The Water Dissensus – A Water Alternatives Forum
Models do not think
by Jonatan Godinez Madrigal, Rozemarijn ter Horst, Bich Tran, Rossella Alba
We are a group of young scholars working together in the Constructive Advanced Thinking Programme framework. We asked and received funding to unpack and discuss 'Controversial tools: researching modelling practices in water governance'.
Models do not think. But they easily become substitutes for thinking. In water science, computer-based models are used as intellectual tools that estimate knowledge about coupled human-water systems. But what is the relationship between numerical models and other human mental faculties like thinking? Should numerical models also have the explicit role of eliciting thinking? Is that even desirable? We have encountered many instances where models are fetishized and expected to provide clear and unproblematic solutions to complex problems, leading humans to foreclose their thinking faculty. This obscures imbalanced power relations, erases people and the natural world, and ignores uncertainties. As a result, it is difficult to achieve a deeper understanding of the social and natural world in decision-making.
In a world that prioritizes STEM disciplines (Science, Technology, Engineering, and Mathematics) in tackling global issues, there are risks and unintended consequences of using models exclusively as intellectual tools. Martin Heidegger's axiom, "science does not think" (1993), challenges the assumption that numerical modelling engenders a profound understanding. Computer-based models in water management provide data-driven solutions, often prioritizing efficiency and optimization. However, Heidegger's assertion prompts us to critically examine the depth of thought inherent in models. Can they genuinely engage with the plurality of meanings and valuations of water, or do they merely provide utilitarian solutions divorced from a more profound understanding?This is not an idle question as "Thinking, no doubt, plays an enormous role in every scientific enterprise, but it is the role of a means to an end; the end is determined by a decision about what is worthwhile knowing, and this decision cannot be scientific" (Arendt, 1981).
What does this mean for the way we engage and develop models? The dominant understanding is that science and engineering are enough to develop good models. However, for some scientists, the modelling process "requires imagination, inspiration, creativity, ingenuity, experience and, skill…Hydrology is an art as much as it is science and engineering" (Savenije, 2009). Therefore, we claim that water modelling as a process needs to change, lest its huge potential be wasted by foreclosing thinking.
Models are useful for computing complex processes that are difficult for humans to calculate. They can help explore the past, present, and future and consolidate hypotheses about the world. However, using them as thinking tools can be challenging because practitioners often confuse them with reality when, in fact, they are only simplifications of reality. Reductionist modelling practices view water as a unidimensional (economic) resource, a 'standing reserve', to be optimized for efficiency divorced from its cultural, ecological, and existential significance, and excluding people and alternative views. For example, in the Zapotillo conflict in Mexico, a high-level international consultancy was hired to develop a hydrological model using WEAP software to settle knowledge controversies around a controversial urban water supply dam. They embarked on a two-year modelling process that focused on the government's vision of dam optimization. They disregarded less impactful alternatives while ostracizing key stakeholders from the modelling process, including the affected communities. The result was a model that lacked legitimacy and did not help transform the conflict (Godinez Madrigal et al., 2020). However, no one seemed accountable for this fiasco. The responsibility was diffused between the model itself, the modellers, and the model commissioners.
To understand this diffusion of accountability, it is necessary to unpack the problematic traits of models as intellectual tools. These include:
- Humans have delegated to models the intellectual faculty to compute and calculate with great capacity. The role of thinking is also present in the modelling process, but it is often obscured when using off-the-shelf, one-size-fits-all established modelling software. This is troublesome because the thinking faculty is expansive and driven by imagination and a deeper engagement with the meaning and significance of phenomena (Arendt, 1971). Established software may not allow for a deeper understanding of reality. Therefore, a strong argument exists for developing tailored and situated models that force the modellers to think of reality beyond pre-established calculations.
- Modelling has an often-unrecognized tension between searching for truth and meaning. Relying solely on models can limit discussions between different types of knowledge with non-modelling actors and foreclose thinking while implementing controversial solutions derived from models. It can also impose a particular worldview on water. "The model hath spoken" has often been used to justify implementing controversial decisions or imposing a particular worldview on water. Arendt warns us that knowledge is a world-building enterprise, as material as building houses.
- Proprietary software can be problematic as it may change its capacities, restrict access, become obsolete, or degrade in quality due to profit-seeking (see 'Enshittification'). This limits human intervention in discussions about the software and its results, particularly with the emergence of AI and machine learning. These models can predict future outcomes without disclosing their inner workings, leading to a loss of knowledge and independent thinking.
If thinking relates to memory as a function to make sense of the past and the will to imagine our common future, then could computer-based models, instead of foreclosing thought, be capable of fostering, as thinking processes, the potentiality of humans to enrich (re)-interpretation of the past, make sense of the present through a plurality of perspectives, and open up the decision space to foster a creative imagination for the future?
There are great examples of using models in ways that potentiate human thinking. In a non-exhaustive list, we can identify alternative modelling processes focusing on thinking (ter Horst et al., 2023): counter-modelling; exposing black boxing; explicitly showcasing the development process of modelling and how modelling decisions affect outcomes; openly questioning modelling decisions and assumptions behind them; foregrounding power relations; calling for particular ethics; and focusing on the process instead of the tool (i.e., companion modelling). So, if there are so many alternative manners of modelling, then why are we still not doing it differently?
We have highlighted the pitfalls and risks of many current modelling undertakings. We are not arguing for doing away with numerical and computer-based models in water management but to rethink their development and use. We claim that an answer lies especially with fostering thinking. We aim to stir and join an ongoing conversation about what models do, their potentiality (all possible ways they can be used), and their affordance (the restrictions inherent in how they are used) in a world where thinking is increasingly rare but the search for truth and meaning is inescapable. However, this search cannot be done without simultaneously considering the uncertainties inherent in models and the power asymmetries intrinsic to different value systems involved in water systems when developing and using models. We must reform the currently dominant development pathway dependent on numerical and computer-based models that are foreclosing thought and instead use models as pathways to expansive human thinking and its collective flourishment.
-----------------------------------------------------------------------------
Rossella Alba studies socio-ecological transformations and inequalities taking as a point of reference infrastructural relations and the governance of resources, and more particularly water. She works in an interdisciplinary manner by combining critical social science research with natural science approaches. ITHESys, Humboldt University, Germany
Rozemarijn ter Horst has been working as a lecturer and doctoral student in the Water Resources Management Group at Wageningen University since October 2020. She studies how quantitative models influence water management and governance. To show how models are political, Rozemarijn ter Horst focuses on case studies in which models are introduced in the hope of reducing or resolving conflicts over shared water resources. In these case studies, she explores how data and technologies play a role in identifying generally accepted development options and how and when contestations (can) take place in this process. The case studies include the federal Kaveri (or Cauvery) River, shared by Kerala, Karnata, Tamil Nadu and the Union Territory of Puducherri, as well as the aquifers shared between Israel and Palestine. Her research draws on science and technology studies and constructivist theories, and she seeks to work closely with those who develop and implement the models. Before working with Wageningen University, Rozemarijn ter Horst worked at IHE Delft on water diplomacy, and remains affiliated as a visiting researcher on cross-border water governance.
Tran Bich is currently a doctoral student at IHE - Delft Institute for Water Education, and at the Technical University of Delft since September 2021. She studies uncertainties in evapotranspiration derived from satellite data and how the implication of these uncertainties in the evaluation of water resources. Prior to her PhD, Tran Bich worked as a research assistant at IHE Delft on the Water Accounting Plus (WA+) framework, which uses open access earth observation data and spatially distributed hydrological models to the assessment of water resources at the basin level. She has conducted several studies on water management and is interested in the uncertainties of these data, models and assessments of water resources.
Jonatan Godinez-Madrigal is a postdoctoral researcher on global transitions in water distribution regimes at IHE Delft. In his research, Jonatan bridges the dichotomy between objective, technical expertise and the more subjective socio-political expertise needed to understand complex socio-ecological issues. By combining mixed methods, such as longitudinal, interdisciplinary, and transdisciplinary research, he is able to simultaneously study the historical, social, and biophysical dimensions of water-related conflicts and socio-technical transitions.References
Arendt, H. (1971). The life of the mind. New York: HMW.
Godinez-Madrigal, J., Van Cauwenbergh, N., & van der Zaag, P. (2020). Unraveling intractable water conflicts: the entanglement of science and politics in decision-making on large hydraulic infrastructure. Hydrology and Earth System Sciences, 24(10), 4903-4921.
Heidegger, M. (1993). Basic Writings. San Francisco: HarperCollins Publishers.
Savenije, H. H. (2009). HESS Opinions "The art of hydrology". Hydrology and Earth System Sciences 13(2), 157-161, https://doi.org/10.5194/hess-13-157-2009, 2009.
ter Horst, R., Alba, R., Vos, J., Rusca, M., Godinez-Madrigal, J., Babel, L. V., Veldwisch, G., Venot, J., Bonté, B., Walker, D.W., & Krueger, T. (2023). Making a case for power-sensitive water modelling: a literature review. Hydrol. Earth Syst. Sci. Discuss. [preprint], https://doi.org/10.5194/hess-2023-164 in review.
Comments 39
Super interesting read thanks a lot for exposing these issues
May I ask what is your background and profession?
Models hold the potential to make water managers lazy and dumb. Lazy to conduct multi-disciplinary research and dumb decisions based on one-size-fits-all research methodologies. Models are but one tool to use in water management issues. With the fast-paced progression of artificial intelligence, modelling such water management decisions could compound the situation.
Thank you for sharing your thoughts Richard. We also think, as you write, that engagement with models and technology should be done consciously, to avoid one-size-fits-all approaches and becoming lazy and dumb. It also inspires us to more consciously engage with concepts and tools in general as well.
The various phenomena that affect natural resources, such as groundwater, are very complex. Indeed, the natural environment is very heterogeneous. The models used are tools that allow us to have a simplified vision of all phenomena, but in no case can not completely replace reality, because this simplification is often the result of not taking into account some parameters, that are sometimes very important in understanding the functioning of the modelled phenomenon or phenomena.
In many published studies in hydrogeology-hydrology, even if the main parameters introduced in the chosen model are well estimated, their inclusion in this empirical model remains a difficult task. This difficulty is increased when there are aquifer formations with some lithostratigraphic heterogeneities difficult to quantify, especially for example, in karst environments (carbonate formations with crack porosity), where it is often difficult to accurately assess some of the flow parameters, which could justify the difficulties encountered when validating certain models with field data (overestimation or sub-estimation of some data, etc.).
To conclude, according to me, the use of models is now essential for a good knowledge, management and protection of water resource but it can not be dissociated from human thought, which will be complementary. Without the human thinking, the model could not be interpreted correctly
Thank you Lila, for your thoughts and sharing challenges related to groundwater modelling in particular. It reminds me of an article by Kroepsch (Kroepsch, A. C. (2018). Groundwater Modeling and Governance: Contesting and Building (Sub)Surface Worlds in Colorado’s Northern San Juan Basin.
Engaging Science, Technology, and Society, 4, 43–66.), in which she questions how decisions are made based on partial information. She describes how a decision is made to pump up more and go for maximization, instead of less and conservation. I am curious what your experiences are in terms of the relation between knowing/not knowing groundwater, and consequently how we relate to it.
Thanks for an interesting piece! I can certainly agree with many of the points raised, which I will not elaborate below.
Models can indeed, sometimes, and with the right people, be invaluable to water resources management and to probe all sorts of hypothesis about the inter-connectivity of land and water resources, their use, and possible future scenarios. More often than not, however, models are not used in appropriate ways. And it is fascinating to see people in their search for colorful images forego their critical faculties; and that trust results that are at times physically impossible (as I have seen in a fancy model developed by colleagues at one of my previous employers, with strong push-back when confronted with the reality that model results outsmarted the laws of physics); plain wrong (because there is little understanding on the multitude of assumptions that need to be understood); and/or really just meaningless. At the same time, modelling has become so much part of the daily routine, that one is questioned for sanity, if one suggests that within the confines of data availability and within the needs of questions asked, any Excel spreadsheet calculation (with a few simple formulas), likely generates results that are more transparent and of equally good (or bad) accuracy. Just not as fanciful looking. Unfortunately, this has become the norm in many places I have worked, and people often rather spend money on remote modellers than hiring people that get their feet dirty.
That is not to say that models and modellers have not done some amazing work. And I personally got to know some groups and people that I personally trust when detailed modelling of complex phenomena is required.
A couple of specific remarks:
(1) I was surprised that not much was said about the question of data (quality, choice, etc.) and about the selection of model tools. Both are I believe critical parts; also as they relate to the politics of things and how complexity and uncertainty is masked, and how results suggest desirable results that do not hold up to even basic critical inquiry. The question of data is the basic principle of "garbage in - garbage out", but made much more challenging when a multitude of scenarios, projections, and assumptions (including from various global databases and regional downscaling) enter into the equation. Testing the appropriateness of these datasets is hardly ever done. Doing so requires skills/money/motivation, and if done can lead to results that question a lot of assumptions and results that were previously developed with poorly developed models and underwrite many key policy document and sector strategies (not something that is all that enticing to people that are engaged to deliver on project/program specific assignments). I was lucky to be part of one such exercise as part of a project preparation I lead and found it intellectually enriching (https://doi.org/10.5194/egusphere-egu23-11342). On model choice, books can be written, and so I will leave it at that.
2) Having to agree with most of your claims and arguments (and seeing many of these issues play out in the open wild), I still wonder what Heidegger's axiom, "science does not think" provides as an enriching entry-point for your inquiry. Leaving the question of whether it is an axiom altogether aside, it appears to me to be rather trivial in the context of modelling (up to this point in time); just as trivial as to acknowledge that models as a result of their atomistic conception and design, will always just be able to answer a fraction of what matters in water resources management and related societal matters. That does not do away that models too easily take center stage and serve as the ultimate justification to push certain investments. But, of course, in quite a number of cases, models are used just the way they were designed for: a tool to test some hypothesis that people can ponder about, argue about, and take into account alongside many other factors that are at least equally and often of more critical importance (aside from questions of physical limits should they arise). I just bring this up, because I tend to agree with those that see Heidegger as poisonous to critical research (as an example: https://doi.org/10.1016/j.polgeo.2020.102159). And I am somewhat convinced that you could reach the same findings and insights, and maybe even more nuanced conclusions, without relying on a scholar that in my book is beyond repair, and even entering into the trenches of finding nuanced language for a Nazi extraordinaire (https://yalebooks.yale.edu/book/9780300233186/heidegger-in-ruins/).
In anycase: good luck with your research and studies!
Reading your piece and typing a quick response, made my evening trainride that much faster.
Thanks for sharing your experience and thoughts Philippe,
I think there is a common bias in all humans to fall for flashy, shiny, blingy things that are perhaps all but empty carcasses or mere shells, but in fact, some simple tools like an Excel sheet can provide you with much deeper insights than even complex models. This debate reminded me of an interesting paper (Venot et al., 2022) of what happens when we let go of the pretence that models can represent physical reality and concentrate on what kind of relationships between people, imaginaries and objects models can enable to resurface: https://www.sciencedirect.com/science/article/pii/S0305750X21003946
Discussing this text with a co-author, we thought that some platonic legacy is at fault, when, following Plato, Ideas that are beautiful are also True and Just. So we're easily misled by fancy interfaces, assuming that something so beautiful must also be True.
Regarding the two issues, you disagree with the text. On the first issue, I think you're right about the need to simultaneously discuss models, data and modelling choices. On the latter, we wanted to provide a glimpse of the varied cornucopia of modelling practices available that can embrace thinking as central instead of foreclosing it. On the former issue, gathering data, I think it deserves its own Water Dissensus contribution only to talk about it, for instance, how the need for more data can be almost a strategy to postpone making decisions in politically contested terrains, as well as excluding certain kinds of knowledge (as discussed in https://www.water-alternatives.org/index.php/alldoc/articles/vol11/v11issue1/425-a11-1-8/file). I think there is a very needed debate on how, for instance, citizen science can contribute to co-producing data and knowledge to fill gaps in physical and social data. For reasons of space, we needed to constrain our argument only about modelling practices, but we could indeed at least mention the topic.
On the second issue, we used the word axiom, understood as "An axiom, postulate, or assumption is a statement that is taken to be true, to serve as a premise or starting point for further reasoning and arguments." On that note, we can call 'Science does not think' an axiom. From there, I personally don't think this axiom is trivial in the subject of water management. For instance, using the word Water Resources, is already a certain kind of wording of a concept that implies what Heidegger criticises as thinking of nature as a 'standing reserve'. A sort of assumed concept that carries within it an intention, which is that water should be used as an (economic) resource. Thinking is what provides meaning to the scientific endeavour, so when we just assume these concepts as natural, then we also limit out thought, in a way that Wittgenstein argued, we cannot think beyond our language, so if our language is contrained by these concpets, we cannot think further than considering water as a resource. Furthermore, I read the document that you shared ('Arcane geopolitics'), and although they bash Heidegger, in the Coda section they acknowledge that "resist [their] gnostic conclusions while taking up what is philosophically insightful in [their] work”, although such insights can “survive if and only if they remain available for a truly critical appropriation”. We believe that is precisely what we did; we used a philosophical insight to elaborate a critical argument while resisting some of its gnostic conclusions. I was aware of Heidegger's problematic past, but I do not believe we should disqualify all of his corpus for that matter. Because 1) We certainly did not only use Heideggers's work to elaborate our argument, but as suggested, we used other author - Arendt to underpin our argument. In Arendt's work, it is acknowledged the mutual inspiration for their work, even with Heidegger's problematic past. B) I don't think that purging and cleaning our discussions from problematic authors is a way to go, lest we fall into the same ideological trap that many of these extreme right ideologies do. I would argue, following Zizek, that the best way to truly take power back from some of these authors, lest risking mystifying them, is to, as suggested by the 'Arcane geopolitics', to critically appropriate valuable insights they formulated.
All the best,
Jonatan
Having enjoyed this particular forum contribution, I would like to provide some last thoughts on your reply.
I will gladly admit, that calling the matter trivial, was probably overly polemic. Nothing about Heidegger is trivial (starting from reading his work, even in my native language german). But - and that is what I specifically meant - the title of this contribution "models do not think" vis-a-vis Heidegger's "science does not think", seem to be pretty common-sense: models do indeed not think (up to this point). They still do, what people tell them to do; even if this might change in the future.
Your association of parts of the discussion with work from (J-P) Venot et al. was curious to me. And I personally, having been unaware of this particular paper by him, could only associate it to some quite important linkages J-P and I have: we were both guided in our PhD research by Francois Molle; worked on a somewhat comparable set of questions; and I assume were having similar inspiring converstations with Francois. J-P, obviously became a successful researcher; while I worked for/with International Financial Institutions, in various roles, throughout my professional life.
On the question of data, there is little to add from my end. Many aspects of the wider questions you ask, are so nuanced that they would deserve their own dissensus forum contributions. And to me, it is testament to the fine balance all of you found, that the discussion that ensued was engaging and triggered different interesting response.
Heidegger, at the same time, will probably remain an issue, where we would both find it hard to come to a common understanding. And I have realized over the last couple of years, that as a result of Germany's and Austria's history (Austria being my home country), there is an understandable reluctance to receive Heidegger's work without taking his politics fully into account (even if just small parts of his philosophical output are appropriated). Reading Benedikt Korf (the author of the paper I referenced) or say Peter Staudenmaier (https://www.marquette.edu/history/directory/peter-staudenmaier.php), both Germans, invokes in me similar concerns and associations, that people from other countries may not necessarily be too worried about.
While prior to 2014, Heidegger's body of work and politics could arguably still be disentagled, the publication of his black notebooks, in my mind, changed the situation fundamentally. That is to say, the onous of justifying one's decision of making Heideggers work an entry to critical studies, now rests pretty firmly on those that conciously make this choice, rather than those critical of it. And indeed, if I recall correctly, there have been numerous critical geographers, political ecologists, and environmental researchers, that have distanced themselves from their earlier inspiration by the work of Heidegger.
To me, a reference to Heidegger simply invokes the same initital reluctance as to scholarly work that via Laclau/Mouffe appropriates some fragments of the work of Carl Schmitt (say the work of Eric Swyngedouw). I - as a layperson - can simply not reconcile any attractive features of the overall critique and worldview, those historical figures advocated. At the same time, I can certainly come up with reasins why a rising authoritarian superpower would find some of Carl Schmitt's positions rather attractive.
As a last point, and this being certainly a most difficult topic, your reference to Wittgenstein's famous quote truly hit me off-guard. For Wittgenstein (particularly his Tractatus Logico-Philosophicus, from which this quote is from) and Heidegger's philosophical work are so far appart, both being key influences and emplematic examples, for the split between the "analytical" philosophical tradition (dominant in english-speaking countries) and the "continental" philosphical tradition that dominated Europe. Again, a layperson (albeit curious to read about this things as a past-time extensively), I struggle to see a way to invoke both in the same argument.
But many thanks again to all of your to share this nice contribution, that made me reflect again a bit about these different topics.
First, sorry for my bad english, I wrote very quickly without rereading.
Thank you for your answer, I find it very interesting and much deeper. I share your analysis.
Indeed, there is much to say about the quality of the data, and especially about their reliability. The scientific data collection networks are not at all the same when you are in an African country or a European or Anglo-Saxon country, for various reasons that we can guess... I have always raised this problem with my students during their studies, especially for doctoral theses!
Dear Lila-Naima,
while my lenghty post was not specifically a response to your valuable contribution, I was certainly inspired to raise the example of analysis done in Bhutan, as this was (amongst other things) targeted at finding analytical evidence to support the widely held believe (particularly amongst local communities) that climate change is the main factor in the drying up of mountain springs (a critical water sources for multiple uses).
We particularly embarked on this work for 2 main reasons: (a) the climate fund targeted for project financing requires (or at least asks for) a solid climate rationale in an effort to ensure that limited climate finance is not spent predominantly on issues that are more the result of human induced change and/or mismanagement, but addresses past and future impacts of climate change. At the same time, (b) the received wisdom on the drying of springs is often associated with land use change, human development in watersheds, and other non-climatic drivers. And consquently the organiization I worked for at that time, was reluctant to finance significant upstream project development work under the assumption that the climate fund might outright reject our proposal without solid evidence. So we hired experts to help us understand the impacts of climate change on springsheds, and to isolate the climate impact from other possible reasons for the decline in spring discharge.
One can go through various IPCC reports, and a growing number of somewhat inconclusive scientific literature, to get a sense of the challenges of doing that. And in our quest to try a relatively simplified approach, we also spent time to organize a 3-day inception workshop, bringing together various modellers and subject matter experts, to agree on the best approaches, datasets, and many other things. As one often encounters, different people have often widely different preferences when it comes to this question. And after the first day of inconclusive discussions, my co-team lead (a highly skilled young women, that has by now taken over the leadership role for this project) and I, requested all the experts to discuss a methodology that could help us make an informed decision on which of all these different data inputs and model choices would be most appropriate for the context of Bhutan. The link just highlights one interesting bit of the analysis conducted; as the two global models we compared with other available options, are among the most widely used datasets for rainfal data in areas that lack a dense monitoring network and long-term data (such as many places in Africa and Asia).
I have often wondered what to tell younger colleagues on how to be careful with modelled results and the data that is used. And it is certainly important to raise awareness to start with. In the end, I have always encouraged them to get their feet dirty, go to the field, avoid getting into the bureaucract dealings of working in international organizations too early, see various places with their own eyes, etc.; in order for them to built a strong understanding what certain numbers and results actually mean when one sees them in reality (and not merely on a computer screen). Many of the most basic errors, that I very often see (and I am not talking about scientific studies, but the application of modelling in 99% of cases where models are used by experts, companies and government agencies), could well be avoided, if the people that crunch the numbers, had a better sense of relating findings to images in the real world.
Lastly: your english and contribution was absolutely perfect, and there is absolutely no need to ever apologies!
All the best in your work
Philippe
We often differentiate between data and model simulations, but what is data? It's important to recognize that the collection of many data types relies heavily on models as well. Consider hydrology as an example: When we collect river discharge data, we're essentially modeling it based on water level readings at specific points along the river or weirs. In other words, the rating curve is a model of the stage–flow relationship. Similarly, measuring rainfall involves collecting readings from tipping buckets and then employing geostatistical models for interpolation. Soil moisture probes typically measure the electrical conductivity (EC) or capacitance of the soil, and then derive soil moisture by modeling the relationship between the observable EC and the unknown soil moisture. All these measurement techniques involve models with underlying assumptions and simplifications. Other sources of data such as remote sensing and climate reanalyses are no exceptions (Tran et al., 2023).
I agree with Philippe Floch and Lila-Naima on the importance of questioning the data. The modelers of a river basin model or a water allocation model should realize that they are, indeed, using outputs of various observation models and it is important to incorporate sensitivity and uncertainty analysis in modeling. The uncertainty of model results is an inseparable compound of many uncertainty sources (inputs, parameters, model structure, ...). Data uncertainty has implications not only for hydrological modeling but also for process understanding and water management (McMillan et al., 2018). Uncertainty studies provide us with an opportunity to reflect on the reliability of models (Kruger and Alba, 2022), prompting us to engage our thinking faculty rather than relying solely on the models themselves.
Thank you for this interesting and timely discussion.
Models do not think. Indeed, it is people who think. Sometimes people can think through models. Modeling is like a language, like mathematics, that enable humans to think clearly, and communicate those thoughts to other people. In terms of language, Ludwig Wittgenstein once said "The limits of my language are the limits of my mind. What I know is what I have words for".
To a large extent, this also applies to people like me who develops and uses models to understand the complex world around us. In my case, I will not generalize, I use models as tools (just like I use mathematics as a tool) to wade through information that is initially complex and confusing to make some sense. I am not wedded to my models and I do not allow myself to become slave to my models. I am more focused on using the models to solve problems or answer questions. Wittgenstein's quote, in my opinion, also applies to models and modeling. Some of the problems we face in reality, e.g., coupled human-water systems are so complex that framing the problems through the lens of models may be the way to open our minds to hidden mechanisms and processes. This has happened to me in my sociohydrology work: models have opened my mind to things that are happening in the real world that I did not know existed before.
However, my models are my own creation. They are fallible, just like we as people are fallible, and so the models we develop or use may be right for the wrong reasons. They may work in some circumstances and not in others. So the danger is when we become slaves to the models and look at the world through our jaundiced eyes. Models currently in vogue tend to reflect dominant views in science and in society, and when you are embedded in the system, we may not realize that there are alternative ways of looking at the world.
The way to address this problem is to have an open mind, to allow multiple perspectives to guide development of models. This is especially important in the age of machine learning and artificial intelligence where the impression is created that models (or computers) do think and can substitute for human thinking and learning. I am hoping that scientists continue to consider these as mere tools that can benefit their thinking and not become slaves. This is going to be a tough ask at the rate things are happening in the technology world.
I would like to add to this perspective where the model becomes a tool, and as mentioned in the piece, helps the thinking process. But even better if it becomes a collaboration tool, perhaps between scientists like in e-WaterCycle, but also between different stakeholders, like scientists, governments and citizens, to enable conversation and to help increase insight into the processes that are modelled. The fact is that the simplification that enables us to model these processes, also enables us to discuss the processes better with people that might not be experts in the topic, and that need the simplification to understand, but even more, to discuss the effects of interaction and adjustments to these processes. We can do this with participatory modelling or by indeed feeding the model with data collection together with the local communities. As long as we keep managing the expectations: this is a tool to try and mimic reality, it is not reality itself. Citizen science that feeds but also perhaps uses models again as tools, as mentioned earlier, is hopefully the way forward, and the process to both make use of the technological innovations but also keep us away from being enslaved.
Thank you for adding these thoughts Sandra, and for placing emphasis on the positive sides of simplifications. I fully agree that it is necessary and can be extremely helpful in facilitating inter- and trans-disciplinary exchanges. Perhaps the art, or challenge to each involved, is then to be mindful of what is lost or altered through these simplifications, and to see how (and what) complexities can be brought in again. This takes time and requires dedicated attention.
Thank you for putting together this piece, and sparking an interesting discussion. In keeping with the ‘water dissensus’ theme, I’m going to focus on the questions that this piece raised for me.
First, the piece is framed around thinking (or not thinking). There are many purposes for models, which are important for how we understand the politics of modelling. Models are used as tools to persuade, to illustrate, to align the management of variable environments with fixed legal standards. Models are used to explore, to test options, to evaluate scenarios. All of these could be considered as forms of ‘thinking’, but this is a different notion of thinking to Arendt’s. Is thinking synonymous with ‘seeking to understand’ in this piece, or is it defined in some broader way?
Second, on a related note, I want to ask if the goal of more tailored/situated models is always to develop a deeper understanding of reality? This seems to be suggested in the piece in several places. If our aim is to instead bring to prominence alternative meanings and values, might this be better achieved by building models which intentionally narrow in on some of the previously overlooked aspects of reality, at the expense of a more complete picture?
Third, the piece states that there is ‘a strong argument for developing tailored and situated models that force the modellers to think of reality beyond pre-established calculations”. Is this intended to be a recommendation for all modellers to do their own tailoring/situating? (a nice provocation, but challenging to make a reality) Or is the argument to have more tailored models available for modellers to pick up? If so, this feels like a technical ‘solution’ (the model will force the modeller) to an ethical problem.
This highlights what I see as a tension between attributing possibilities and outcomes to models themselves or to the people who use them (the cover photo illustrates this well). I feel a similar tension when I read that relying on models can limit discussion and also that people use models to justify particular worldviews. There is a different degree of human agency in these two scenarios. Perhaps there’s no simple answer to this tension, and it depends on the context, on the degree to which models are fetishised, or using in cunning ways. Thanks for the invitation to think about these issues - I’m completely in agreement that we need to explore better ways of doing water modelling.
For Arendt, there is a big difference between thinking and knowing (seeking to understand). In the text, we claim that mainstream modelling has concentrated on seeking to understand at the expense of reconciling the fact that there is a plurality of meanings (thinking), present and advanced by different groups inhabiting the same territories. Only marginal efforts have been made to do this, like with companion and participatory modelling.
In relation to your second and third questions, I'm not sure if there is a trade-off to strive for a more complete picture. Perhaps with a perspective of knowing, then there can be such trade-off. However, with a perspective of widening thinking instead of narrowing it, I think situated models can be more suitable for this. When one uses off-the-shelf models, we need to adapt our information and system understanding to the variables the software requires to run. This can skew the modelling process to make the model run instead of concentrating on what is at hand, the central question that is relevant to the actors. Modelling (especially the seemingly complicated ones) can be so enticing...just to think that one can capture on a computer the complexity of reality can be very seducing, but also deceiving. Situated and tailored models might look boring, as someone above mentioned about a plain excel sheet, but because of that more grounded (humble?), and then human agency through thinking can take the wheel of decisions and meaning.
Thanks for your comment and the discussion.
Thanks for the reply, but I'm still quite confused about the different terminology around thinking, knowing and modelling purposes here. The difference that Arendt sees isn't clear to me from the quotes in the text, but from your comment I get the following definitions:
Thinking = 'reconciling the fact that there is a plurality of meanings'
Knowing = 'seeking to understand'
If I understand you correctly, thinking is what we should be trying to do better, rather than knowing. At the same time, the text suggests that current approaches have been unable to reach a deeper understanding, and I would personally think that achieving a deeper understanding of a situation is often quite a useful goal.
Returning to the definition of thinking, if we understand reconcile as 'bring into agreement or harmony', I'm not sure I see this reconciling as automatically the most important task for a modelling project. To use a very obvious example, if company X sees a river as a good place to dump pollution, while community Y uses the river for fishing, bathing etc then maybe reconciling "the plurality of what the river means" is besides the point, if our aim is more-than-human flourishing? But then, perhaps the next step in this argument is that in such a case we don't need a model at all!
I hope that's a helpful provocation.
Dear Elliot, thank you for your provocations. Indeed, we state that modelling can often lead to foreclosing thinking, or reducing what we think is a world that consists of a plurality of meanings. We invite people to engage with models reflexively and critically, to enable thinking and seeing the multiplicity or complexity.
I think a clarification would be useful, as the word 'reconciling' that Jonatan used was not in the sense of reconciling the plurality (as that would be more related to 'knowing'), but was used to invite people to reconcile with the fact that the world is complex and multiple.
And yes! This also leads us to suggest that in many cases, modelling is not a helpful exercise - as you also suggest. Or other discussions/interactions are required to understand the complexities before modelling is done. There are several modelling approaches that take such a step (and sometimes that leads to the decision to not build/apply a numerical model at all), such as companion modelling.
Very interesting reflections. Much can be said to continue this discussion. A couple of points below:
- It would be nice to start opening up the taxonomy of “models”. The term, as currently is in the text, gives the sense that it is a rather uniform category. It is not the same to talk about a biophysical model (e.g. hydrologic models) as talking about other modelling efforts that aim to capture social complexity (e.g. System Dynamics or Agent-Based models). See the paper by Bots and van Daalen (2008) for a more detailed reflection on this matter and the book of García-Díaz and Olaya (2017) for examples.
- More attention should be put on the alignment between a model’s analytic capabilities and its ability to represent a specific problem or system at hand. Here is when we evidence issues such as the one presented by Godinez-Madrigal et al. (2020). In that case, a model was built in misalignment with the issues or processes that were aiming to “solve” or represent. For wicked problems such as the one discussed above, there are established modelling approaches (interestingly, not quantitative, but still qualitative modelling) that are better suited to deal with deeply contested issues e.g. Soft Systems Methodology and Critical Systems Heuristics (Mingers & White, 2010).
- The issues that occur in the intersection between society and the environment quickly escape the realm of objectivity as they imply the ponderation of questions of value, therefore having ethical implications. This is a critical consideration for water modellers to be reflective on both their models and model-building approaches, with important implications for building models in collaborative ways. See a more detailed discussion in Amorocho-Daza et al (2023).
Thanks for sharing your thoughts and sparking the discussion around models and thinking!
References
Amorocho‐Daza, H., van der Zaag, P., & Sušnik, J. (2023). Ethical considerations of using system dynamics in participatory settings: a social‐ecological‐systems perspective. System Dynamics Review. doi:10.1002/sdr.1755
Bots, P. W. G., & van Daalen, C. E. (2008). Participatory Model Construction and Model Use in Natural Resource Management: a Framework for Reflection. Systemic Practice and Action Research, 21(6), 389-407. doi:10.1007/s11213-008-9108-6
García-Díaz, C., & Olaya, C. (Eds.). (2017). Social systems engineering: the design of complexity. John Wiley & Sons.
Godinez-Madrigal, J., Van Cauwenbergh, N., & van der Zaag, P. (2020). Unraveling intractable water conflicts: the entanglement of science and politics in decision-making on large hydraulic infrastructure. Hydrology and Earth System Sciences, 24(10), 4903-4921. doi:10.5194/hess-24-4903-2020
Mingers, J., & White, L. (2010). A review of the recent contribution of systems thinking to operational research and management science. European Journal of Operational Research, 207(3), 1147-1161. doi:10.1016/j.ejor.2009.12.019
Indeed, Henry, we should discuss a taxonomy of models. We've been working on it lately. The big question is to find the essential characteristics that can make a given category unique, to find its essence and not its accidental characteristics. In biology, the essential characteristic that defines a species over others is the reproductive organs. Can we find such a defining feature for models? I'll look into the papers you suggested for inspiration.
On the second point, I think the conflict was a mixed bag of modelling issues. Because on the one hand, the research question behind the UNOPS model was a valid but biased question. The government, as a key actor, was heavily invested in proving that an interbasin water transfer scheme could be the solution to the water shortages in cities. The process of structuring the problem had already been discussed behind closed doors decades before the modelling even started. If we think of models as a process, then it would have been necessary to start the process with, as suggested by you, qualitative modelling. But on the other hand, at the point in time when the conflict was already intractable, modelling was just the continuation of war by other means, which required such a quantitative "representative of reality" approach. Modelling was instrumentalized to become a tool for frame switch in the social arena. Therefore, what we did to replicate their results using their own approach was needed to show that even then, with their own epistemological turf, the water transfer scheme was unfeasible.
And finally, regarding the ethical implications. I think you're completely right; modelling has ethical, but also philosophical, material, social and political implications; making it a very complex element that will only increase in complexity and omnipresent in many spheres of society the years to come. Because modelling has the inherent trait of easily become instrumentalized, or worse, to take a life of its own through automatic (underlaying suppositions buried deep in the mathematical soil) decisions, it's so important to lay the ground for reflexive and precautionary modelling.
Thanks Henry!
I share this interest in thinking about taxonomies of water modelling. But, I would suggest getting away from the biological metaphor and from finding essential characteristics. Perhaps it's more useful to think about typologies instead (which I understand as a looser form of categorisation that allows for multiple overlapping sets of categories - eg. by structure and by purpose). As a human geographer, I can't quite get behind the taxonomy of Bots and van Elden and its separation of physical, individual and social, but I appreciate the attempt to make these distinctions. It seems there's useful things those of us interested in water models could learn from NRM and socio-ecological-systems fields, though in the case of water 'Physical system model' is the main category and certainly needs to be broken down further.
I'm with Siva above
Models are useful as tools to help people to think, when they are applied for that purpose. Too often they are promoted and sold (and I use the term advisedly) to people and organisations as a substitute for thinking or even to confuse thinking.
We should always interrogate the intent of the instrument.
In South AFrica, the application of models for the management of a large hydrological system over three decades has guided operational decisions and proven absolutely invaluable.
Where practitioners and polticians have chosen not to be guided by similar models in similar systems - they have suffered disastrous consequences - see the Cape Town example.
The one challenge we have with the modelling of essentially technical systems is that we need to interface them with social (and political) systems. How do people behave when they're given information? How can that be guided rather than manipulated?
A couple of references illustrate these issues:-
M.S Basson, J.A Van Rooyen, 2001. Practical application of probabilistic approaches to the management of water resource systems, https://doi.org/10.1016/S0022-1694(00)00367-X
Muller M 2018. Cape Town’s drought: don’t blame climate change. https://www.nature.com/articles/d41586-018-05649-1
" 2018. Three Challenges to Keep Hydrologists and Planners Busy for Another Decade (September 18, 2018). Available at SSRN: https://ssrn.com/abstract=3287160
" 2019. Some systems perspectives on demand management during Cape Town’s 2015–2018 water crisis, International Journal of Water Resources Development, DOI: 10.1080/07900627.2019.1667754
" 2018. Decolonising engineering in South Africa – Experience to date and some emerging challenges. S Afr J Sci. 2018;114(5/6), Art. #a0270, 6 pages. http://dx.doi. org/10.17159/sajs.2018/a0270
" 2021. Why-full-dams-dont-mean-water-security.
https://theconversation.com/why-full-dams-dont-mean-water-security-a-look-at-south-africa-160898#:~:text=When%20asked%20whether%20water%20supply,like%20a%20household's%20bank%20account.
Dear Mike Muller,
Thank you for your reflections on our water dissensus. We, the authors of the water dissensus, are part of a project in which we aim to declare models to be 'controversial tools'. We do this explicitly to engage with models in a way that supports thinking, and to be mindful of the wanted and unwanted effects of the model ánd modelling process. We centralise the model in this project, to be able to decentralise modelling and re-centralise the empirical in decisionmaking. The resulting approach would then not to interface models with the social and political systems, but to put the social and political systems first, and identify how models can serve (interrogate the intent of the model) - and also where they cannot/should not, or only should for a limited time or specific purpose.
In 2004, soon after my position as Program Director in Hydrologic Sciences at NSF ended, Vijay Gupta used all his influence in vain to try to get me on the editorial board of Water Resources Research. My fault was that I was “too controversial.”
In my 50+ years of experience since I started college, I have learned several things. One is that it is difficult to write a good essay without passion. More importantly, in my experience at least, it has been impossible to think philosophically and generally about a scientific issue without having explored its underlying physico-mathematical structure in depth. Well, at least in such a way that it might be useful to others.
As for Heidegger’s quoted statement, “Science does not think,” altered to “models do not think,” the question is: what does one wish to gain from this perspective and why would one care? The participants go on to frame the discussion in terms of “water resources” and the associated models. Do they realize that, as has already been argued in 1987 by Vit Klemes in “A hydrological perspective,” the term “water resources” already tends to set the boundaries of inquiry within technology, rather than science? The reasons for this are also well-known and arise from a desperate need to solve a problem now and use the solution as a guide for management. Of course, some of our most venerated journals have “water resources” in the titles.
From Klemes: “For the most part, hydrologists are still dancing to the tune of ‘constantly trying to find better ways to manage water’, which is not theirs but technologists' task, and are either unable or unwilling to say "Enough is enough - - our task is constantly to seek better solutions to the water balance equation.’ ” Here it can be added that the solution of this equation was named the central problem of hydrology by Robert Horton already in 1931. Certainly, much could be said about what constitutes a “solution” to this equation.
Of course, models do not think. If they ever do learn to think, there will be even less motivation for scientists to learn to think than there is now. What is alarming is not that “science does not think,” but that those who wish to be scientists do not. Ever since I once became program director in Hydrologic Sciences at NSF, I became aware that most hydrologic “sciences” research was not science, but phenomenology, as stated again by Klemes in his reference to the Radio Yerevan joke about what hydrologists actually do.
Review the literature for the water balance and what do you find? The most commonly cited model, used to test his land surface model for accuracy by Manabe in one of his papers cited by the Nobel Committee, is the square root of the product of two models that were guessed for the function at the turn of the 20th Century. The reasoning? One tended to overestimate evapotranspiration and the other tended to underestimate it. Can somebody tell me what the hyperbolic cotangent function might have to do with the water balance (the sum of two exponentials divided by their difference)?
In 1998, Bernabe and Bruderer tested a method of predicting flow in a disordered network in which the local flow resistance distribution was known and found that of the general theoretical approaches known for treating this heterogeneity, the most accurate was the critical path analysis from percolation theory. You might think that, if it is possible to verify this using a known system, the method might have been applied in less well-understood systems. The same result for electrical networks had been obtained 20 years earlier. By good fortune, I had completed my Ph.D. in physics under the person who had invented the technique.
Coincidentally (in a temporal sense), I had tried this method in 1998 in Transport in Porous Media to develop a procedure to calculate distributions of the hydraulic conductivity. In 2008, I extended the method to address solute transport. Four years later (Ghanbarian et al. 2012), we showed that it can generate predictions (that is, with zero adjustable parameters) of complete, non-Gaussian, solute arrival time distributions more accurately than fits with dozens or more unknown parameters. By continuing this research, it has been possible to find the limits on vegetation growth and soil formation due to the heterogeneous soil network. Finally, the results were combined with an ecological optimum (divvying up the water between soil formation and vegetation growth so as to maximize the net primary productivity NPP of ecosystems) that actually allows prediction of the water balance, associated stream-flow elasticity, and NPP (see the highlight from 2023 in Eos, https://eos.org/editor-highlights/how-much-terrestrial-precipitation-is-used-by-vegetation about streamflow elasticity from AGU Advances (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2022AV000867), and its linked paper with the same theory and theoretical parameters in WRR from 2024 on NPP https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023WR036340 ). One short quote from the highlight in Eos: “The contribution allows us to predict the impact of changing climatic conditions on the interplay between climate, vegetation, and water resources.”
Thus, has it become possible actually to use science to find solutions to problems of water resources, too. However, the process was a 25-year long adventure, that overlapped for 6 years my 20.5 year search for a permanent position. It also involves some 75 refereed publications and five books.
In this regard, I actually was informed that I was the most cited author in SSSAJ for the year 2013. However, that number of citations is only about 1000 and the most cited paper of the three I published there that year was a critical review on tortuosity (first author Behzad Ghanbarian, at that time my Ph.D. student). In any case, the number of citations does not really count for much. Did you know that a paper that has been cited over 30,000 times, making its author the most cited author of SSSAJ in 1980 – and thus probably of all time – actually contains an internal mathematical contradiction (and that this contradiction was pointed out in a journal edited by the same author, as well as in our 2013 What’s Wrong With Soil Physics, SSSAJ paper)? This, of course, means that the paper with 30,000 citations is, a priori, not science, bringing up yet another quote of a quote by Vit Klemes: “Hall criticized the common practice whereby ‘speculative assertions . . , become scientific fact. . . [by a] proper number of citations in the literature." (emphasis added.). It is, actually clear that, within the theoretical framework of hydrologic science extant, the supposition that the saturation-dependence of the hydraulic conductivity can be obtained from a suitable average over capillary bundles, is no more than a speculative assertion, as an analysis in terms of the paths with the lowest possible cumulative resistance shows (Bernabe and Bruderer, 1998 again) that the entire potential field is controlled by a few (critical, in the sense of percolation) conductances, makes the assertion also demonstrably false. The assertion is only true if, cast in percolation theory, the critical volume fraction is actually zero, yet this number is used as an adjustable parameter in the tortuosity correction.
So the field of hydrologic science (and its related field, soil physics), almost 40 years after Klemes’ essay, is still on ground as firm as a southern California soil liquefied by an earthquake. The real question is: Does anybody care about it? In another invited essay (Hunt, 2022), I pointed out that a problem that is bigger even than that of models that are irreproducible, is the accompanying cynical suggestion that, if no model is correct, what justifies the preference of one model over another? Indeed, why bother to build a solid foundation at all? After such a long career of solving hydrological problems I hope to see some echo of the passion for getting things right in our next generation of hydrologists. This would go a long way to answering our question from WWWSP (Hunt et al. 2013) of whether soil physics is a branch of physics that deals with soils, or a compilation of phenomenologies “fudging” (Klemes, 1988) physical properties of soil. It seems to me that this is a concrete guide to the navigation around some of the basic issues in this discussion.
Bernabé, Y., & Bruderer, C. (1998). Effect of the variance of pore size distribution on the transport properties of heterogeneous networks. Journal of Geophysical Research: Solid Earth, 103(B1), 513-525.
Ghanbarian-Alavijeh, B., Skinner, T. E., & Hunt, A. G. (2012). Saturation dependence of dispersion in porous media. Physical Review E, 86(6), 066316.
Hunt, A. G., Ewing, R. P., & Horton, R. (2013). What's wrong with soil physics?. Soil Science Society of America Journal, 77(6), 1877-1887.
HUNT, A. (2022). What perspective will best lead soil physics into the future?. Journal of the Japanese Society of Soil Physics, 152, 3-6.
Klemeš, V. (1988). A hydrological perspective. Journal of Hydrology, 100(1-3), 3-28.
Allen:
I saw your contribution by accident, as I opened a suggested webaddress autofilled on/by my phone. And I was surprised to see the number of posts since I last responded. Sitting around, I looked casually over them, and found your's curious and in parts puzzling.
While I would certainly feel compelled to respond to many of them, and spend another 30mins to dig up scholarly references to make my response look more respectable (something not as hard in today's age), I will limit myself to answering a specific question you asked: "The participants go on to frame the discussion in terms of “water resources” and the associated models. Do they realize that, as has already been argued in 1987 by Vit Klemes in “A hydrological perspective,” the term “water resources” already tends to set the boundaries of inquiry within technology, rather than science?".
I can, of course, not know what other participants experience when using those terms; and how typing or uttering the words shapes their thinking or narrows their view. Personally, I am well aware of all sorts of critique associated with it: from the one you mentioned (even if Klemes is totally unknown to me, his general thought and argument about the issue are of course not original as such, as much as he may have been one that introduced it to certain), all the way to arguments claiming those words are part of a plot for capitalist world dimination and wilful environmental destruction. Some putting this into context to the "rotten core" of western thought all the way back to the Greeks; others (more modestly) just going back to Marx; and others invoking all sorts of intellectuals since then.
You may be surprised, that an increasing amount of students and young graduates today, are well aware of it, and still use it. During their studies, they have have been exposed to different heterodox scholarship. And so have I (not because of the curriculum, but as I was curious and guided by people with a critical attitude about a certain narrowness about the scientific worldview or engineering). Sometimes I found merrit in the arguments; at other times I found them intelligence destroying.
I started using those terms, as my job title suddenly became "*** water resources specialist"; a "**** water resources management ****"; or "water **** engineer". Nothing of this changed my views; neither were they changed by using the word myself (a quite normal process, as many institutions simply call it that way, and will edit one's work quite freely, or insist on changes to comply with general expectations). For outside of an academic environment or scholarly article, one chooses words for all sorts of reasons, apart from this specific aspect.
But words really dont matter much. Wheter a person (a) sees a river and believes it should be turned into a cascade of oversized bathtubs; (b) looks at a stream and close by cultural lands, and only sees diversions and intensified hightech irrigated agriculture; or (c) looks at a forst and wants to immediatly turn the whole thing into a cashcrop plantation, is really not fundementally the result of the use of a word.
People (and society) put meaning into these words much more fundementally than the other way round. And those are really quite different in different parts of the world; with different cultural and philosophical heritage; and different traditions. Some put meaning unconscously; others actively engage with the purpose to redefine the meaning of words. Some do so to further worldview A; others to counter the narrowness of worldview A, push worldview B; and others want to change completely to worldview C. Some believe certain words are irredeemably associated with a way of thinking that led to catastrophy and introduced their own lexica that only a few initited will ever understand (think Adorno). Others dont believe so and keep chipping away in an effort to gradualy shift or defend meaning.
That is to say: one should simply not draw to hastely a conclusion of why people use or dont use certain words; that any of the meaning Klemes has associated in his critique is not something constantly in flux; or that anyone doing so is an intellectually impaired advocate of certain aspects of water management that you feel strongly about. For I could have used a wide variety of words instead of "water resources" in a response to a blog post: water, H2O+, or liquid fluid sustaining the beauty of the world. Just as much as the use of "management", does not force one to see dikes, spurs, concrete and rebars when being asked to help understand increasing levels of extreme flow events in the mid-hills of the Himalaya, in an effort to prevent further loss of life.
Some certainly do believe that concrete is the all powerful solution; and that the marvels of modern technology is the ultimate saviour. Others do not. And more and more people have rather sophisticated views on how to balance things. But only in some very specific cases will societies decide to tarnish the use of a word to an extent that makes it socially inacceptable: and "water resources" or "management" is not one of them for the time being.
Philippe
Please excuse my rather polemical tone; and I would obviously not use „intellectually impaired“ but rather „ignorant despite knowing better“.
Both a result of my early morning typing, and as I am rereading some wonderful polemics from a century ago.
Philippe,
Thank you for your thoughtful response. I urge you to read Vit Klemes' article. He was one of the people involved indirectly in the founding of the Hydrologic Sciences Program at NSF. Ed Waymire and Vijay Gupta were heavily involved in writing the Eagleson Report (Opportunities in the Hydrologic Sciences, NAS, 1991). Vijay recently passed away (as did Eagleson and Ignacio). I think that Vit's comments still hold today. That the Budyko (and Choudhury, and all the other) phenomenologies for the water balance can be treated as theories is a mystery. In any case, long after we proposed our solution to the water balance, I went back and read pages 65-66 of the Eagleson report and found that we almost perfectly followed the proposed template. Ed Waymire has confirmed - calling it a remarkable perspective. In the process, I think that we also have done what Vit Klemes and J. C. I. Dooge said was what was necessary.
Is this self-advocacy? Of course. If I know that we accomplished what the HS program was set up to do (and it is further confirmed in Eos by Alberto) and I am 69 with a job that is threatened by an insanely stupid Board of Trustees (etc.), would I not be irresponsible if I did not?
This is why I suggest reading these two pieces of the literature. Confirm for yourself (ves).
Sincerely,
Allen
Many thanks for your kind response, Allen.
Rereading my earlier post, I would not have answered as politely - not one of my strenghts of course.
I downloaded the paper and will carefully read it (the first para already reads fitting), and take a look at the other citations you shared. For I read as a passion and pasttime. I will be happy to respond to you by email (as I believe I already found your contact details).
Kind regards
Philippe
I will be glad to hear from you. Best wishes, Allen
It’s nice to see that the originators of this Dissensus, that are still in their early careers, have come reflect on the validity of the models, that they have been taught to use. As far as I know, such models now come pre-packed to fit a specific group of problems, and limited possibiIities for modification. They may not fit so well, when it comes to real world application. Here, I take it that you are referring to systems models for forecasting something. For such models, we will never understand enough, something my systems guru, C. West Churchman, never failed to point out. He had had the American philosopher Edgar A. Springer as his mentor, and He differed from the the philosphers, that have been discussed in this Dissensus. His pragmatic stance never allowed him to understand philosophy as a merely theoretical enterprise; rather, philosophy to him was an intellectual effort to improve social practice. Thus, Churchman referred to his systems as inquiring system. You need to know the system context, and its history, in order to be able to develop, or find, a model that is tailored to the problem context. He also warned about the the enemies of systems; aestetics, ethics, politics, and religion. He was also the first to write a paper on “wicked problems”. These are problems that defy exact solutions. Thus, we now have Mode 1, where your models might be adequate,as well as Mode 2 science, where they are not. Mode 1 refers to “Old School” problems, while Mode 2 refers to post-normal science and wicked problems, which require enlarged pier groups. Such problems don’t allow for any development of confidence intervals. Here, we need to rely on social assessments of what is a “good enough solution”. The thinking needs to come the modelling efforts. Thus, I recommend you to be humble, and realize that a good definition of the problem at hand, is the key to a meaningful model. As the saying goes, garbage in – garbage out.
The central problem of hydrology, the business of hydrology, the most complex problem in hydrology, is the water balance. It is not a Mode 2 problem. It is a Mode 1 problem.
The days of normal science are not over in hydrology. It is still possible to formulate hypotheses and test them, and, when the predictions succeed, to solve actual problems. It may not be an easy process, however and may take a quarter of a century. Certainly, it is made harder by refusing to adopt the appropriate tools for the work. Rejecting philosophy is not a productive avenue, as the scientific method is still applicable, nor is defining problems as wicked, because they involve systems, or because they appear to be wicked. So-called Mode 2 problems may be transformed to Mode 1 problems by using the appropriate tool. Casting science as post-modern definitely does not help when the solution is known; it only serves to foster the cynicism about an unverified assumption of the fundamental indistinguishability of models (or, going backwards in the present context, of thinking, or of science).
It has been possible to unite the understanding of: flow, hysteresis, transport, weathering, vegetation growth and productivity and the grand problem of hydrology, the water balance, together with plant species richness as a function of climate, within a single framework. If I choose, I can write a 4th edition of our Lecture Notes in Physics book to include these last few topics, or write a new Springer book in their geophysical monograph series; the 2nd edition of Networks on Networks is about to be submitted to the publisher, our book, “Hydrogeology, Chemical Weathering, and Soil Formation” was an AGU “best-seller,” and our work on the water balance in AGU Advances has already been characterized in Eos by Alberto Montanari as predictive. There are only limited fundamental inputs necessary: 1) the soil is a network, 2) interactions between the soil and the plant networks are best treated by acknowledging that the soil is a network, 3) the physical properties of the soil network are best treated using percolation theory, because it quantifies the paths of least resistance, and 4) it is possible to use an ecological principle that maximization of net primary productivity guides the evolution of the plants. Failure to acknowledge the success of the result does not absolve responsibility to use the best science; (willful) lack of familiarity with the law is (culpability) not excusable in court.
Sociology is probably all Mode 2, but turbulence and its derivatives may well be Mode 2 also. I was thinking about problems in the subsurface and stand by that.
Dear Allen Hunt,
It certainly never was an easy task to assess the water balance, because you have only point measurements to assess areal or volumetric entities. By now, as you may know, we are experiencing a global warming, that has considerable impacts, on two of the components of the water balance equation, namely evaporation and rainfall. Our hydrological statistics toolbox has simply become obsolete. If there are people living within the basin, they will, as you point out, have an unpredictable impact on the water balance.
The IPCC forecasts are the results of work by thousand of scholars, but they are still expressed in terms of intervals, labelled as low, most likely, and high, respectively. These are just some mean values for the entire planet and, as well experienced by now, there are huge differences between different parts of our world.
You mention your book on chemical weathering and soil formation. I guess it is based on historical data, maybe in a geological timescale. I don’t really believe, that you would be able to forecast how such a process would develop for the next 25 years within a particular drainage basin.
In 1988, Vit Klemes, President of the IAHR, pointed out the desperate status of research in _Hydrological Science_. 3 years later, the US NSF HS program was founded with principal goal to solve the water balance ("define the terrestrial water fluxes"), as advised by Dooge, Klemes, and Horton over 50 years earlier. Klemes was full of hope for changes. Nearly 40 years on, the phenomenology of choice for the water balance has not changed (still Budyko). As noted, that formula is ad hoc in the extreme. As noted, other common results, such as the van Genuchten formulas for the unsaturated hydraulic properties have no scientific basis. The status of hydrologic science and soil physics has not changed. Solutions based on percolation theory that outperform those based on averaging have been given. They conform to the vision of those who put together the Eagleson report. They took 25 years to develop with great labor and little or no institutional support. If you look carefully at the literature from this perspective, you will have to recognize the lack of progress. In 1981-1982 Muhammad Sahimi, for example, showed that hysteresis in wetting and drying was a consequence of the distinction in percolation between allowable and accessible sites in a network. This information is unavailable in the hydrology literature today. For a summary of the (non-) intersection of percolation results with hydrology, please see the Wiki site, http://www.history-of-hydrology.net/mediawiki/index.php?title=Percolation_Concepts_in_Hydrology, invited by Keith Beven. It has plenty of demonstrations of the value of using a science that quantifies the effects of heterogeneity on flow and transport, rather than treating them as a problem.
It really does not matter whether you believe that I can solve your catchment with the equations. What matters is whether you can solve the problems of your catchment with those equations. They are simple to use. If you wish to find out where we are today, look at what these people said over 30 years ago. Other references are given in a previous post.
Allen Hunt
Totally agree with what is stated in the text. Models are tools that have to be socialized and contextualized in the territory. Otherwise, technocratic management will be addressed in water management and that will even generate disputes and conflicts due to the solutions derived from the use of the model, for example hydrological ones. This is why it is important that technical water decisions be passed through a "filter" or framework of governance and human rights. Unfortunately, many engineers are not familiar with these comprehensive, multidisciplinary approaches.
In a recent case from 2023, the National Water Commission of Mexico decided to deforest 24 kilometers of an urban river (Santa Catarina) in Monterrey, Mexico in order to increase its 'hydraulic capacity'. This decision was based on a model (https://www.youtube.com/watch?v=PstsCvD41cw) but the decision was never communicated or consulted, so in the territory this decision generated an important social mobilization and legal instruments were imposed to stop deforestation works (https://animalpolitico.com/estados/colectivo-suspension-frenar-desmontes-rio-santa-catarina-nuevo-leon).
Expertise is no longer enough if decisions regarding water are not democratized, and even more so when there are ecological and social impacts.We need multidisciplinary and transdisciplinary water management.
Best!
Water systems and models inherently diverge in their ontological nature. Water systems are dynamic entities, constantly shifting and evolving, whereas a model represents a static snapshot of a particular scenario. This fundamental difference implies that we cannot rely on a universal model to address every problem. However, when we narrow down the scope of models to tackle specific issues or understand conflicts, the influence of power becomes increasingly apparent. Nevertheless, we cannot disregard the utility of models. The pressing question then arises: How can we develop responsible models? Who qualifies as a responsible modeler?
My dad once explained me the difference between a thought and an idea: thoughts are fuzzy and only untangle themselves and acquire meaning once we put them into language. I see models as a powerful tool to turn fuzzy "collective thoughts" into actionable ideas.
The more I dive into the world of using models for different stages of the policy life-cycle, the more I see their value in facilitating collective thinking processes over any other of their functions, such as sources of truth. This means that the process of making the model and untangling these collective thoughts is as valuable or more than the resulting model in itself. And that once you have a model, its value resides in its capacity to trigger and guide discussions between sectors and disciplines more than in any insight that it might provide.
If we see models as a technology to do transdisciplinary decision-making and we couple them with inclusive and representative participatory processes, they can help us tackle the ever-growing complexity of the challenges ahead. If we do narrow-sighted, sectoral models that then become ubiquitous sources of truth, they will just help us continue digging the hole we've been digging for some time.
What a superb dissensus. A quick comment, drawing on some emails with Rozemarijn, on the emerging papers from the Special issue on the politics of quantification, and drawing on my own experience with two models that computer 'real water savings' in irrigation (mine and FAO's REWAS. Links below).
I consider the FAO REWAS model to be so wholly inadequate that it is 'not even wrong'. (Link below). Whether you accept my view is neither here nor there but I would like someone with an irrigation background to defend why it omits so many factors that determine real water savings. Or can it be that we seem not to know how to differentiate irrigation models in terms of their workings, objectives and provenance? Into this gap or lacuna steps political claim-making of a model's utility.
I muse that models become political because we don't really have good Darwinian winnowing procedures for dismissing environmental models that a) have long habituated or default usage; b) are uncompetitive in explanatory and predictive terms; and c) that foreclose robust debate and learning. In terms of Darwinian sweeping out of poor models, we also lack an empirical feedback loop that might determine which models to reject (or improve). We should recall that irrigation systems (and other socio-ecological systems) can continue to operate/produce food/consume water at a relatively low equilibrium. In other words system failure, guided by what a model tells us to do, is not stark or on-off. (To help anyone think about what I am saying, look at SpaceX; if their models are inadequate, their rockets will crash and explode. They are constantly iterating towards better designs and models). I will leave it there and post this now.
https://en.wikipedia.org/wiki/Not_even_wrong
https://www.sciencedirect.com/science/article/pii/S0378377423003025
FAO = https://www.futurewater.eu/projects/training-package-for-water-productivity-and-real-water-savings/