In our conversation today with three researchers from the Swedish University of Agricultural Sciences (Elin Röös, Robin Harder and Johan Karlsson), we discuss what food systems models can and cannot tell us about what type of future food system we'd like to create, previous projects modeling food systems at different scales (bioregion, country, continent), and how our values influence what questions we ask from a model and how we interpret its results.
For more info and transcript, please visit: https://tabledebates.org/podcast/episode14
Ep14: Elin Röös, Johan Karlsson and Robin Harder on "Values in food systems models"
Welcome to Feed, a food systems podcast presented by Table. A collaboration between the University of Oxford, Swedish University of Agricultural Sciences and Wageningen University. I’m Matthew Kessler. And today we explore the topic of Food Systems Modeling at different scales and how our values as researchers might impact those models.
We’re trying out a new format for episode. We’re sharing a roundtable conversation that co-host Samara Brock and I with 3 researchers based at the Swedish University of Agricultural Sciences, Elin Röös, Johan Karlsson and Robin Harder. Each of them has worked on modeling the environmental impacts of individual foods, diets and whole food systems at different scales, from a bioregion, to a nation, to a continent.
We first start our chat unpacking what exactly is a food systems model, and what might we learn when we model different future scenarios? We discuss how our values might influence what questions we may ask from a model and also how we might interpret the results. It’s a grey area in the research community and we speak directly to some of these tensions while reflecting on our own biases.
Also, when speaking with researchers and systems thinkers, you might hear this phrase a few times
I guess my answer is it depends-
It think it really depends on-
It really depends-
So again it depends on the model-
So you'll be hearing different voices on this call. And we'll first start with the round of introductions.
I'm Elin Röös. I’m a researcher and teacher at the University of Agricultural Sciences in Uppsala. And my background is in life cycle assessment, LCA and environmental modeling of food and food systems, agricultural systems and so on. And I work on many interdisciplinary projects all related to achieving a more sustainable food system.
Hi, my name is Robin Harder. I'm calling in from Knivsta, south of Uppsala, in Sweden. I'm a postdoc at the Swedish University of Agricultural Sciences. And I'm working with recirculating nutrients from human excreta and wastewater to land and food systems.
My name is Johann Karlsson. And I'm a PhD candidate at the Swedish University of Agricultural sciences in Uppsala. And I work on the role of livestock in sustainable food systems.
So, you all work with quantifying the environmental impacts of foods and food systems using different types of models. Some of you may have heard of LCAs, or life cycle assessments. But this is only one type of analysis. Elin, can you describe some of the different models used to assess the environmental impact of foods and food systems and on what scales they operate?
Yes, let’s start with LCA which is a method that’s been around for quite some time and has used a lot in different sectors to quantify the environmental impacts from different products and services. And it’s been used quite a lot also for food. So in an LCA what you do is you follow the product from the cradle to the grave, as it is called in LCA lingo, and what we mean with that is you look at all the emissions and all the resources – energy, water, and so on, land – that is used along the whole life cycle of the product. So beginning with the farm inputs, the emissions of the farm, packaging, transports and so on. And then you relate this impact to a certain amount of the product, for example 1 kg of let’s say beef or pork or potatoes. So this method allows us to compare different products. For example how large is the climate impact of 1 kg beef in relation to 1 kg of pork and so on. But you can also quantify other environmental impacts, not just the climate impact. You can look at eutrophication or acidification.
So these LCA results, which are also commonly called footprints, like the climate footprint or the water footprint of a certain product, they can be really useful for different purposes but there are some clear limitations to what LCAs can give us answers to.
And can you talk about some of these limitations?
One thing is that we need to look at the amount of different foods consumed in total to be able to say something about where the major impacts come from. So for example LCA can tell us that a certain type of food has a very high impact, but maybe that food is consumed in very low quantities. So it’s not really a problem when you look at the total amounts consumed. But what we can do is we can use these LCA results, or these footprints, to calculate the impact from the whole diet or the whole country by multiplying the footprint results for 1kg of food with total amount consumed of that food. Then it becomes much more useful when we have the full environmental impact from a certain diet or from a country or the world as a whole. And what we can also do then is compare this result with some kind of threshold for what we can consider is a sustainable impact from the food system.
Right, another nuance related to scaling is understanding why exactly one product’s impact is higher than another. For example, it could be really to how a product is transported rather than how it is produced. Foods transported by flight are really bad for the climate, but they are consumed in very low quantities. A 2020 study by Our World in Data showed that 60% of food is transported by boat, 30% by road, 9 % by train and only .16% by plane.
Johan, you’ve worked on a report modeling different futures scenarios for the Nordic Food system. Can you share an example of how scaling these findings from an LCA can potentially be misleading?
Yeah, so a thing that is lost when you take LCA based data and try to scale that to different scenarios for changed diets or changing consumptions, what you lose is these interconnections that are inherent in food systems. A concrete example is that if we look at a farm based LCA, we might be interested in the relative environmental performance of substituting different feed components in the animals’ diets. So one example there is we might substitute soybean feed with rape seed meal that is a byproduct of from rapeseed oil production. So looking at this from a farm scale we would likely find that substituting soybean meal for rapeseed meal will lower environmental impacts because we would not incur as much deforestation for instance. But then looking at this from a more food systems perspective we would find different resource limitations within the food system. We would find that we only have a certain amount of rapeseed meal that is produced from the amount of rapeseed oil that we produce. So here we would see that we wouldn’t be able to incorporate or substitute all soybean meal for rapeseed meal for instance. So depending on the scale of animal source food production you would have in your scenarios, you would be able to incorporate different proportions of these byproduct feeds into livestock diets. The magnitude of production changes the product, the environmental orland footprints. Looking at this from a food systems perspective we would probably need to reduce the total amount of animal source food production.
Maybe then just to conclude this part on different modelling approaches, we should acknowledge that there are many different types of models you can use to assess the impacts from the food system. So LCA and these biophysical models, they’re two examples. Economists use general equilibrium models to investigate how different policy and different developments effect land use and greenhouse gas emissions and jobs and so on. So these equilibrium models, that’s another type of looking at the food system and modeling these effects and its built around economic optimization and behavior of consumers and producers. So the logic in these models are built around how people have historically reacted to prices and policy and so on. They are often quite complex to run and understand. What we've worked mostly which these more simple, purely biophysical models that I would say, they are a bit easier to understand and a bit more transparent maybe. But they don't have any economic logic in them. It's more like describing the flows of biomass in agriculture and in the food system. For example, instead of relying on historical data on how consumers react to prices, and so on, the demand, the diet , it is set as an input to the model. So we can change in any realistic or unrealistic way. So in these biophysical models, we are free to explore any option here in terms of diets and productions and systems and so on, and whether that is realistic or how it can be accomplished, it’s not really part of this type of modelling exercise. And of course because of that, we can’t say anything about how the future will likely develop. It's not a prediction, it’s more like an exploration of the different options we have. So depending on how we eat and produce food, we can say something about how the environmental impacts are likely to develop.
We see from this description that when modelling food systems that it’s important to consider not only to the temporal and geographical scale of the model, but also whether you are looking at an individual food product, a particular diet, or the whole food system.
So, it seems like there's a lot of different approaches and variables that can impact results of models. So things like how boundaries are drawn, what's included, what's not that whether you're using an economic model logic or not. So if two different researchers were tasked with the same research question, would they come up with the same results? And if not, why not?
Well, as a researcher, I guess my answer is it depends, which is quite common answer for researchers. And why it depends, I think it is, it is important distinguish between what is referred to sometimes as the analytical choices and pre analytical choices. So the analytical choices are choices you make in your actual model. Whereas pre analytical choices are choices you make before you even start modeling or before you use the model. So, those pre analytical choices could be decisions about what do you deem relevant enough to even include in the model? Or what is your what is the starting point in my field human expert management, the choice could be? Do you see the human excreta management as a part of a resource recovery challenge in waste management? Or do you see it as a nutrient supply challenge to land and food systems? And that obviously impacts what you're starting to look at in your analytical model. So, having this in mind, I would say if two researchers are tasked with building a model that is based on a similar pre analytical choices, I would say chances are that the outcomes would be rather similar. However, if the pre analytical choices are widely differing, I will also expect the outcomes of the models to be very different than and that's really what, what in my experience, often if I don't agree with a model, it's not that I don't agree with the model as such that it's more that I don't really buy the assumptions that were made before the model was being built.
Hmm. And those assumptions are something we're going to dig into a little bit later. Are there other perspectives on whether or not someone would land on the same answers if they were asked the same research question?
I think it depends on how well defined the research question is. If we take a very general question like, could organic farming feed the world? We can try to answer that using a model and you can get very different answers depending on how you set up the model and what assumptions you put in, in terms of diets and yield improvements and impacts from climate change and so on. So for example if you assume that we will continue to consume according to current trends then of course organic can be really bad because it can cause a lot of deforestation. Because we would need more land since organic yields are at least currently lower. But if we instead assume that we can also change diets to more plant based and we can also reduce waste then organic can feed the world on the currently available land and deliver the benefits of organic agriculture and feed the world. So you can get very different answers. I mean it’s just like Robin said, it clearly all depends on these predefined assumptions that you make before you start running the model.
One aspect that I've been thinking a bit about is that when you are sort of in the practical work of constructing your model, you spend a lot of time making sure that you didn't screw up anywhere within your models. So you look at your results from a number of different angles. And then you see some results that make you go, Wait a minute, that can't be right. And then you really go into detail to make sure that you didn't make a mistake somewhere. And so I think that sort of this realization, looking at results that, wait, this can't be right. It's very much dependent on your sort of preconceptions about how society works, and how food systems in particular work. But I think also within the construction of a model and itself, there is an influence of values and preconceptions by the researchers that that will affect the results and may lead to different results, depending on which researcher tackles the same question.
Clearly people have different normative values that they might bring to the model in the questions that they ask. For example, a normative value might be making sure that everyone has access to sufficient calories and nutrition, or it could the wish to conserve a certain percentage of land for biodiversity, or it could be a demand for better animal welfare standards. I'll open this question up, if someone can maybe share some insights into their work with modelling the impacts of the food system and how evidence or values-based assumptions have come into play?
Well, a model can be used to assess any type of normative or explorative future that we want to look into. For example, we might be interested in how far emissions can be reduced with a certain technology or how much emissions would be reduced with a certain type of waste reduction or dietary change, or of course any combination of such mitigation options. So usually we use models with some type of scenario or a bunch of scenarios to investigate different alternatives.
So maybe here I should clarify the difference between a model and a scenario. The model is the more technical thing that is made up of the mathematical equations used to calculate different impacts. As far as possible this model is based on empirical data and mathematical relationships. These biophysical models in themselves, they are not full of values. But for example, to determine how much carbon dioxide that is released from burning a certain amount of oil, that is given by a chemical formula that we use in the model.
But the scenario then, that is what that determines what we input into the model and these scenarios are of course full of them values and assumptions. That concerns how will food be produce and which type of production systems, how will diets look, how fast will certain technology be deployed and so on. So we can also work with scenarios, just qualitatively, describing these futures in words. But we can also use these models to put some numbers to it, it adds an extra dimension of feeding back what it would actually mean in terms of foods produced and different impacts of greenhouse gas emissions, land use and so on. That adds valuable information to the scenario.
We can develop scenarios in many different ways, that’s a whole science in itself. And sometimes they are developed by the researchers but often it is done with stakeholders. So commonly, actually as researchers we are modeling someone else's vision. And that’s exactly what we did this with a set of Nordic stakeholders. There were 5 NGOs I think that had this common vision of how to develop agriculture, and the food system in the Nordic Region. So maybe Johan you want to say a few word about this because you worked most closely with them.
Yeah, so we had this group of NGOs that were more or less environmentally focused NGOs and some small-scale farming organizations, from the Nordic countries. And yeah, as Elin said, we tried to make them provide the sort of the more normative decisions on what we should model. So we tried to make it very explicit that these are the normative choices and, for example, saying that byproducts should be used to feed animals. This leads to certain results, or, in this case, we modeled organic production systems. So that was another sort of normative choice that we should look at scenarios for organic production, which resulted in certain crop rotations and certain yield levels. So we tried to make those things very explicit. But at the same time, I also think that because we did this in a collaborative way, with a lot of information going back and forth between ourselves with researchers and NGOs, so I think we sort of influenced each other in a way and sort of the conceptual models in our head as researchers also influenced how we constructed our models and the results we got. So I think it's really hard also to pinpoint your own values and norms that are so interlinked to our existence in some way. So it's very hard to really see what influence your values have on the models you construct.
So you're touching on something I wanted to ask about the participatory modeling process that you undertook. You sort of outline how values shape models, but it seems that what you're talking about, and I'd like you to elaborate on a little bit more is that, in fact, participating in a modeling process can in turn shift our values, norms and assumptions about food systems? And could be used as a sort of a deliberation tool to bring different people together, possibly. Have you seen it be useful in that regard?
I can maybe give one example from the work with the Nordic stakeholders that I thought was quite interesting. And because this worked built on previous work where we had used bioenergy in agriculture to replace fossil fuels, so producing some bio energy from biomass harvested from agricultural land. And then when we started to work with these NGOs, they were quite opposed to using any agricultural land for bioenergy, because that's what was in their kind of strategy that we need all the land to produce food. So we said okay, we'll just have to remove the bioenergy then and we'll have to replace it with fossil fuels. And we did that and we kind of showed them the results and we said okay, but we still have this food waste and manure, we could produce some bioenergy here without using land actually. And there was some discussion back and forth if they would accept that or not, because it was kind of against their strategy in a way and then finally that they said, “Okay, let's use that bio energy because we don't want the fossil energy either.” So I think it was way where the modeling and this whole process actually, yeah showed the possibilities and have them change their mind a bit and had them change their values a bit or their views on bioenergy.
So this type of participatory scenario building process can help more clearly articulate what are the tradeoffs when pursuing different futures. Elin shares anexample about how to deal with managing resources that are more abundant in some parts of the world.
And also something that we discussed a lot was fish. Because Norway has a lot of fisheries as you know. How to kind of divide this fish should Norwegian people be allowed to have a lot of fish in their diet because their a fishing nation or should it be divided equally among the Nordic countries, or even globally, maybe we should just share all the fish. That's what we did in the end. We just shared the fish, didn't we, Johan, if I remember correctly now.
Yeah, we ended up looking at sort of global, projected sustainable yields and splitting that by the global population to reach sort of a fair share of wild fish in the diet.
Yeah, but that was something we discussed a lot. And I mean, you could also read, like, maybe we should rely more on local production. And then tmaybe it make sense that Norwegians eat more fish than the rest of us. So that was clearly a normative decision that we had to take.
Johan offers another example from his work in the Nordic region in the contested debate about how to use and think about Semi-natural pastures.
Semi-natural pastures in the Nordic region are really important hotspots for biodiversity and hold many cultural values. So to keep these landscapes open and preserve these values, we generally need to have livestock grazing there. But another argument for maintaining livestock production in these areas is the potential for food production where we wouldn’t otherwise be able to produce food. And looking at this at first glance, it seems like a really valid argument. But when we actually quantified how much of the Nordic diets in our scenarios that could be attributed to biomass grazed from semi-natural pastures. We actually found that their contribution in terms of protein or calories produced in relation to what produced through feed and food grown on arable lands, their contribution was quite minimal. So I think this was a really interesting result, and a way that quantification can be important to provide insights into which arguments for something that are valid and which arguments are maybe not so valid.
So you've all conducted research at different scales and we wanted to talk with you about those projects. Robin, you were part of a food system project that modeled the Okanagan, which is incidentally where I'm from. So I'm curious for you to briefly describe that project and what specifically you were modeling.
The Okanagan food system project was a project that I was able to participate in while I was visiting Vancouver as a visiting postdoc, and there was a team from the Institute of Sustainable Food Systems (ISFS) at quantum Polytechnic University. along the lines of what Elin outlined with the participatory processes. The idea was that the different stakeholders, different actors in the Okanagan would be involved in exploring possible food system futures for the bioregion, and the indicators that we looked into as a team were as broad as looking into consumption and production in terms of food self-reliance, but also habitat like ecological aspects, water use, but also economic indicators like jobs created, tax revenue, etc. And we also, on like the farming side, what I was mainly focusing on was the nutrient management. So how much nutrients are needed for certain food system scenarios? And what is available in different types of organic residuals? And how does that compare to one another? So do we have an accumulation of nutrients in the region? Or do we lose nutrients in a region due to exports? That was my contribution. But I said the whole project was much broader than just the nutrients and I looked into a whole suite of indicators.
And did you find anything in that project that was surprising to you?
Well, to me, I think from the nutrient perspective, what I found surprising is that the way local food self-reliance can be defined is subject to some debate. You could either, for example, consider a cow that is grazing in the Okanagan, and getting additional feed from imported feed, you might consider that local production. Or you may say, “No, well, actually, the feed is coming from somewhere else. It's not local.” In a similar way, if you produce chocolate in a country that doesn't have cocoa, or you could say it's a local product, because it's produced there, or it's not local, because to cacao comes from somewhere else. I mean, it's a similar discussion, but really, what surprised me is those assumptions like whether you make one or the other have a tremendous impact on how things pan out in terms of nutrient accumulation. So that's something I wouldn't have expected to be that strong.
So, as modelers, like you're probably when you're reading or looking at other people's models, you're looking for the analytical framing, the assumptions they're making? Are you also reading for values? Because you seem to be modelers who are tuned to this way of thinking? Or do you look for an underlying values and think through how they might have shifted different models?
Yeah, I think after a while you, you kind of get used to recognizing different patterns. And I think one very important thing that determines also the outcome a lot is what you think about human behavior, and to what extent that can be changed or not? Because we’ve done some very optimistic scenarios and modeling exercises where we assume that we can change diets very drastically. And you can get fantastic results when it comes to greenhouse gas emissions. But, I mean, how realistic is that? So that's on the one extreme, and then on the other extreme, you can find scenarios and modeling exercises that just keep diets constant. Because people won't change. We see if you look at trends, when people get richer, they want to eat more livestock products. So maybe somewhere in between is maybe where we hopefully will end up. But I think how you see, what is actually possible to change in terms of human behavior, that's really crucial in scenario development. And also, how much trust you put in technology? That's the other one. So how much of these environmental problems can we solve with technology? So if we assume that in a few years, we'll have feed additives to give to ruminants so that they don't emit so much methane, or we can keep them in the barn and clean the air? I mean, then we can have lots of ruminants, of course. So yeah, that's the other really important value or belief in in, in both technology and behavior change, I think you can see that quite easily, like how the scenarios are set up based on this.
It would still be in my experience some kind of inference in a sense that I mean, the more you read and understand different perspectives, the more you get a feeling of which direction it goes. But sometimes I do find it difficult myself, because it's rarely explicitly articulated what the values are. I mean, if it's like modeling assumption, like we chose this parameter over this parameter, because of we know that the data quality is better, or it's more representative. These are often very clearly articulated. But when it comes to the values, I mean, even like do you see humans as a part of nature or something that is above nature? I mean, these are the core underpinnings I think, that have a very strong influence, but they're not articulated in that sense. So I would agree with Elin that that you do get a feeling of, well, it tends to be this way or that way. But you can't always be sure whether you correctly interpret because it's is not articulated. Now, that's what I find most difficult.
That is such an interesting observation. It makes me wonder if these papers could benefit, you know, you have a method section. Could they benefit from a value section? Johan, you said that it's really hard to identify your own values, but I wonder if it's something that could be useful to surface?
Yeah, in social science areas. You would sometimes include a specific section in the papers that makes these authors’ values explicit. the paper.
So some of these “values” sections might take the form of a positionality statement, where the researcher acknowledges that their background and life experiences influences how they see, understand and engage with the world. Even the exercise of writing the statement also forces the researcher to reflect on their own values.
And I think, really, modeling is close to Social Sciences in a way. It's very quantitative and very tight natural sciences. But it's still so much as we have discussed influenced by different values. And it is really society that we are modeling. So it's really close in a way to social sciences. And I think it could be beneficial, and also good for the researchers to be forced to think more about your own values and how that might influence what you do and what you don't.
Yeah, I think this is really interesting to think about how this would play out, because I think the values that we are aware of, those are less of a problem. At least for me, I'm aware of some things that I have very strong opinions on. And then I find myself going always in the opposite direction, because I'm so aware of this. So I don't want this to influence my work. So I sometimes I even get some feedback like oh, you could you could go even further here to state your conclusions, but I've been so afraid of doing that, because I know that I have this strong opinion. So I'm more worried about these things that I don't realize myself and how they influence my work. But of course, I mean, it's always good to try to reflect on why do I take this make this choice now? Where does it come from? And can I really justify this now.
I think an interesting exercise here is what was once suggested to me by a professor in Vancouver, whom I work with Gunilla Öberg. She said like, of course, we as researchers, we are kind of trained to spot in consistencies. And maybe you should be as well, you should apply as much scrutiny towards your own work and really try to, to basically challenge your own thinking in a similar way, like you would challenge other people's thinking. It made me realize that there's a lot of assumptions and a lot of values that I at that point wasn't aware of myself.
For me, it would be an interesting thing to do not to surface values to say, we're removing these biases, but just to sort of do the classic Science and Technology Studies acknowledgement that we all have these values. And we're trying to have these technical arguments that are actually values arguments. And so if we can shift the focus and be able to discuss these things, we might be able to have a more robust conversation.
I think we all are more or less coalescing around this as an issue, as something that would be better if it was surfaced and addressed. And a question for you, Samara is, you’re engaged in a lot of these conversations now with different organizations that are aiming to transform the food system. And you're also looking at the values that they're bringing to the table. So I wonder what types of exercises or practices might be employed to try to achieve the surfacing of these values?
I mean, I was really intrigued by the concept that the participatory modeling process could be more of a deliberative process in terms of helping people to not only surface their underlying assumptions but see how those underlying assumptions shape models. Because often in these food systems debates that I'm sort of still in meshed in right now people are talking past each other, and not actually having a productive dialogue in terms of changing their own opinions or shifting or learning things.
I ask this also, as this is part of TABLE’s mission. So I'm just curious to hear your all of your views. What would it effectively change? How do you think deliberative processes would look different if the values of people participating were made more explicit?
It might be opening up for questioning narratives in terms of what does the model actually cover? What is considered relevant? Is the story that we're trying to explore, is it plausible given the results. I think with a participatory process, if you have pluralistic views, if you have people with different values and with different perceptions of what is important, I really think that could help really scrutinizing those pre analytical choices much better than if you just have a rather closed process.
I really agree with you that this sort of having pluralistic values, included in stakeholder groups for instance, is really an important way to perhaps diversify what we model and how we do it. But I also like, just from my own experience, working with this project in the Nordic countries, where it was a really tight group of NGOs, they were really similar. And they had been working together before. So they were sort of on track, this is what we want to do. And I think this was an important success factor that it worked out as well as it did. I think it is also really a challenge if you have a really diverse group of stakeholders to make them sit down and collaboratively sketch out future visions, I think it's really important work. But I also can see that it might be really, really tricky.
We spoke earlier with Jessica Duncan, who reflected that dialogue processes aren't meant specifically for consensus. They're supposed to engage with conflict and to find areas of compromise and find areas where compromise is not possible. One thing I wanted to touch on is what’s not included into the models. And how, what isn’t included, how that might affect results. I think Johan, you have an example of the Nordic food system process, where someone posted the question to you why reindeers weren’t included in the model. Can you speak a little more about that experience?
Yeah, exactly. We had this number of seminars presenting the work we did on the Nordic countries. And on one of the seminars there was this person who was representative of the Sami people. He asked me, “so why didn't you include reindeer in this. It’s a really important food producing activity happening in the Nordic countries?” And I really didn't have any good answer to why we didn't. And I think this was really influenced by partly because both us as researchers and the different representatives of the NGOs were all originating from sort of the more south-central parts of Nordic countries, while reindeer herding, is done the Sami people further north. So that was one influence. But I think there's also sort of larger things at play looking at the historic suppression of Sami people in Sweden and other Nordic countries. And after I got this comment, I started thinking a bit more and I started looking into statistics and then I realized that reindeers are actually not included in the Swedish Agricultural Statistics next to the other farm animals. And the exact reasons why this is so, I'm not sure. But I definitely think it has sort of historic values influence how statistics organizations are constructed and what we collect statistics on and what we don't.
Yeah, that’s really interesting, and I guess they do, either Robin or Elin have any examples of what is it Other models are what just specifically difficult to incorporate into a model that might be an important factor.
I would say there is three reasons that I see why something would not be included in the model. It's either because pre-analytically, it's being considered irrelevant or not relevant enough to be modeled. And then if you come to the actual modeling process, other things that might constrain something making its way into the model is: you have no data, or poor data, or you're lacking a conceptual understanding of how to actually model. How to describe what's happening with a certain process. I can give you an example in the Okanagan, well, like everywhere else, there is about 17 plant nutrients that might be potentially of interest, we only had data for nitrogen, phosphorus, potassium, and magnesium. And that's also only because we did a little trick with using data from other similar areas, not the Okanagan. So that were the four nutrients we considered and all the other 13, we had no data, so they didn't make it into the model. These are the three broad reasons that that I see. And sometimes it might be problematic, and sometimes it might be fine. That really depends.
And returning to the point of data availability that Robin mentioned, it's like this different food system models are usually based on national statistics. Because that's the data we have, since we can't really go out and measure in any other way, specifically for a research project. And these statistics that are collected are not collected for research purposes. The background to their collection is more sort of perhaps you have some agricultural subsidy schemes that you need to monitor and therefore you collect statistics on different things. And also it tends to be these more commodified crops and animal products that are in statistics. And therefore these more commodified crops and animal products are the ones that get modeled. So I think it's sort of a lock in risk. Looking at these models, where we model the current industrially produced crops and livestock products, but we don't include the products that are perhaps today more niche in the models. Because there is no data for them for the more rare types of crops and the more rare animal species and breeds. Yeah, I think there is some risk of sort of lock into current structure of food systems.
Thank you all this has been super, super interesting. I'm going to ruminate on the question Matthew you asked about values, I have some ideas. I'll ruminate without releasing any methane. Sorry, that was a dumb food system joke. Oh, my goodness. (laughs). We just wanted to wrap up with the final question. Because you are people who have dedicated your time to doing models, we wanted to ask you, why are models important? Why do you keep doing them? And in what way should we be using them to make the most sense and how they can play a role in impacting the world?
From my sides, I see models as tools to help us structure our thinking. Ideally, in a coherent way. So if it's done in a rigorous way, where you both challenge your pre analytical assumptions, but also your analytical modeling choices, I really think it can help exploring different courses of action, like what would happen if we do action A or action B. Maybe also surface some connections that we might not intuitively comprehend. So that, for me is a very important outcome of modeling exercises.
But I'm also convinced that the full potential of models can only be achieved if it's like a process where you bring in different ideas of how the reality works, different simplifications of the complex reality. Otherwise, you may end up with a perfectly beautiful model analytically, but it might be questioned in terms of its relevance by other people who have a different understanding of how the world operates and have a whole different set of values.
So for me, it's a tool to facilitate discussions. It doesn't replace a process where people together figure out what to do. It's not like a model can tell you do X or Y. I don't think that's, that's what models are supposed to do. In my opinion, it's more like helping that process. But in the end, it's human beings who have to take the decisions, assisted by models, but not just because the model says X or Y this is what we do.
I agree with Robin. And to add to that, I think the end use of all of this work doing food systems models, is so for me, at least always that they somehow end up in the public debate and shape policy decisions. So somehow, the end purpose is always to change the future I think, in one way or the other. But then, in the ideal situation, stakeholders shaping public debates and policy need to rely on both these sort of quantitative models that we are talking about, but also on a lot of other ways to assess how a better future can be shaped.
Yeah, I think models can help us - the types of models that we're using, like these biophysical models can show you what options you have, and the magnitude of the impacts of different choices you make. And then you will need to add a layer of what do we want on top of that. They can only tell you this is what you get based on different choices you make. But this discussion of what do we actually want how it has to be done in parallel.
Thank you all so much for your, for your time. And the interesting discussion.
Yes, thank you. That was a great conversation.
And that wraps another episode of the Feed podcast. A big thanks again to the researchers at the Swedish University of Agricultural Sciences, Elin Röös, who is also one of TABLE’s research directors, Johan Karlsson and Robin Harder. Links to the articles we discussed in the conversation and many more other useful food systems resources are available on our website, https://tabledebates.org/
You can subscribe to Fodder, our weekly newsletter, to stay up to date with the latest news and opportunities on food systems sustainability topics.
TABLE is hiring three new key staff members. You can find links to the job descriptions and how to apply on our website. Deadline is Sept 6 2021.
If you enjoy listening to the show, you can support our efforts by rating and reviewing us on Apple podcast or wherever you listen. It really helps us reach new listeners!
This episode was edited and mixed by Matthew Kessler. Music in this episode by Blue dot sessions. We’ll be back in a few weeks wrapping up our Scale theme, sharing our findings on different ideas about how local or global our food system should be.