Looking more closely at the emergence of smart farming developments, another crucial consideration pertains to the innovation processes that try to generate and integrate new and striking configurations of firms, farmers, and research institutes. New networks of diverse stakeholders are formed; extant networks are re-made, potentially rendering invisible the role of some actors and their interests. In a sense, then, smart farming can resemble the notion of “innovation by withdrawal”,which departs from the view that innovation is “structured around the introduction of a new element, an artefact, a way of operating, a service, and its success is dependent on the number of adopters and the significance of the entities which are articulated with it” . Yet, as demonstrated by the case of no-till farming in France , withdrawal of one element relies on making visible hitherto invisible or overlooked elements , while maintaining problematic practices.To respond, an alternative to focusing on introduction or withdrawal is to recognize that agricultural innovation, like any other practice, is always a topological affair: it is about overseeing and managing configurations of humans and materials and how they flow through a system or across a specific domain, such as a field . Where there are blockages, conduits can be installed to increase flows; where there are leaks, plugs are required. Farmers shift and prod to adjust arrangements of materials or relations with a view to addressing problems, such as falling yields, vulnerabilities to climatic variability, or exposure to viruses. A mechanism of implementing innovations such as those associated with smart farming is to reconfigure topological arrangements. As demonstrated by literature on agricultural biosecurity,farming practice needs to be viewed as occurring against the topological backdrop of an “entangled interplay” , with numerous “contingent intra-actions” occurring across multiple risky “borderlands” .Proximities,distance,nft system connectivities,and modulations of presence/absence figure in the effort to create desired outcomes,with new insertions or removals pursued in efforts to control or steer activities in defined or experimental ways.

Whether the risk is a matter of falling yields or exposure to viruses, it pays to acknowledge the ongoing relationship between farming and innovation through a topological lens; that is, to dwell on farmers as active agents of topological transformation, even if they are rarely acting alone. One way to combine an analytical concern with digital life, corporate interests in establishing a certain type of smart farming, and topological transformations is highlighted by literature on smart farming innovation processes. Consider, for example, how new configurations come into the picture. As highlighted in research on smart farming in Canada, and as referenced by Relf-Eckstein et al.,the Canadian government has recently established innovation ‘superclusters’ to examine and exploit technological opportunities. A recent outcome is an industry-led consortia called Protein Industries Canada, which includes a partnership whereby Lucent BioSciences “will use the hulls of pea and lentil seeds which are a co-product from value-added processing completed by AGT Foods and Ingredients [to create] Soileos: a novel carbon-neutral micro-nutrient fertilizer that uses organic fibre as a carrier to provide micro-nutrients to plants” . As this case suggests, the ‘smart’ in smart farming can involve astute and imaginative arrangements to make new products and chase after profits in novel ways. A similar picture emerges in the Netherlands where a “golden triangle” of agricultural research, industry, and government aims to create “new business ecosystems consisting of focal firms, their suppliers, complementor firms, and customers” . A key feature is the leading role of the Dutch firm Philips, which occupies a prominent position in high-tech urban agriculture , a growing smart farming sector, by “providing the essential technologies, registration of patents, and creation of new business opportunities” . Meanwhile, in the larger and more traditional Dutch agricultural sector, the Food Valley Open Innovation Ecosystem includes “the Wageningen Campus and the planned World Food Centre in Ede” and creates ties between research and development centres run by large firms such as Friesland Campina and Unilever and wider networks of small-to-medium enterprises and startups.

There are 15,000 scientists across Food Valley, with twenty research institutes, 1440 food related, and 70 science related firms . Such configurations of firms, farmers and research institutes will likely create new smart farming products and services and build on Dutch successes in exporting around €9 billion worth of high-tech agrifood innovations, including “energy-efficient greenhouses, precision agricultural systems and new discoveries that make crops more resistant” . Put differently, the topologies of smart farming point toward new forms of “path creation” that involve but also often extend beyond farmers. It is instructive that smart farming today is bound up with efforts to use ‘open innovation’ processes that facilitate co-design or co-innovation between agricultural technology providers, farmers, and others. Such an approach can “blur the boundaries between scientists and agricultural system stakeholders, between agronomists and farmers, and between actors in the agricultural sector and those designing in other sectors” . The virtue of “participatory design processes involving farmers” is that it can yield new tools, such as dashboards , to help farmers understand agro-ecological conditions. There are, however, no guarantees that smart farming developments will yield effective configurations. Indeed, there is significant evidence that smart farming developments are hamstrung not only by the instrumental logics underpinning technology providers but also by ineffective coordination and inadequate arrangements of materials or skills. Consider here the push to develop automated body condition scoring and a soil water outlook tool for Australian dairy farmers. Noting that these versions of smart farming innovation involve “a unique innovation challenge [not least because of] the new knowledge demands for farmers in a highly dynamic, technology-driven environment” , one finding is that the new tools and practices confront limitations in the way agricultural relations are configured with respect to the wider institutional milieu. Making the most of the soil water outlook tool, for example, required but did not receive sufficient input from the Australian Bureau of Meteorology “to help farmers to link the SWO with seasonal climate outlooks” . Then, with regards to automated body condition scoring, the new technology led some farmers to think “maybe we don’t need the [farm] advisor as often,”,with the upshot that “some tools were potentially replacing the skills of advisors”.Yet, because “more remote monitoring of key performance indicator data via online software” can enable farm advisors to make fewer farm visits, smart farming in this context conceivably increases the sense of isolation many farmers already experience .

Elsewhere in Australia, smart farming developments call attention to a different dynamic between farmers and advisors. In some rice farming regions, advisors might be expected to be the “sense makers” who can explain and encourage farmers to adopt new technologies; but in fact one consultant respondent claimed “it’s mainly been the farmers dragging the agronomists along” .At the same time, “insufficient support structures” , for example regarding data compatibility or standards, can hold back adoption and frustrate farmers who are “prepared to use evolving and uncertainty-generating technologies” but find that their knowledge is not effectively tapped. In Canada, ‘broadacre’ smart farming developments occur amid the “critical constraint” of labour shortages and demographic change, but even here “adoption is lower than anticipated” , with one explanation focusing on tensions around what happens to data produced on farms. A problem yet to be overcome is industry self-regulation of data usage and a lack of certainty about the legal ramifications of smart farming. Thus, “[u]ntil clarity is brought to the issue of data, the industry is at risk of losing farmer’s trust and potentially hindering innovation opportunities at the farm level” . Although there are examples from the literature which demonstrate that smart farming innovation involves an ongoing process of trying to reconfigure arrangements of sociotechnical relations, I argue a more accurate and urgent conclusion is to emphasize the ‘mis-configured innovations’ of smart farming. One of the main features of smart farming concerns the limited parameters within which innovations operate. In Canada, for example, an element in smart farming arrangements is models and platforms designed for commodity farmers, not those “farmers working outside of the dominant industrial model” . In effect, “the maps created within those big data platforms developed by industry are made meaningful only if one adheres to a rigid conventional farming strategy of seeding in neat rows separated by areas of soil free of weeds” . A similar result emerges in Australia where observers note that farmers want autosteer technologies, new imagery services, levelling and GPS guidance because “if they’ve got efficient layouts, laser levelled, they’ll make significant water savings and they’ll have reduced labour inputs as well” . Smart farming therefore means that food producers contemplate, “standardizing the environment” in accordance with the commercial imperatives of farmers operating large holdings and using expensive machinery to generate predictable topographies that fit with the new topologies required to make smart farming technologies effective. Built-in biases pervade all algorithmic systems ; the biases in smart farming might only pertain to environments in the first place but they can have broader political-economic effects.

As such, the core problem with the various reconfigurations underpinning smart farming developments is not simply that the absence of one or other action or reform can limit their impact,hydroponic gutter but rather that smart farming innovation processes begin and proceed without adequately conceptualizing the underlying obstacles and limitations confronting food producers today. Technological innovations that reinforce power asymmetries regarding data ownership, for example, or that fail to challenge implicit biases toward certain types of environments, render some interests invisible while reifying specific types of logics, such as narrow measures of economic efficiency. Like any innovation, insertion, or reconfiguration, smart farming entails topological transformation; but problems emerge when the “quieter registers” of smart farming make it possible for “powerful actors to make their presence felt at one remove, to reach into the everyday life of distant others” , for example by dispossessing them of valuable data or establishing algorithmic biases toward standardized farm topographies.It can make sense to use devices or services in new arrangements that create efficiencies or give food producers new access to information that can inform decisions. However, because these developments always by necessity involve reconfiguring arrangements of sociotechnical relations, agricultural innovation processes will continue to introduce new misconfigurations when they pursue discrete solutions to specific problems, rather than integrated developments based on incremental adjustments in information-intensive iterative processes that target systemic or structural change. As insisted upon by scholarship on food sovereignty in critical agrarian studies , the urgent challenge today is to conceptualize a planetary land, agrarian, and food system in which food producers and consumers everywhere are confronted by, but examine ways of overcoming, the same problems of neoliberal capitalism dominated by transnational corporations, authoritarian governance, and climate change. In the shadow of the corporate food regime, producing food in the Netherlands or Canada is bound up with the realities of producing food in India or Kenya. Further, the dynamics of digital life mean smart farming innovations in one place will inform and conceivably move the ‘planetary cognitive ecology’ generally, with unpredictable but connected results playing out elsewhere. The products of smart farming will only reinforce problems if they yield new patents for agricultural technology providers in a place such as Ireland , a widening yield gap between capital- and labour-intensive agrarian systems , or if they increase the likelihood of ‘smart’ food production in one region leading to food dumping in another . Per the vision of a Common Food Policy in the European Union , rather than seeing smart farming developments “reinforcing existing production models, leading to trade-offs between different environmental impacts, or between environmental and social sustainability,” the task is to reorient innovation “towards low-input, diversified agroecological systems” . In the light of these challenges, a sustainable and successful smart farming innovation process requires what we might imagine as the coproduction of ambitious ‘topological repertoires’ that make ongoing assessments of absence, presence, proximity, and reach at the scale of a structure or system and then pursue appropriate technological solutions from the ground up.