September 01, 2023

The “Immaculate Conception of Data” – and why it’s a problem

Episode 2 of 'Who will control the food system?'
Food Barons report cover art showing peasants resisting corporate digital giants

Tune into the next episode in our latest podcast mini-series, Who Will Control the Food System, where we uncover just who's pulling the strings of industrial agriculture, dissect the latest corporate strategies, and take inspiration from the peoples and movements fighting back.

In this second episode, Zahra Moloo talks to Kelly Bronson, a social scientist at the University of Ottawa in Canada, about her research into the secretive legal agreements surrounding agricultural big data, to trace how it is used and with what consequences. In particular, what happens when big data is embedded in pre-existing arrangements of power and corporate strategies?

Take tractors, for example. ‘Digital’ tractors are not like the vehicles of times past. They have built-in sensors that collect data and stream it to cloud-based infrastructure. Critically, the digital business model means that the farmer does not own the tractor, or the software that is embedded in the tractor, or even the data that is generated by the equipment. Rather, when a farmer purchases a tractor from a farm machinery company such as John Deere the farmer only receives a “license to operate the vehicle.” It is the company, Deere, that owns all of it.

Not only that, but the farmer also has to pay (in addition to paying to use the tractor) for automated data services or data support services that will provide him with technical advice – which the farmer must follow – on what, when and how to plant in his own field. 

What is this data that is generated from the tractor? How is data more generally captured in the context of agriculture? Who uses it? Why doesn’t the farmer own it? 

In this second episode Zahra Moloo and Kelly Bronson talk about Bronson's new book, “The Immaculate Conception of Data: Agribusiness, Activists, and Their Shared Politics of the Future.”

Listen in as we explore these questions!

To find out more about the digitalisation of food and agriculture you can also watch our animation “Big Brother is Coming to the Farm: the Digital Takeover of Food” (available here in Arabic, Bahasa Indonesia, Bisaya, English, Filipino, French, Italian, Portuguese, Spanish and Swahili – and with a version in Hindi on the way).



(Note that the transcript has been lightly edited to make it easier to read.)

The “Immaculate Conception of Data” – and why it’s a problem

The food we eat can go through a long journey before arriving on our tables, from farms and fields, to markets or supermarkets. And then to our homes, each stage in this journey can look very different, depending on who controls the process. On a farm, is it a small-scale farmer who plants and harvests your produce? Where do they get their seeds from? Do they share seeds amongst themselves as they have done for thousands of years? Or do they buy them from companies? Why, during the pandemic, did nearly a billion people go hungry while food and agriculture giants made exorbitant profits? In this podcast series, we will look at who controls our food systems and how those trends are changing.

We will investigate which companies are starting to take greater control of commercial seeds, farm machinery and grocery retail. We will look at how and why big tech companies like Amazon, Microsoft, Alphabet, Google and Alibaba are moving into food. And we will examine new trends unknown to most people, like carbon farming and digital platforms. Join us as we take a look at who controls and who will control our food systems.

When a farmer purchases a tractor from the farm machinery company John Deere, he receives a license to operate the vehicle. These days, tractors are not like the old clunky vehicles of times past. They have built in sensors that collect data and stream it to cloud-based infrastructure. The farmer does not own the tractor or the software that is embedded in the tractor, or even the data that is generated by the equipment. It is the company, John Deere, that owns all of it. Not only that, but in addition to paying for the tractor, the farmer also has to pay for automated data services, or data support services, that will provide him with technical advice on what, when and how to plant in his own field. These tractors are mostly used in the Global North such as in the US, but they are also being increasingly deployed and promoted on big farms in the South.

What is this data generated from the tractor? How is data more generally captured in the context of agriculture? Who uses it? Why doesn't the farmer own it?

I am Zahra Moloo. In this second episode of Who Will Control the Food System, I will speak to Kelly Bronson, a social scientist at the University of Ottawa in Canada. Bronson is the author of the Immaculate Conception of Data: Agribusiness, Activists and their Shared Politics of the Future. Her book looks behind the secretive legal agreements surrounding agricultural big data, mostly in the US and Canada, to trace how it is used and with what consequences.

Kelly Bronson, thank you for speaking with me today on Who Will Control the Food System. Let me begin with an obvious question about the title of your book: The Immaculate Conception of Data. In immaculate conception, as we know from the Catholic Church, is the idea that the Virgin Mary, the mother of Jesus, was free from the original sin of Adam. What exactly do you mean by the Immaculate Conception of Data?


I  thought of calling the book: Big data, big power. A big part of what I tried to do in the book is peek behind the technologies and the legal agreements and other kinds of social infrastructures surrounding these new digital tractors and other kinds of digitized farm equipment to get at who was doing what with data, and for whose gain. But in the end, I called the book The Immaculate Conception of Data, borrowing from this religious story or metaphor because the book is really grounded empirically in my own kind of data collection: I spent time with data scientists working in the private sector for companies like John Deere, but also Bayer Monsanto, Farmers Edge. I also spent time with scientists working in public contexts: for the Canadian government, for the US government, but also with activists, self-labeled farmer hacktivists, who were playing in the space of using data to drive food system reformation. Initially, when I set out to write the book, I kind of knew these communities, some scientists in the private sector, and I knew the farm hacktivists and activists a bit more because to be totally honest, that's where my interest lays, my kind of food politics.

I sort of assumed that there was this model, we could call it “digital agriculture”, that was more on an industrial scale, like John Deere's model: selling tractors to mostly commodity growers, as you mentioned, mostly across North America, intended for export markets; industrial agriculture. And then I thought that there might be another model for the realization of a kind of data-driven agriculture, which is more activist, facing local food systems, fostering practices like agroecology or regenerative agriculture.

At the beginning of the book, I would sort of compare these two systems and people working in these two spaces. But then I noticed, and this is where the title comes from, that even though there are these pretty real differences in how the tools are being used, which kinds of data are being collected, from whom and for whose benefit, I noticed there was a similarity. And that was that across the board, everyone was talking about data in this really particular way, which is as if data on their own were a thing that just dropped from the sky, as if they were immaculately conceived, instead of being brought into being, by powerful companies for example.

And people were talking about data as if they are agential. Borrowing from that religious metaphor, that sort of story, confers Mary and God a lot of power. Because if you can just do something and make something happen, that's quite powerful, that kind of omnipotence is a part of that religious story I wanted to get at as well, using the immaculate conception of data. Because even the activists who are really working (and I'm on their side) for food system reformation, were really talking about data as if the data themselves were going to get us somewhere. I think that way of talking about data, using this framework, the immaculate conception of data, is really problematic because it renders less visible the really powerful actors and interests. Let's just call it what it is, the corporations who have for a long time really dictated the directions, technological but also social and environmental, of the food system.


And so you mentioned this term “data-driven”, we hear this all the time, just as ubiquitous now: this data-driven this, data-driven that, and I'm sure there are other terms that you also explore in your book that point to this idea of data having its own agency. From your book and your research, did you find that this narrative, or these terms, were being used in an intentional way, or has it just become a way of thinking that has sort of fallen upon us?


As part of the research process is that one looks systematically at the data, using an approach where you can sort of generate meaning from your field research in a pretty systematic fashion. And it wasn't until I started to do that, it's called coding, that I noticed this common way of speaking. I kind of assumed, these were just terms of phrases that have become ubiquitous, fallen upon us as you put it. I quite like that. This is just part of the kind of landscape now, or the lexicon around digital systems, and it's actually part of the lexicon that exceeds agriculture. So the immaculate conception of data framework is actually everywhere. You'll probably start to notice it after me having drawn attention to it.

But then I realized, at least in the context of those people with whom I was spending time, that the answer to your question is: “it's tactical”. So this is what I mean. I realized that none of the scientists, none of the activist scientists or the private sector scientists, none of the business people at conventions - because I did go to a lot of agricultural technology and also just AI and data technology conventions - none of the participants would probably hold true to the premises behind the immaculate conception of data. Nobody would say: oh, yes, of course, data are just come to be. They're not natural.

You know, in fact, it was these very same people who were working tirelessly and showing me the very complex and difficult work they were doing to bring data into being, to clean data, it's so-called cleaning data, to manage data infrastructures, to make meaning from data.

And so I started to realize it was actually part of a kind of rhetorical strategy. Talking about data using the immaculate conception is a way of talking about data that works, because if something is just brought into being, then the person who is part of that system is considered omnipotent, or very powerful. And so I realized it was part of a kind of longstanding cultural approach, actually, long before data, so-called big data or artificial intelligence, long before these tools, applied to food production.

We've had, at least in the western world, an approach toward technology where if one can speak for a technological system as free from politics, as immaculately conceived, then that's conferred a lot of power to people speaking for this technology. It's been a winning strategy, and so I give stories in the book, drawing from speeches at the colonizing of the Americas, where the so-called founding fathers, or the “civilizing politicians/technologists” in North America had a very similar immaculate conception way of talking about technologies like the telegraph, for example, or the rail line, these powerful agents of change that were going to help civilize the Americas. And that's very problematic. I'm not saying this was a good way of talking, but it's a way of talking that worked to bring people on to a particular approach.


I would like to go into more of the details of data, how it's collected and how it's used. You mentioned the John Deere tractor. So to take this example, the data that's taken from this tractor and other data that's taken from the farms is now considered a very valuable resource from the world of agriculture. So what exactly is the content of this data? Why is it so valuable?


I think one important thing to point out is the broader context where people - both in the public realm but also scholarly, were thinking about companies like Facebook, and what they were doing with personal data. I started to notice in 2016 and 2017 that there was not really a conversation around who is collecting data and for whose gain, or even for what purposes, pertaining to what we might call environmental data - and I would include in that agricultural data. But a lot of the attention, popularly and scholarly, is given to data that's collected on people and collected from participation in online environments like social media. I started to notice that at a time when there was a really ballooning public conversation around the money that was to be made from the collection and use - and even misuse of these data. If we think of the Cambridge Analytica scandal, the gaming of the electoral system in the US or the UK, around the initial Trump election and the use of Facebook data to do that… meanwhile there were very few people paying attention to environmental or agricultural data.

There are a number of ways the data are collected. They're collected from what's called precision agricultural equipment. So new tractors today are digital vehicles. They're licensed: they're not owned, just like our cell phones. Farmers who buy a new John Deere tractor, they have to also sign a license agreement to use the data systems on that tractor. And the tractor passively collects data from the start. It’s like when you enter an online environment, data are collected on your whereabouts, even on your scrolling behavior, and most of us know about this.

Similarly, in a tractor, the moment the farm worker or laborer opens the tractor door, data are collected on the laborer, the whereabouts, the tractors movements. Data are collected from the environment on the crop, the planting, the soil, moisture, pH, chemical makeup… But also more broadly, those data are brought together with other environmental data into systems that are meant to generate advice for farmers on how to manage the farm and in particular, where to plant in the field, how to manage chemicals, when to plant, when to seed, et cetera.


So one example of these precision agricultural equipment is the tractor. What are some of the other equipment that is collecting data from the farm?


I talked with a lot of scientists and even if you go to John Deere’s smart farming website, you can see there's a great kind of visual of all of the different sources of data that the company brings together to generate so-called data-driven advice, or advice that's generated from the coalition of data across sources using machine learning or AI systems for farmers. So here are some other equipment, drones, for example. Not every farm would use drone technology, but many in these envisioned smart farm environments do.

The sources of farm data are systems - and they're in many cases private systems - of weather data collection or other environmental data collection. For example, the Canadian or the publicly traded company Farmers Edge, have their own private system of weather stations that participating farms use to help collect data that then get aggregated across farms. Those data are then combined with satellite data. Again, in many cases, these companies are partnering with private satellite companies, like Planet Labs, or they're buying up public or just using without even needing to pay, they're using publicly collected environmental data and combining those data with the data from these privately owned or licensed data collection sources to generate what is called big data.

The uses are to generate advice for farmers. One beneficiary there is the farmer who's using or, as you said at the beginning, paying for - and that's one distinction between agricultural data use and AI use versus social media - we don't pay for our participation on Facebook. But farmers pay several times: they pay for the equipment that collects the data, they sign a license agreement, and they also then have to pay for the so-called data-driven advice. They will get what's called precision advice: Where exactly in my field needs attention in terms of chemical attention, for example.

So farmers are one beneficiary from this kind of data use, used with these AI systems. So, Farmers Edge has My Farm Manager, which is an app accessed on a tablet or a smartphone that would help generate maps that are specific to farms and also advice that's specific to particular farm operators or farm environments.

But there are other uses too. First of all, there's a potential obvious conflict of interest. So for example, I found in a close reading of the terms of use agreements and a close reading of what these different AI systems advice for farmers that Monsanto or Bayer Monsanto calls the advice prescriptions. In the so-called AI-driven or data-driven prescriptions written for farmers, the systems only suggest to farmers using chemicals that are already part of the ecosystem of products within a corporation. So there's an obvious vested interest or conflict of interest, where the AI system is generating advice for farmers that they ought to use a particular chemical also sold by the system generating the advice.

In the book, in trying to do this work of figuring out how these systems work, I found that I could not validate the algorithms that generate the advice. Those are all proprietary. And I actually couldn't even talk with the scientists who are building them. In some cases, scientists did speak with me. In some cases, I had to sign a non-disclosure agreement. But trade secrecy law was also preventing a full disclosure of exactly the steps by which this advice gets generated. So that's one potential, sort of social and environmental issue: the fact that there's a potential vested interest in reproducing particular use of chemicals that make these companies very wealthy, and that also we know have harmful social and environmental impacts in the food system.

But the other uses of these data exceed the farm environment. It's really hard to find evidence for this. So I have to be really careful in the claims that I'm making, because it's very hard to trace exactly what's being done with the data. But we know that data have been transferred among input supply companies across the food system or food chain. We know there have been data transferred between John Deere and Monsanto Company. We can infer, based on what we know from, for example, the use of the selling of data by social media companies to advertisers, that the data themselves are an asset.

So we know that companies who collect the data, like the equipment manufacturers, stand to gain a lot of money by selling packaged data systems to, for example, another kind of input supply company, like a chemical company. And I had actually someone in industry say to me that this was being done, that we know that these data sets are going to be valuable for insurance companies, and in particular, reinsurance companies, like Lloyd's of London or Swiss, Ag Swiss, that make a lot of money from loss, from farmers' misfortune in the global food system, by being able to use modeling and AI to predict loss. Therefore, they literally bank on farmer misfortune. So those are some of the uses of data across the food system.


Do you know from your research what kind of relationships, what kind of legal and corporate frameworks would allow the data to be shared across different companies? For instance, you write about how Bayer has the capability to access data from almost half of all the farmers in North America. What are the kind of relationships that would allow, let's say, Deere to share that data with Bayer Monsanto?


Maybe there are two important things to point out there. There's the legal infrastructure in particular that allows companies like John Deere to collect data and use them in particular ways without compensating farmers, without being completely transparent to farmers or other food system actors about the uses or potential misuses of data. And those legal mechanisms are intellectual property or copyright mechanisms, and in particular, terms of use or license agreements.

And now farmers have to sign these agreements if they're going to use these systems, just like we have to sign away these privacy agreements for our participation online. One sort of side note is that these agreements are really difficult to read, but those are the legal mechanisms that allow companies to do things with the data that might not always be in the interest of the person who generates the data, the farmer in this case.

I found in these terms of use that there was some evidence, especially if I looked across years, in changes in the language in these legal agreements where there was clearly room being made in these legal agreements for uses of data like the selling of data to third parties, for example, transfer of data from an equipment manufacturer to a chemical manufacturer.

And so that's an important site actually for thinking about power in the food system in answer to who controls the food system and how. There's really important work to be done by legal actors and activism to amend this legal infrastructure. This is actually one way in which these seemingly brand new or revolutionary data systems and AI systems in agriculture really follow from historic agricultural technologies, namely GMOs or genetically modified organisms. It is the license agreements around seed systems that really allow for some of the misuses of this system in terms of environmental justice.


What's really quite shocking about this is, as you mentioned before, is that when we use online digital platforms like Facebook, we don't pay to use that data. But farmers are both providing the data and having to pay to use their own data. So is there any way that the public or farmers can access what is being done with their data in this context of a seemingly one-sided flow of data from farmers to companies?


No, I don't think so. And that is, again, a slight difference between the social media space and the farming space. There aren't really good ways right now for a farmer to access those data. There has been some visibility around this via the conversation around the right to repair. I haven't found, actually, in my reading of the legal mechanisms that surround these digital tools, any sort of way that farmers or others can easily access both the data, but also, even myself - as a critical researcher, from a scientific or technical perspective - it's almost impossible to validate these AI systems because people don't have access. The legal scholar Frank Pascal calls it a "black box", or I think he says a "pernicious black box" surrounding data and AI systems.


We're going to take a short break. When we're back, we'll be speaking more about the digitized futures imagined by actors in the industrial food system and also at how activists imagine a digitized future.

Welcome back to Who Will Control the Food System. This is our second episode, and we are speaking with Kelly Bronsen, social scientist at the University of Ottawa. She's talking to us about her book: The Immaculate Conception of Data. You mentioned that you write about this digitized future imagined by actors in the industrial food system, and you present a picture of what that would look like. Can you describe what you mean by a digitized future? I mean, what would that look like?


To describe or to get at that vision of the future, what I would call the dominant vision or the one being promoted across the food system by really powerful actors, whether it's agricultural input supply companies like Bayer Monsanto or equipment manufacturers like John Deere, but also groups of academics, the folks working technically in this space. For example, there's a whole journal devoted to so-called digital agriculture. It's called Precision Agriculture. The academics working in that space are promoting this vision, the dominant vision. There are also supranational policy organizations that we know are really powerful in setting agendas for the global food system like the UN FAO or the World Bank. I would say they promote this vision too.

This vision is one where data are collected on farms, whether it's through crowdsourced data via smartphone use or whether it's through these precision tractors or precision equipment, combined with satellite data. All of is this is fed to cloud-based infrastructure and AI systems are used to make sense of these data. Some people think of that suite of tools used together as the fully realized smart farm. Some academics describe this as Farm 4.0. It's the kind of idea that farming or food production will be perfected in the future using these digital tools.

The vision is really that these tools will be used and that farming will be made more precise. Digital tools, and AI in particular are a perfection of human reasoning or human cognitive capacity. In particular, people argue that it will lead to a kind of productivity gain. Farmers are basically going to out-compete their neighbors because they're going to make more money through a productivity gain.

But there's also another promise and it's an environmental one. If you look closely, as I did in the book, at the promotion, the promise is that farmers are going to make more money and produce more food, but they're going to do so with less environmental impact. And this is a really interesting message that I explore in the book. The head of so-called smart agriculture at Bayer Monsanto said that in the future, they were not going to make money from inputs at all. They were going to have replaced inputs in farming with information, which actually begs the question of how they're going to function as input supply companies. The message is really that farmers are going to make more judicious use of harmful inputs like agricultural chemicals (such as Roundup Ready), or water, because they're going to make more precise decisions informed by the digital tools.

And so this is the vision that farming will continue business-as-usual in terms of a continuation of the industrial model: commodity crops grown for export market. But it's an amendment of the system. So I would say that's what I would call the dominant vision of the future of farming under digital tools like data and AI.


That does sound like a very seductive vision if you look at both the idea of productivity and of environmental sustainability. It seems like quite a few people, farmers, activists, when faced with this sort of critique of digital agriculture are kind of curious whether there's any truth to this. Can these digital technologies and data actually be useful for productivity or for environmental good? Is there any truth to this seductive vision?


I think there is some evidence that there is a productivity gain in yield and there is some evidence - but not a ton - that there is potential for an amendment specifically around chemical use and impact on, for example, groundwater. This is not because of a reduction in the use of chemicals, but because as one soil scientist in the private sector said to me: “no, no, by precision, we don't mean that it's less chemicals, but it's more precise use of chemicals”. I think that's one place where maybe we see some evidence.

I would say there's not nearly enough evidence to support the really widespread enthusiasm and the funding behind these so-called digital agricultural systems. One thing that I try to say explicitly, especially in the last chapter of the book, is that I think we really need to be cautious: I'm of the mind - and this is my food politics - that we need more than just a reformation of the global industrial model of food production, but we need a broader and more radical change, a disruption in the dominant mode of food production. We need more small farmers, we need more support for peasant farming, we need more biodiversity and investment in historic tried and true techniques like polyculture, cover cropping, and agroecological and regenerative agriculture.

And so there's reason to believe that actually the promotion and the use of these data and AI systems in agriculture will actually prevent food system trajectories toward a broader amendment of our current dominant system because they're biased toward the global industrial model. Agriculture is just one example. In the same way, we know that Google search is biased toward particular bodies. If you look at manliness, you'll get a particular way that Google search combs data because of gender biases, or racial biases in the data that feed Google search. Similarly, if you even just look at a company like Bayer Monsanto, the data are only collected on agronomic crops.


That actually brings me to a question that I found based on something you said in your book that was very interesting. You say that one would imagine that activists in the US and Canada would have oppositional ideas about industrial agriculture, but you say that activists also have an imaginary farm - a digital farm in the future. So can you describe this? What does a digitized future of food production look like to activists who you would imagine would be oppositional to this idea of a digitized future and to the industrial food system?


I really only spent time with farm activists who were excited about data systems already.

So that's a kind of bias actually in my research design. But actually at the outset of the book, I was looking for people in the agricultural space, activists or scholars who were critical of these systems and it was hard to find them. There wasn't a whole lot of critical attention. And so I did end up spending time with these activists who were excited about these systems. And yeah, what's their vision of the future under these systems? Well, it's radically different from the dominant vision in terms of the kind of agricultural system they think so-called digital agriculture data and AI and agriculture will support.

The activists I spent time with in Quebec are a group called Autoconstruction, which is part of a broader agricultural cooperative called Capé, a group in the US called Farm Hack, and another related group called the Gathering for Open Agricultural Technologies or GOAT. These groups really envision that data can be used to disrupt the dominant industrial model of agriculture by supporting alternative agricultural systems like regenerative agriculture, like agroecology. In fact, there is an open source platform called FarmOS that's written open source with the participation of farmers that is totally different from the systems being developed by companies like Bayer Monsanto.

This farm management system can use data and AI to help direct farm advice and collect data on different crops, not agronomic. It collects data on compost piles or manure heaps. In that sense, the vision is different. In the kind of agricultural system that these activists imagine, the digital helping or serving is totally different. The model of using the data, or the ethos, I would say, surrounding the use of data and AI is totally different from the private sector. The activists are motivated by open data sharing, open access, copyright commons. They're working with farmers and incorporating a diversity of farmers’ feedback. And yet, as you pointed out, the subtitle of my book is : Agribusiness, Activists and their Shared Politics of the Future. This is something that really surprised me that the activists do share in one thing with the agribusiness. A shared way of talking about data, back to the immaculate conception of data, as if they just dropped from the sky and as if data - as opposed to all of the people and the powerful interests behind data - are going to get us to the future.

I felt really conflicted in pointing this out in the book, because my politics are really with the Farm Hack folks and the activist folks. But I felt it was important to point out because I really do believe that the words we use and the ways we conceive things or imagine things, the way we talk about things matters. And I could see that in talking about data as immaculately conceived, that the activists in a way are working at cross-purposes with their own interests because if we talk about data as if they just dropped from the sky, then we're not asking questions. And by “we”, I mean all of us, consumers, scholars, activists, we're not asking questions about, okay, who collected those data? Where did the data go? How are they managed? Who's making money from them?


The other thing that you mentioned in your book that I was very interested in hearing you talk about more is that farmers in North America seem only moderately interested in what agribusiness companies are doing with their personal data. And you say that for years, being a successful farmer in North America has meant behaving like a corporation. Can you explain more what you mean by this?


As you said at the outset, the book really does focus largely on a global northern - and in particular a North American context. But I would say that there has been more attention, and certainly today, in the Global South around data and data justice and data colonialism, as we might call it. Who's using data for what purposes and who's gained, including in agriculture?

But yeah, I was surprised when I looked in North America at farmers, especially those who had literally bought into these data systems. Largely in Canada and in the US, it is about farmers such as in the Midwest who are operating on thousands of acres or hectares across that kind of land scale of operation. They are the farmers who were really buying into this suite of technologies, what some academics and industry call the fully realized smart farm or farm 4.0. However, I would say that there isn't a really widespread adoption of this fully realized smart farm in Canada or anywhere in the world. Canada, the US and Australia are maybe ahead on this among some farmers, but even compared to, for example, genetically modified organisms, the suite of digital tools have not found widespread application.

So I did find something curious: I kind of thought that the farmers would be resistant in a way to the uses of the collection of data, that they would be suspicious of the companies potentially collecting the data and then using the data without compensating the farmer. But I didn't find that. What I found is that the farmers who had bought into these systems were actually pretty fine with the company. They didn't really want to necessarily access the data behind these systems because they have sort of bought into the logic of farming as a business. They see themselves as a busy farm operator or manager, as opposed to necessarily a hands-on farm worker. They were pretty content to just deputize and offload the sort of data management capabilities to the in-house data scientists and AI expertise in a company like Bayer Monsanto.

That was different from the activists. In fact, all those farmers who were using the full suite of digital technologies, they didn't seem especially suspicious or troubled by the corporate uses or potential misuses of their data, but they did tell me they were worried about government oversight. One of them said: “I'm worried about big brother, that these data could then get into the hands of government actors”, and he didn't really want that. They didn't want a kind of oversight by government actors over farm operations.


Very interesting. It was quite surprising to hear farmers not being concerned about big companies taking their data, but I suppose that would be very different from activists in the South or farmers in the South who are championing agroecology and such approaches.


Absolutely. If you look globally, there's all sorts of activism around data misuses and ownership of data under frameworks like data colonialism or indigenous data sovereignty that's really being spearheaded by Via Campesina, and other peasant farming organizations and groups from the Global South.

What I found was a surprising lack of suspicion among the farmers who have bought into these systems, part of it comes back to that question about who's collecting the data. And if we think about the history of corporate agriculture or industrial agriculture, companies like John Deere, Bayer Monsanto or Monsanto have courted relationships with farmers. They've developed these relationships of trust with farmers. When I talk to farmers in North America, if you're a Deere man, you're a Deere man, and sorry for the sexist language. It's a cultural relationship and in the space of data and artificial intelligence, this is a particular advantage for these companies because we know that whoever is collecting the most data and the earliest has an advantage when it comes to machine learning and AI. If we think about systems like Alexa or Siri, they have a market advantage because machine learning and other kinds of artificial intelligence get better the more data they're so-called fed.

And back to your question about how these data are being used, I really saw in my reading of the license agreements and the terms of use that these companies are now using data to further these very particular client relationships. So for example, the license agreements, just like those for GMOs, really disincentivize a farmer from moving from one company system to another. The tractor’s data system is not interoperable with a different AI system by a different input supply company. So John Deere's system is not interoperable with other systems. And that locks them in.


Are there any developments in terms of regulating the way in which data is being used? Where we are now with this whole digital infrastructure in agriculture and this extraction of data, what is there to counter what's happening with big data?


Not much. For Canada, for example, we've got systems of voluntary regulation, meaning that companies themselves regulate, using their own internal policies, which you can read about. And I did so for the book, reading the corporate websites. Those are visible and made visible to potential users. But again, there's only a kind of modicum of choice there. Because if you're already a Deere user, you're in some ways locked into that system. And actually, if you buy a new tractor, it's necessarily a digital tractor. It's necessarily collecting data. So you can't even really opt out.

But the other way that the companies voluntarily “regulate” is through these legal terms of use or licensing agreements. And so that's why reading those is really important. That's the regulation system we currently have. There's really nothing internationally that is cracking down on these companies in terms of potential misuses of the data. One could imagine ways in which, for example, competition law would be used to crack down on input supply companies. I mean, they're already oligopoly companies.

The sale of Monsanto to Bayer was almost disallowed by the US courts. One could imagine how the US courts would have taken an anti-competition stance and disallowed that acquisition, but they didn't. That would have been a really important move in the sale of Monsanto to Bayer in 2018, the biggest acquisition in all of history.

Competition law could be used similarly to try some of these companies to push for things like interoperability, or to push for things like a lack of transfer of data between powerful input supply companies, especially for uses in terms of reinsurance. I feel like governments across the board, not just in terms of agriculture, have been caught sitting on their hands or not quite keeping up to pace with the pace of development in terms of AI.

With the drop of chatGPT in November for example, governments are trying to catch up and develop legislation, particular for big data and AI. In Canada right now, we have a proposed Bill C27 for data and a proposed AI and data legislation, but neither in Bill C27 nor in the AI and data legislation, do we have anything particular to agriculture. And it's very similar across the board. There's maybe a little bit more proposed regulation or legislation in Germany and in other jurisdictions in the EU than there is in North America, but I would say across the globe, currently, it's no actual regulation, and it's company self-policing. And so, that's one site for activism. I would say that's something we really need attention on.

Thank you so much, Kelly Bronson. That concludes our second episode in our podcast mini series Who Will Control the Food System. To find out more about agriculture and big data, please visit ETC Group's website, ETC Group also has an animation which breaks down the digitalization of agriculture, called Big Brother is Coming to the Farm, which you can find on our website. You can find all our podcast episodes on the ETC Group website or on Spotify on ETC Group podcasts.


Please consider supporting ETC's unique research and advocacy with a tax-deductible donation. Donate here