1. Introduction
The term
foodshed [
1], like the term watershed, is used to describe a geographic area, in this case, one within which food flows from sources to consumers. Foodsheds have been described at scales ranging from local to global. Concerning the foodshed, the use of the term
power is varied but does follow some common threads. A review of the existing and relevant literature from public health and food systems research showed that having access to food was historically an indication of power [
2]. Over time, power became associated with the ability to control the production, processing, and movement of foods and the resources required for their making. Other writings use
power to refer more pointedly to an individual’s or organization’s ability to influence policy change and provide control over some aspect of the food system [
3,
4,
5,
6,
7,
8,
9,
10]. This influence may come from control over the built environment [
6], having an economic advantage [
8], and/or the ability of those with influence to maintain the standing they have in a situation [
9]. Ostrom defined power as a function of the value of the opportunity and the extent of the control [
11]. The term
agency, which has a long history in institutional and development economics (viz, Cheung’s framing of sharecropping contracts as principal-agent problems [
12]), connotes an individual’s ability to advocate [
4] for themselves and to act as a semi-autonomous agent within social-constructed systems of economic and political constraints.
Different disciplines have emphasized different, yet interrelated, aspects of the role of power in the food system. The role of differences in the level of power between small farmers and large corporations [
4,
8] or between farm laborers and landowners within the foodshed has been a central focus of agrarian studies, particularly about food sovereignty. For instance, in Argentina, there are many farm laborers whose livelihoods depend on landowners’ willingness to hire and pay them [
13]. This is not an uncommon phenomenon; it is also an issue for migrant workers in North America [
14], where employers exert power over undocumented workers, leveraging the threat of deportation. Similarly, power and related conflicts are common themes in the literature on natural resource management and international agricultural development across the Global South [
15,
16]. In social work, power relations are typically described on a local or individual scale in terms of oppression and empowerment of the oppressed [
13]. From a political economy perspective, the role of power during policy formulation and implementation [
2] is a central concern. Take, for instance, a large corporation or company’s opposition to policy changes that may force them to change product formulations to be more nutritious but, instead, the corporation leverages its power to deploy the “part of the solution” strategy. In the “part of the solution” strategy, companies (e.g., Mars, Unilever, Kellogg’s, and Coca-Cola) have pushed back to acquire supposedly healthy brands and made commitments to reformulate their products to reduce sugar or sodium in their products [
17]. Similarly, several wealthier companies control consumer choice architecture and can use their financial power to exert political power, allowing them to influence policy, often to block change [
8,
18]. In political science and ecology (with the perspective of looking at ecology and society, rather than biotic aspects of the environment), interdependencies within systems (e.g., political, administrative, and social-ecological) occur due to power differences [
19]. A common thread among these different perspectives on power from different fields of study is that power mediates changes in the system including the type, speed, timing, and scale of the changes, which type of changes occur or do not occur, how policy changes are implemented and by whom, and who benefits and who loses.
Power imbalances are ubiquitous across the food system, crossing local, regional, national, and global scales. These imbalances have consequences, often leading to adverse social, economic, and health outcomes, political marginalization, impoverishment, and exploitation of those with less power [
2], and reinforcing inequalities and injustices, leading to social tension and, sometimes, civil unrest [
4,
13]. In the case of
Tareferos, a group of farmworkers in Argentina, exploitation by farm owners meant the laborers had limited access to healthcare or retirement funds. These problems are common for farm workers and small-scale farmers around the world. For farmers in Honduras, loss of land meant loss of food sovereignty and food insecurity. Power imbalances and policy choices often are mutually reinforcing. In the United States, “redlining”, which is a structurally racist policy meant to keep Black people and other racial/ethnic minorities out of neighborhoods where White people live, have perpetuated social, political, and economic inequalities and played a role in the development and spatial arrangement of “food deserts” and other areas with reduced food security [
6]. A study undertaken in Australia found that power imbalances in favor of supermarkets negatively affected the nutritional value of the food available to consumers, impacting public health outcomes [
7]. In short, power imbalances in food systems (and other systems, too) create a variety of complex, far-reaching problems, many of which can be difficult to identify, understand, and address. Failure to address these manifestations of power imbalances perpetuates social justice and arguably undermines food system sustainability and resilience [
20].
Redistribution of power is always challenging, if not impossible. In public health, power imbalances have been identified as a barrier to deliberations about the community health [
21], which makes it harder to even start the conversation about redistribution of power within a community. A prerequisite to addressing these challenges within the food system is the engagement of diverse voices and the participation of marginalized stakeholders. Community engagement is important for addressing power imbalances, because it challenges the structural powers that oppress marginalized stakeholders, and shifts the decision-making authority from those with power to those with less or no power [
22]. Multiple factors need to be considered to create opportunities (sometimes called “safe spaces”) where it is possible to truly hear diverse voices and engage in collective efforts aimed toward positive change. Creating such opportunities in a foodshed is complicated by several factors. One factor is that foodsheds are complex. There are many stakeholders within the foodshed, including producers (e.g., farmers), processors (e.g., butchers), distributors (e.g., transportation), retailers (e.g., retail food stores), consumers (e.g., households) and resource management (e.g., food waste recovery). Each of these stakeholders is affected directly by challenges in the foodshed and could benefit (or suffer) from changes in the food system. Another factor is the relationship between equity and sustainability and how this relationship affects each stakeholder and interactions between stakeholders in the foodshed. Somewhat paradoxically, the pursuit of a more equitable food system may involve tradeoffs with sustainability, and vice-versa [
23]. Additionally, consumers and producers with higher income and more political and economic connections (social capital) have greater power than those with lower income and fewer connections [
20]. Those with power typically are the voices that dominate discourses around the food system and health policy [
24,
25]. As a result of these factors, it is difficult to amplify the voices of those with less power and include their views in an authentic manner. Empowering the powerless in the food system by including them in the design, governance, operation, and policies that shapes foodsheds, including both the structure and function of food systems, may lead to a more sustainable and sustainable foodshed.
Understanding what power means to each stakeholder in a foodshed is a key step towards identifying existing power relationships. Identifying existing power relationships and reimagining those relationships may help to achieve a more equitable and resilient food system. Of course, it is the exercise of power in the pursuit of narrow self-interest and in opposition to broader common interests (say in public health, childhood nutrition, environmental health) that is the canonical problem underpinning so many policy failures. Through the development of methods and tools for understanding these relationships and convening and lifting marginalized voices, the long-term goal is to mobilize food system stakeholders for the identification and development of collaborative solutions that shift outcomes toward the broader public interest [
20,
24,
25].
One tool for bringing together diverse voices is the use of community-based system dynamics, specifically group model building. The community-based system dynamics approach was developed as part of system dynamics modeling, which is a methodology from the field of systems science [
26,
27,
28,
29,
30,
31,
32]. In the community-based system dynamics approach, the dynamics, feedbacks, and nonlinearities within a system are identified by community participants using facilitated discussions and group activities [
33]. Bringing together these diverse voices of participants allows participants to develop solutions to problems within a system that they care about and work together in, but may not know how each organization’s decisions or policies affect other organizations and, eventually, the entire system [
33]. This approach to problem-solving prioritizes inclusion and collaboration with multiple stakeholders, with the goal of co-creating solutions to challenges that benefit multiple stakeholders. These solutions ideally transcend the power differences concentrated in the vested interests of each organization or stakeholder by finding a common good. Perhaps, more realistically, and in the context of a foodshed, the hope is to mitigate the impacts of power imbalances by engaging diverse voices and mobilizing marginalized communities to address food system challenges that injure them directly.
Examples of the community-based system dynamics approach, in particular, group model building, in the food system literature include studies on food insecurity [
34], food consumption [
35], obesity [
32,
36], food supply and demand [
37], food-related policymaking [
38], and program evaluation [
39]. In each of these studies, input for the modeling was gathered from a variety of stakeholders through facilitated discussions. The facilitated discussions generated various types of boundary objects. In some cases when data were used to build a system dynamics model, the authors used those boundary objects and translated model-driven insights into policy actions. A boundary object has many definitions [
40], but we are using it here to mean textual and graphical representations of ideas, perspectives, and mental models of group model building workshop participants [
29]. Boundary objects may serve to identify new ways to build data discovery and data sharing capabilities among food system stakeholders.
The community-based system dynamics approach offers a way to develop boundary objects in a manner that mitigates power imbalances and, in the long-term, may facilitate data discovery and data sharing among stakeholders. Food system stakeholders often do not know about each other’s data holdings, which prevents them from asking new questions or sharing their data with others. If we assume that all food system stakeholders have some power to share data, then it is reasonable to also assume that the level of power varies and remains unknown to other food system stakeholders. Based on Ostrom’s definition of power, the value of the opportunity (i.e., sharing data) and the extent of control (i.e., the capacity to share data) is, perhaps, most appropriate to bring up again because food system stakeholders may vary in how they value data sharing. Similarly, the extent of control to share data may vary due to different capacities and access to resources needed to share data. Therefore, identifying power relationships among data holders and re-orienting those relationships might be one way to enable more data sharing among food system stakeholders to solve foodshed challenges. Identifying such relationships using a methodology that minimizes power imbalances among stakeholders is a novel feature of the community-based system dynamics approach.
While current studies on data sharing via data commons or shared vocabularies using ontologies (e.g., FoodON [
41]) mainly focus on developing the infrastructure for data sharing, little attention has been paid to the role of power in data sharing among food system stakeholders. To make progress in addressing this problem, we set out to conduct a group model building workshop with food system stakeholders to start to develop a class called ‘Power’ within an existing ontology, known as PPOD for “People, Project, Organizations, and Datasets” [
1,
42,
43]. An
ontology is a set of concepts within a specific area or domain, which describes the properties of those concepts and the relationships between the concepts. A
class is a core element comprising an ontology [
44]. The PPOD ontology was developed for foodshed and food systems research and applications. Ultimately, we aim to integrate the class Power within the PPOD ontology. A novel feature of our study is extending the community based system dynamics approach, specifically group model building scripts, for enhancing the PPOD ontology. This is a novel feature, because typically ontology development is an academic exercise where community stakeholders are not directly involved in the process [
45].
The addition of the class Power is a foundational step in eliminating barriers to including diverse voices in shaping and solving challenges in the foodshed through data sharing, transparency, discovery, data-informed debate, and co-creation of data-enabled solutions. It is conceivable that such a classification system could enable the re-imagining of existing power relationships within the food systems to make progress towards greater equity, diversity, transparency, and accountability of food system actors. It is hoped that our work can provide tools to enhance organizing and collective action to confront powerful vested interests standing in opposition to the public good.
3. Results
The workshop produced a variety of boundary objects in the form of visualizations, notes, and graphs. We digitized these boundary objects for reporting purposes. We did not change any wording in any of the boundary objects shown below, but added clarification to some of the words where necessary.
From the first activity, a concept map emerged showing the relationships among production, composition and variability, and health (
Figure 1). Relationships between each concept were indicated by lines, but we did not ask participants about the direction or magnitude of the relationship. Our main objective for generating a concept map was to engage participants from diverse backgrounds and experiences. We also sought to help participants visually understand how the concepts that they were familiar with were related to concepts from other disciplines or domains. Based on the concept map, which we left up on a separate whiteboard for participants to view and refer to throughout the workshop, participants identified 34 different stakeholders with varying levels of power and interest in addressing the research question (
Figure 2).
The power vs. interest graph was the output from the second activity of the workshop. The graph showed that lobbyists and the food industry had the highest power and lowest interest (top left in
Figure 2), low-income communities had the highest interest and lowest level of power (bottom right in
Figure 2), and logistic firms, legislators, and consumers had the highest power with the highest interest (top right in
Figure 2). Participants grouped the 34 stakeholders that they identified into six broad categories of stakeholder. These stakeholder categories were business (industry), business (farmers), government/regulators, healthcare/public health, consumers/activists/civil societies, and labor (
Table 1, top row). See note in
Table 1 for the difference between business (industry) and business (farmers).
One output from the third activity (
Figure 3), which further simplified the six stakeholder categories from
Table 1, identified feedback loops among business, government, and consumer stakeholders. Specifically, participants identified how consumers influenced businesses through their money and their voice and influenced government through their money, voice, and votes. Voice [
48] was broadly meant to convey the social capital of a stakeholder category that the stakeholder may use to influence the actions of another stakeholder. Examples of how consumers used their voice to influence other stakeholders included consumer feedback, online reviews, and advocacy. Participants also highlighted that business stakeholders influenced consumers through products and marketing and influenced government through votes. Examples of votes were lobbying efforts and other ways that businesses influenced government (broadly including agencies, elected and unelected officials, and regulators) activities, decisions, and policies. Lastly, we observed in
Figure 3 that the government influenced both businesses and consumers through its regulatory authority. These broad types of influences or expressions of power, such as money, votes, and voice, among stakeholders, were investigated further in the last workshop activity.
Another output from the third activity identified the current and future influences or power relationships among the six broad stakeholder categories (
Figure 4). Through this activity, the diagram of current relationships (black arrows in
Figure 4) showed that business (industry) influenced the greatest number of stakeholder groups (e.g., government, farmers, and laborers). Comparatively, the influences of business (farmers), laborers, consumers, and healthcare sectors are currently limited. In terms of future relationships (red arrows in
Figure 4) that may lead to a more sustainable and equitable foodshed, we observed that most participants indicated a need for greater power and influence for consumers, workers, and the health sector vis-a-vis government and business (industry) stakeholders.
The last output from the third activity was the set of statements that further clarified the relationship, in terms of power, agency, and influence, among the six broad stakeholder categories (
Figure 5). Based on statements from participants, we noted twenty different statements in the following format: noun → verb → noun, where nouns were objects (e.g., stakeholders) and verbs described the nature of the relationship between two objects in terms of current power relationships. The most common verbs were “funding” and “regulating”. Less common verbs were “demonstrating”, “advising” and “training”. Based on these verbs, the participants generated five broad categories to describe the nature of the power relationships among groups of food system stakeholders: providing the infrastructure; regulation and mandates, media action; providing funding; and withholding resources.
4. Discussion
In the context of the initial research question, our results highlighted current imbalances in power relationships among food system stakeholders. Although the question was specific to certain components of the food system (e.g., food composition and nutrient availability), the concept map (
Figure 1) identified connections among many different parts of the larger food system, and key stakeholders varied in their levels of power and interest (
Figure 2). Going one step further, the information that we collected on current relationships among food system stakeholders (
Figure 3 and
Figure 4) highlighted the importance of increasing the level of power among key stakeholders, such as consumers, workers or labor, and the health sector. Our findings (in
Figure 2 and
Figure 3) highlighted how those who currently have higher levels of power may be able to reinforce their level of power through the positive feedback loops (
Figure 3). Collectively, these findings point toward a path for re-orienting current power relationships (
Figure 4) in a manner that could lead to more equitable and sustainable solutions in food systems.
An insight from the workshop was the diverse set of words used by participants to describe how power or influence is practically applied among stakeholders. For example, the statement in
Figure 5 that “Government regulates food content for consumers” identifies the role of regulation in terms of how a government entity or stakeholder (e.g., Food and Drug Administration) has direct power and influence over food manufacturers and indirectly on consumers (e.g., people who eat food that is regulated for its quality, safety, and nutritional value by the government). Another example is the statement “Advertising food influences choice through media of consumers”. In this statement, participants suggested that it was the choice of consumers that is influenced by food advertisers through any of the several types of media (e.g., news, print, social). The words “choice”, “regulate” and words like them, respectively, including other tenses (e.g., choose) or synonyms (e.g., control or police), provide a useful starting point for developing a class Power within the PPOD ontology. Another way to go about creating the shared vocabulary for the class Power would be for developers of PPOD to look up synonyms for the “power” from the dictionary or select concepts from the relevant literature. This alternative approach would have been much less time-consuming than organizing a workshop. In contrast, we engaged directly with food system stakeholders and empowered them to identify the nouns and verbs by “workshopping” concepts with participants and, thereby, creating a shared vocabulary for the class Power. Our approach started with community engagement rather than the typical research paradigm of generating knowledge and solutions and then going to the community to test their usefulness and utility in real-world settings.
Our current study is an initial attempt to define the class Power within PPOD. We envision a future scenario where a fully developed and instantiated version of the class Power within the PPOD ontology may allow food system researchers and stakeholders alike to elucidate the power relationships between organizations and people. Knowledge about such relationships may be used to understand how power relationships may promote or block the sharing of datasets among food system stakeholders. We plan to use the class Power based on the initial set of verbs gathered through the last workshop activity (
Figure 5) and enhance the vocabulary for the class Power (and PPOD) through additional testing with stakeholders through similar workshops, interviews, and focus groups. The following are two examples of how such a scenario could play out in real-world settings.
For the first example, imagine a food system stakeholder (provisionally, call them Organization A) who seeks to answer a question for which they do not have data. Organization A could use a PPOD-based visualization to identify relevant datasets to their question and identify relationships between themselves and other organizations (provisionally, call them the set of organizations B = {B1, B2,…, Bi}, where i is the total number of organizations that have the data sought by Organization A. Using the class Power in PPOD, the type of influence between organizations A and each Bi could be visualized by Organization A to identify which Bi has power relationships that are more likely to lead to the sharing of data sought by Organization A. To extend this example to food system researchers, the class Power in PPOD may allow them to identify which power relationships may promote or make it more difficult to share data among food system actors. The strength of the PPOD ontology is that, in the future, it may allow for such an analysis to be constrained to specific organizations, datasets, people, and projects.
Another example comes from building upon prior work [
49] of some of the authors of the current study. In prior work, these authors linked issues and indicators related to sustainable sourcing using the PPOD ontology. In the future, a fully developed class Power for PPOD may be used to identify the power relationships among organizations around a specific sustainability issue(s) or indicator(s). If the identified power relationships are one-sided or non-existent or weak, then collaborative and trust-building activities may be needed to facilitate data sharing around a specific sustainability issue or indicator. Such activities could be conducted by food system researchers or community stakeholders who specifically focus on a specific sustainability issue or track specific sustainability indicators.
Our study had several limitations. First, we conducted the workshop as part of a conference organized by the International Center for Food Ontology Operability Data and Semantics (IC-FOODS). Therefore, the participants who attended the conference and were self-selected to participate in the workshop are not representative of the full set of stakeholders in the food system. Additionally, our participant recruitment strategy may have biased the set of concepts, stakeholders, description of power relationships, and verbs generated to describe the nature of the power relationship. Although we acknowledge this limitation, it is also worth noting that the purpose of the workshop was to test whether the community-based system dynamics approach could be modified to generate knowledge that would ultimately be used to define the class Power within the PPOD ontology. Therefore, we plan to supplement the initial set of knowledge products/boundary objects generated during the workshop described here with ongoing work that will involve more representative food system stakeholders. Another limitation of our study was the lack of follow-up with participants to validate our interpretation of the boundary objects that they co-created during the workshop. This limitation was mainly due to the nature and format of the workshop, which was a half-day workshop at the end of a multi-day conference. One consequence of this limitation may be that our interpretation of the verbs identified in
Figure 5 may not be reliable because we did not seek further clarification from participants about the reasons behind verbs and nouns they wrote about during the workshop activity. Lastly, a limitation of our study was that the research question was generated by the study team rather than the workshop participants. As a result, it is possible that some participants may not have been very familiar with the issues and topics. This may have been a source of confusion in participant responses during workshop activities, such as the concept mapping activity.