Innovation, like evolution, is a process of constantly discovering ways of rearranging the world into forms that are unlikely to arise by chance—and that happen to be useful . . . . [I]nnovation is the most important fact about the modern world, but one of the least well understood. . . . The striking thing about innovation is how mysterious it still is. No [scientist] can fully explain why innovation happens, let alone why it happens when and where it does.

—Matt Ridley, How Innovation Works (2020)

Technological and institutional innovations in agri-food systems (AFSs)Footnote 1 over the past century have brought dramatic advances in human well-being worldwide. Yet these gains increasingly appear unsustainable due to massive, adverse spillover effects on climate, natural environment, public health and nutrition, and social justice (Barrett 2021). How can humanity innovate further to bring about AFS transformations that can sustain and expand past progress, while making them healthier for all people and for the planet that must sustain current and future generations?

Recent scientific studies of global AFSs bring out clearly the challenges we face. Some emphasize the environmental and climate unsustainability of AFSs (GloPan 2016, 2020; IPCC 2019; IPBES 2019; Willett et al. 2019). Given projected growth in human populations and incomes, and the headwinds of the climate and extinction crises, satisfying future aggregate demand for food will put unprecedented pressures on finite water, land, genetic, and atmospheric resources. The risks of enormous and potentially irreversible ecological damage are no longer under serious scholarly dispute. Moreover, beyond the longer-run pressures wrought by inevitable food-demand growth, building evidence raises concerns about AFSs’ resilience to sudden weather, environmental, disease, economic, or political shocks. Such shocks appear to be rising in frequency and/or intensity, and commonly cascade, with one triggering another (Maystadt and Ecker 2014; Von Uexkull et al. 2016). And shocks to AFSs increasingly appear to feed sociopolitical instability around the world in a potentially vicious cycle (Barrett 2013).

Other recent studies point to AFSs’ failure to advance the well-being of all persons, in at least two distinct ways (GloPan 2016, 2020; Haddad et al. 2016; FAO 2019; HLPE 2020). First, today’s AFSs fail to ensure healthy diets for all—a necessary condition for food security.Footnote 2 Second, AFSs do not provide equitable and inclusive livelihoods for the roughly half of the world’s labor force—more than 1.3 billion people (ILO 2015)—who work in agri-food value chains (AVCs).Footnote 3 Far too many people who labor on farms, in processing facilities, groceries, restaurants, or elsewhere within our AFSs fail to earn a living income or to control essential resources such as land, or risk serious injury or illness, or are victims of forced labor. Women, indigenous populations, racial and religious minorities, and young people are disproportionately disadvantaged for a variety of systemic reasons. Despite the unprecedented productivity and prosperity enabled by technological advances and institutional and policy reforms in global AFSs over the past century, far too many people still face chronic or episodic undernutrition, diet-related health risks are a growing problem, and AFS jobs are among the most dangerous and exploitation-prone on the planet.

The COVID-19 pandemic has laid bare previously under-recognized fragilities that pose yet another hidden cost of the modern AFS: uninsured risk of catastrophic disruptions. Past, sometimes-single-minded pursuit of lower food production costs and consumer prices brought valuable efficiency gains. But it has also led to such AVC specialization and concentration based on economies of scale and scope that many producers and sub-systems struggled to adjust to a massive systemic shock. Advances in logistics and market integration enabled reasonably quick stabilization of food supplies and prices in most places. But we should understand the COVID-19 pandemic as a warning shot across the bow of AFSs. As scientists expect natural and manmade shocks to grow in frequency and severity, enhancing AFS resilience grows ever more urgent and may entail building in some redundancy as systemic insurance (Webb et al. 2020).

This book originated as a report commissioned by the Cornell Atkinson Center for Sustainability in response to an invitation from the journal Nature Sustainability, which—in collaboration with its new sister journal, Nature Food—wanted to devote its 2020 expert panel to this topic.Footnote 4 The panel brought together experts who come from many different continents and who span a wide range of disciplines and organizations—from industry and universities to social movements, governments, philanthropies, institutional and venture capital investors, and multilateral agencies.

The panel synthesized the best current science to describe the present state of the world’s AFSs and key external drivers of AFS changes over the next 25–50 years, as well as tease out key lessons from the COVID-19 pandemic experience this year. As is increasingly widely recognized, the costs that farmers and downstream value chain actors incur and the prices consumers pay understate foods’ true costs to society once one accounts for adverse environmental, health, and social spillover effects. Inevitable demographic, economic, and climate change in the coming decades will catastrophically aggravate these problems under business-as-usual scenarios. Innovations will be needed to facilitate concerted, coordinated efforts to transition to more healthy, equitable, resilient, and sustainable AFSs.

In deliberating about needed innovations, the panel concluded that four key AFS features must continuously remain front-of-mind: decentralized individual and collective human (H) agency that drives systemic change, the intrinsic heterogeneity (H) of AFSs locally and globally, pervasive spillover (S) effects, and the essential role of scientific (S) research. Attention to these HHSS (pronounced “his”) attributes is essential to avoid adverse unintended consequences and make real progress.

The panel then developed a shared vision for the AFSs of 2045–70, beyond the 2030 horizon of the UN Sustainable Development Goals (SDGs). We summarize that vision in four core AFS objectives: healthy (H) and nutritious diets, equitable (E) and inclusive value chains, resilience (R) to shocks and stressors, and climate and environmental sustainability (S), summarized in the acronym HERS. AFSs are immutably HHSS. The task is to make them equally HERS. Failure to address the HERS objectives risks catastrophic failure, even existential threats, under business-as-usual scenarios. Faced with multiple, high-level, pressing objectives, AFS adaptations cannot attend only to unidimensional concerns, whether about climate, environment, health, employment, equity, productivity, or resilience. Both tradeoffs and synergies exist among these design objectives. For that reason, among others, we therefore need bundled responses to address looming challenges and to realize the considerable promise of a rich pipeline of emergent technologies, a portfolio to deliver on multiple objectives that no one innovation can simultaneously satisfy.

With a shared assessment of current state—and of inexorable drivers of AFS change—and a shared vision of desired future state firmly in mind, the panel then undertook a detailed review of scores of innovations at various stages of development and implementation.Footnote 5 The pipeline of emergent technologies is full of promise.Footnote 6 A disproportionate share of them are digital innovations, but the abundance of agronomic, genetic, mechanical, and social science advances available to advance HERS objectives is undeniable. One cannot help but conclude that existing and imminent knowledge really are not the factors limiting progress in addressing the formidable challenges facing AFSs now and in the coming years.

The limiting factor is more sociopolitical: insufficient leadership, political will, and willingness to find cooperative solutions rather than winner-take-all outcomes. All new technologies must navigate a complex maze of biophysical, political economy, and sociocultural obstacles to adapt and scale, and thus they need companion interventions to accelerate them to implementation and diffusion. Furthermore, every innovation we studied will almost surely have unintended impacts on non-target outcomes, and the resulting tradeoffs naturally spark opposition by groups concerned that change might hurt them. The panel therefore heavily emphasized the importance of coupling technical advances with social and policy change, into socio-technical innovation bundles customized to each AFS context’s needs to realize the HERS objectives. But identifying and bundling the right innovations is an intrinsically social process, one that demands cooperation that is in shorter supply than are brilliant scientific insights.

This can be summarized in the conceptualization of the AFS innovation cycle depicted in Fig. 1. Human agency drives the AFS innovation cycle. External drivers (e.g., demographic change, income growth, climate change) influence collective objectives (e.g., HERS outcomes) and actor-specific objectives (e.g., firm profits or political power) and, jointly with those objectives, induce myriad innovations by individuals and organizations. Innovations (represented by puzzle pieces) draw on different (natural or social) science-based methods (represented by different colors) to generate products, processes, or policies with distinct designs and purposes (represented by different shapes). Transformation accelerators—key enabling societal features—help AFS-specific stakeholders redirect some ill-fitting innovations back for adaptation to the local context and accelerate combination of other innovations. To become implementable and scalable, socio-technical innovation bundles need appropriate, context-dependent pieces and the right composite shape to fit local purposes. Implementation and scaling then generate feedback that affects external drivers, and in combination with those external drivers, generate outcomes. Monitoring key performance measures (KPMs) informs assessment of those outcomes—and of individual and combinatorial innovations—and helps direct adaptive management of synergies and tradeoffs among objectives, renewing the AFS innovation cycle.

Fig. 1
figure 1

The agri-food systems innovation cycle

Co-creation of socio-technical innovation bundles necessarily requires multi-party cooperation among public and private sector organizations. The panel therefore developed some process and action recommendations to guide AVC actors as we navigate together from the present, precarious state to a HERS one in our children’s future. This requires some basic rules of engagement, including discussing KPMs to monitor progress. After all, we manage what we measure. Significant public investment and trust in first-rate science will be necessary but far from sufficient. Investment increasingly turns on performance assessed, for better or worse, by KPMs. The institutional, policy, and sociocultural accelerators of technological adaptation, diffusion, and upscaling are essential complements. Hence the need for different AVC actors’ active engagement in the AFS innovation cycle.

One central message of the report is that in championing the foundational role of science and engineering to enable sustainable progress, too many high-level reports inadvertently downplay the equally crucial role of human agency (NASEM 2020). We therefore focus not only on prospective innovations but just as much on the necessary actions by actors throughout AVCs.

Change only comes about through the actions of people and the organizations they comprise. Impactful innovation can originate among actors anywhere along the food value chain, induced by any of a host of motives. So, too, can obstruction. Throughout human history, the greatest progress has come through innovation, be it in biophysiochemical technologies (e.g., improved plant and animal genetics; new medicines, transport, or computing equipment) or institutionsFootnote 7 (e.g., formal policies such as rules of tenure over land and water or contract law, or informal sociocultural practices such as cuisine). In order to harness the potential of the breathtaking pace of innovation today in digital, genetic, and other spaces, many different actors—consumers, retailers, restaurants, distributors, processors, farmers, input manufacturers, governments, charitable organizations, etc.—must engage in honest, constructive dialogues of the sort we undertook with the objective of co-designing contextually appropriate socio-technical bundles of innovations that can enable navigation away from looming dangers and towards a HERS future.

In order to enjoy HERS agri-food systems at a horizon of 25–50 years, we must invest and innovate today. We will reap then what we sow now. Innovation takes time. The lag from scientific discovery to its implementation in new technologies to productivity or other improvements at sufficient scale to be detectable in industrial, sectoral or national data is typically 15–25 years (Adams 1990; Chavas et al. 1997; Ahmadpoor and Jones 2017; Baldos et al. 2019).Footnote 8 This compels decentralized, coordinated action by public, private, and civil society actors throughout AVCs, starting immediately. Redirecting the course of AFSs presently headed towards climate, environmental, public health, and social justice disaster will require all hands on deck, working together with shared responsibility to do the hard work of navigating away from danger and towards environmentally and socially sustainable AFSs to sustain future generations. We are concerned, but ultimately optimistic that from the grim turmoil of 2020 will emerge greater unity and resolve to successfully address the systemic issues that bedevil AFSs locally and globally and that imperil our children’s and grandchildren’s futures.

Ultimately, the analysis presented in the ensuing pages culminates in seven essential actions that must guide agri-food systems transformations. In no particular order, these are:

Develop socio-technical innovation bundles: Despite the abundance of rapidly progressing innovations across all stages of AVCs today—in digital, genetic, and other spaces—no magic scientific or engineering bullets exist. Few, if any, innovations can adapt and scale effectively without essential supporting policies and institutions. Innovation is as much a social process as a scientific one, and no innovation we could identify can effectively target all four HERS objectives simultaneously. We therefore need a portfolio approach to deliver impact and to maintain necessary balance among objectives. The creative destruction of technological change inevitably generates both winners and losers, and new technologies will almost surely produce both positive and negative spillovers across HERS objectives. Co-creation of bundled approaches is therefore essential to enable packages of new technologies and practices to emerge, adapt, and diffuse to scale within, and across, contexts, and to generate beneficial impacts with limited, or no unintended, net adverse consequences.

Reduce the land and water footprint of food: Meeting future growth in food demand while reducing AFS land and water use is both necessary and inevitable. We cannot effectively tackle the climate and extinction crises and reduce the risk of zoonosis-driven pandemics without reducing AFS terrestrial and marine footprints. Decoupling food demand growth from land and water use is perhaps the most essential and challenging transition task we face. That process must be actively and cooperatively negotiated among diverse stakeholders.

Commit to co-creation with shared and verifiable responsibility: The complex pathways from innovation to scaling to impact necessitate co-creation of locally contextualized socio-technical bundles. Because human agency drives everything, all parties need incentives to act, including explicit sharing of both the responsibility to address emergent challenges and the benefits from innovation. Shared responsibilities must be matched with verifiable key performance metrics, agreed sanctions for transgressions, and safety-net protections against losses. Co-designed socio-technical bundles can accelerate human agency to facilitate, rather than obstruct, beneficial innovation and minimize unintended consequences.

Deconcentrate power: Many components of candidate solutions are well known, but impeded by concentrated economic and political power or by the marginalization of key stakeholders. The powerful can too easily obstruct progress (e.g., via catch-and-kill acquisitions, political lobbying, patent thickets). Reducing market and political power imbalances and broadening participation in innovation dialogues can accelerate innovation. Novel financing of discovery for open-source innovation, reforms of intellectual property regimes, and more robust enforcement of anti-trust laws can accelerate beneficial transitions, as can more concerted government and civil society efforts to facilitate participatory dialogues to foster co-creation of effective solutions.

Mainstream systemic risk management: The COVID-19 pandemic underscores the rising importance of building effective systemic risk management for AFSs. Most governments already appropriately mandate many forms of individual insurance (automobile, fire, health, etc.) so as to resolve market failures and avert catastrophic spillover effects. We increasingly need analogous approaches—both risk reduction and risk transfer mechanisms—to address low-probability, high-impact events (e.g., pandemics) or a combination of events (each with higher individual probabilities) that jointly cause a high-impact event (e.g., the 2007–08 food price spike).Footnote 9

Develop novel financing mechanisms: AFS innovations and systemic risk management require massive up-front investment of hundreds of billions of dollars additional resources annually. This is feasible but demands creativity, especially to mobilize private resources beyond public spending and philanthropic investments. The world is awash in investible resources, with historically low interest rates and high equity market valuations. The COVID-19 pandemic has proved that governments can quickly mobilize massive public funding when the stakes are high and solutions are urgently needed. Meanwhile, a growing community of private investors recognize the complementarities between longer-term financial and non-financial outcomes. Novel methods to mobilize the financing necessary for transforming AFSs are rapidly emerging.

Reconfigure public support for AFSs: Governments play two essential roles: investing in essential public goods and services—including basic science and education, reliable data, and appropriate, effective regulation—and facilitating dialogue to find cooperative solutions. Far too much current government agri-food spending is misspent, especially the roughly US$2 billion/day that goes to environmentally harmful farm subsidies that impede necessary innovation and disproportionately benefit better-off landowners, many of whom do not actively farm themselves. Governments must crowd in far greater private investment in AFS transformations by redirecting public resources towards social protection programs, agri-food research, and physical and institutional infrastructure (e.g., universal rural broadband access, extension services, product standards, food safety assurance). Governments also play an essential role convening civil society dialogues to facilitate discovery of, and support for, appropriate socio-technical bundles. Governments likewise must lead in co-developing and endorsing commitment frameworks, and complementary indicators and accountability mechanisms to ensure effective implementation of identified cooperative solutions at national, regional, and global scales.

Four Key Features of Agri-Food Systems and Agri-Food Value Chains

As the first-ever United Nations (UN) Food Systems summit approaches in 2021, many people and organizations are thinking carefully about how to transform contemporary AFSs to more effectively advance the 17 SDGs (Fig. 2) set in 2015 by the UN General Assembly with the intention of achieving each of them by 2030. SDG 2 (Zero Hunger) perhaps draws most attention in discussions of AFSs, but strong connections exist to virtually every one of the other 16 SDGs as well.

Fig. 2
figure 2

The 17 sustainable development goals

AFSs consist of webs of interactions among human actors, non-human organisms, and abiotic processes, with complex interlinkages across trophic scales, economic sectors, geographic space, and time. Everything that goes into growing, capturing, storing, transforming, distributing, or eating food fits within AFSs. The literature is rich with various representations of AFSs (Ericksen 2008; Global Panel on Agriculture and Food Systems for Nutrition 2016; HLPE 2017; Fanzo et al. 2020), all of which necessarily oversimplify so as to emphasize specific foci appropriate to their immediate purpose. But across the myriad AFS depictions, the four key HHSS features stand out as especially relevant when trying to promote beneficial innovations.

Inevitably decentralized decision-making within AFSs underscores the first key feature of agri-food systems: human agency. Our emphasis on AVCs follows from the centrality of decentralized exercise of human agency by actors each pursuing objectives that may, at times, conflict with one another. Command-and-control systems do not work because the interests of the powerful still prevail, even if power is conferred through political rather than market processes. Rather, societies must find ways to reconcile multiple, sometimes-competing objectives in pluralistic systems.

This often means fostering collective action. Hence the importance of mechanisms to improve coordination and align incentives, and the generation of behaviors that produce positive externalities, as well as of innovations to reduce negative externalities in those areas where coordination routinely fails. The structures and processes through which people and organizations acquire, maintain, and exercise sociopolitical power and cultural influence matter enormously to whether, and what sorts of, coordination will emerge. Hence the rising global chorus for more explicitly incorporating human agency in the conceptualization of food security, so as to elevate the right to food already recognized in treaties, including Article 25 of the 1948 Universal Declaration of Human Rights; Article 11.2 of the 1966 International Covenant on Economic, Social, and Cultural Rights; and in the constitutions of dozens of countries (Vidar et al. 2014; Gundersen 2019; HLPE 2020).

In recognizing the central role of human agency in AFSs, we also need to avoid the common temptation to focus excessively on either end of the value chain: upstream farmers and/or downstream food consumers. Most value addition, employment, etc., occurs between the farm and final consumer, and the relative importance of the post-farmgate stages of value chains inevitably expands with income growth and urbanization. Mid-stream value chain actors—many of them large, private corporations—too often lurk in the shadows of policy debates. These actors can, and must, be mobilized as equal partners in the co-creation of innovations to accelerate AFS transformation.

The intrinsic heterogeneity of AFSs is their second key feature. The coordination mechanisms and science necessary to internalize or mitigate externalities so as to avoid catastrophe and to foster continuous improvement vary enormously across geographies and agroecological and socioeconomic contexts. One-size-fits-all solutions do not exist. The panel therefore eschews ranking specific innovations, as performance will typically vary by context.

We adopt the approach of the Food Systems Dashboard, an excellent new tool that curates myriad data sources to enable visualization of key data series at country, regional, and global scales, and emphasizes five AFS types (Fanzo et al. 2020; Marshall et al. 2021)Footnote 10:

  1. 1.

    Rural and traditional: Farming is dominated by smallholders, and agricultural yields are typically low. Most farmers focus on staple crops (and retain much of their harvest for their own consumption) and a limited number of cash crops. Food imports and exports represent a small percentage of domestic consumption and production. Supply chains are short, resulting in many local, fragmented markets and limited non-farm AFS employment. Limited cold chains and storage facilities cause large food losses, which may also disincentivize diversification into perishable foods. The quantity and diversity of foods available varies seasonally, often with a pronounced lean season. Food is mainly sold through informal market outlets, including independently owned small shops, street vendors, and periodic markets. Supermarkets are uncommon, especially outside of major cities. Mandatory or voluntary fortification guidelines for staple foods are common in order to combat micronutrient deficiencies.

  2. 2.

    Informal and expanding: Average agricultural land and labor productivity and access to inputs (e.g., improved seeds and fertilizer) are higher than in traditional systems and rising. Modern food supply chains are in place for grains and other dry foods, which include processors and centralized distribution centers. These are also emerging for fresh foods, though traditional supply chains continue to dominate due to cold chains and other market infrastructure that remain underdeveloped. Processed and packaged foods are available in both urban and rural areas. Food processing may incorporate a combination of locally sourced and imported ingredients. Demand for convenience foods increases as the formal, non-farm labor force grows and includes more women, with urbanization and income growth also playing a role. Supermarkets and fast food are rapidly expanding and attracting more middle-class consumers, although informal market outlets still dominate food retailing, especially for animal-source foods, fruits, and vegetables. Few food quality standards are in place and advertising is not regulated, though many countries have fortification guidelines for staple foods.

  3. 3.

    Emerging and diversifying: Large-scale commercial farms increasingly co-exist alongside large numbers of small-scale farms, all of which enjoy enhanced market integration through better communications and transportation infrastructure. Food supply chains for fresh foods, including fruits, vegetables, and animal-source foods, are developing rapidly. Supply chains are elongating, with urban areas relying on food imports and rural areas relying more on export markets than in more traditional and informal food systems. Processed and packaged foods are widely available in rural areas, with less seasonal fluctuation in availability of perishable foods. Supermarkets are common even in smaller cities, although most fresh foods continue to be purchased through informal markets. Food safety and quality standards exist, but mainly within formal markets due to limited government monitoring capacity. A greater proportion of countries in this food system type have adopted food-based dietary guidelines.

  4. 4.

    Modernizing and formalizing: Larger farms rely more on mechanization and input-intensive practices, resulting in higher agricultural land and labor productivity. Food supply chain infrastructure is more developed, resulting in fewer food losses beyond the farmgate, although waste and spoilage at the retail and consumer end of the supply chain remains a challenge. Food and beverage manufacturing, food retailing, and food service capture a significantly greater share of consumer food expenditures. Dietary diversity rises, with regional specialization in agricultural production and imports of foods enabling more year-round availability of diverse foods. Multiple supermarket and food service chains exist within cities and larger-sized towns. These chains capture a large market share of fresh foods and are more accessible to lower-income consumers. Government regulation and monitoring of food quality standards are more common.

  5. 5.

    Industrialized and consolidated: Farming represents a land- and capital-intensive business, dominated by a small number of large-scale, input-intensive farms serving specialized domestic and international markets (e.g., horticulture, animal feed, processed food ingredients, biofuels). Market consolidation is common both upstream and downstream, as a shrinking number of large life-sciences firms supply patent-protected farm inputs while large processors, manufacturers, and retailers procure directly from farmers, reducing the number of intermediaries along the supply chain. Supermarket density is high in urban and metropolitan areas and even most medium-sized towns have access to multiple chains. The formal food sector represents nearly all domestic food consumption, including fresh foods. Luxury-oriented food retail and food service expand, creating greater quality differentiation in the food retail and food service sectors. Pockets of food insecurity still exist, often referred to as “food deserts,” alongside employment, income, and wealth disparities. A greater proportion of countries in this food system type have adopted policies that ban use of industrial trans fats and reformulate processed foods for reduction of salt intake.

At a coarse scale, simply using the typology method to assign entire countries to individual AFS types drives home several key points. First, a plurality of humanity currently lives in countries dominated by rural and traditional systems (Table 1). Population growth and migration patterns will only reinforce this need to invest far more effort and resources in AFS innovation for the Global South. Second, most of the Earth’s land mass is in the most advanced (industrial and consolidated or modernizing and formalizing) systems (Table 1). These places present especially large opportunities to transition working lands from growing food to sequestering carbon to reduce harmful greenhouse gases (GHGs) and to reap the resulting mitigation benefits: harvesting renewable energy and restoring habitats. Third, although discussions of AFSs commonly revolve around the extremes of this continuum—focusing on either the smallholder farmers that predominate in rural and traditional systems, or on the large-scale industrial farming and food corporations of the industrial and consolidated systems—most of the world’s population resides in countries dominated by transitional states. The opportunity to shape those transitions is especially profound.

Table 1 Human population and land area by agri-food system type

Of course, many AFS types can co-exist within a country or even a metropolitan region. Typologies allow for cross comparisons of trends and emerging patterns at whatever level of aggregation or disaggregation the data permit. As we highlight below, the impact pathways one envisions for different innovations fundamentally turn on the characteristics of the local AFS one targets. We depict key patterns in AFSs today with reference to these five typologies.

The third key feature that stands out as especially relevant when trying to promote beneficial innovations is that the closely coupled nature of AFSs implies that actions anywhere have spillover effects or externalities elsewhere in the system. Examples of negative externalities abound in AFSs. Some food processing practices that reduce costs, thereby making food more affordable (e.g., by adding inexpensive fats, salt, and sugars) have adverse public health consequences. Fertilizer misuse or overuse on farms can lead to nutrient runoff into waterways that causes downstream eutrophication or harmful algae blooms that harm fisheries. Many food system processes contribute massive amounts of GHGs that adversely affect the global climate, including land clearing; tilling of soil; agrochemical applications; the digestive processes of vast numbers of ruminant livestock; and the burning of fossil fuels, either directly by farm machinery and transport equipment or indirectly by utilities that provide electricity to milking parlors, manufacturing facilities, retail outlets, etc. Equally important, however, are positive externalities that arise from other behaviors—from animal and plant disease controls that limit the spread of harmful organisms, to scientific discoveries that cascade into further innovations.

AFSs’ pervasive externalities imply a divergence between the market price of foods and their social cost, once one factors in environmental, public health, and other externalities. This divergence reflects a market failure; markets typically cannot internalize spillovers easily. A range of groups are working on true cost accounting for food, often relying on life-cycle costing and similar methods to try to capture the full impacts of each product, inclusive of indirect impacts on the natural environment, public health, etc.Footnote 11 Governments must play a role in addressing the gap between market prices and true costs through regulatory, subsidy, and tax policies. But private companies and investors can do so, as well, including through innovative financing mechanisms of the sort we discuss later.

But no matter the policy instrument or pricing method governments use, they quickly confront the “food price dilemma” (Timmer et al. 1983), wherein price changes invariably cause both winners and losers. For example, higher food prices to reduce the environmental impacts of agri-food production generate environmental gains but also equity losses as foods become less affordable to the poor. Hence the central importance of technological advances—and especially socio-technical bundles—because these offer the chance to obviate the food-price dilemma and generate gains in one or more dimensions without having to impose losses on others. Advances will not always be “win–win”; a “win-neutral” is still an unambiguous improvement. The central task of innovation systems and the design of transition pathways is to identify bundles of technological and policy/social innovations that together enable what economists term “Pareto improvements” (i.e., advances for at least some without making anyone worse off).

The pervasive externalities that arise from AFSs’ deep connectivity through various abiotic, ecological, and human processes often induce a tempting conceit that one can optimize AFSs. But billions of individual food consumers, farmers, firm managers, workers, etc., make decisions and act every day, pursuing their own motives within the constraints specific to their time and station. No one has authority or control over even significant sub-systems, much less the whole. Rather, AFSs are highly decentralized networks of agents making interdependent decisions semi-autonomously. Moreover, we often overstate how well we can quantify and compare trade-offs of often fundamentally incomparable multiple objectives.

The fourth key feature of modern AFSs is the central place of science—for discovery, invention, adaptation, and engineering—which is necessary to maintain and advance innovation and systems performance in virtually any dimension. The panel is alarmed by how widely—and perhaps increasingly—sound scientific advice and evidence is being ignored by business, community, media, and political leaders, as well as by everyday decision-takers. Scientific research remains essential to unlock better ways of more efficiently using the Earth’s finite resources, of combatting changing threats, and of seizing emergent opportunities. The evolutionary nature of the AFS structure implies a never-ending need for scientific research to continuously adapt to evolving systems. Hence the importance of ongoing, generous public and philanthropic funding of basic science, a pure public good on which private investors can build. Indeed, scientific discovery generates some of the greatest positive spillovers as new findings diffuse and adapt broadly throughout AFS, lowering food prices to provide consumers with more affordable and safer foods, and farmers and firms with more productive digital, genetic and mechanical inputs and management processes. The world has previously faced daunting AFS challenges and, through science, emerged stronger; we can do it again (Barrett 2021).

Together, these four essential features of AFSs—summarized earlier in the simple mnemonic HHSS—must remain front-of-mind in promoting innovations within AVCs: decentralized human (H) agency, the intrinsic heterogeneity (H) of AFSs locally and globally, pervasive spillover (S) effects, and the essential role of scientific (S) research. They are foundational to the panel’s assessment of the rich pipeline of emergent AFS innovations and our recommended action plans to facilitate necessary transitions in the decades ahead.