling process, and the reality it conveys be transparent, explicit, and part of the decision making process or of the research exercise? Is there space for a counter-modelling exercise, or counter-discourse? If a model excludes certain issues or is not a good tool to represent culture or place for instance, how then to design a process that allows space for this?

All these questions sparked fruitful discussions between modellers and non-modellers, as well as between natural and social scientists on how models are constructed and made, and on what impact do they have. In the scientific literature, very little has been written on this topic. Nor are discussions on the impact of modelling and accountability common – yet – in the water sector. We are therefore eager to continue the conversation and if you are interested in this topic and would like to join our efforts, get in touch!

---

---