New tool evaluates sustainability of complex systems for strategy and communication

Factory Safety Math Meets the Farm: New Tool Scores Sustainability by Place, Not Just Practice

Sustainability has a measurement problem, and a new tool developed with University of Turin funding thinks the fix lies in mathematics that factories have trusted for decades. In a study published in the Elsevier journal Environmental and Sustainability Indicators, researcher D. Biagini presents I-sustain, a farm-level method that transplants the logic of industrial risk assessment—the engineer’s pairing of how often something happens with how bad it is when it does—onto the evaluation of agricultural sustainability. The twist is that the method refuses to judge a farm in isolation. Every function a farm performs is multiplied by a measure of the vulnerability, exposure and needs of the territory it occupies, so identical practices can legitimately earn different sustainability scores in different places. Laid out in the first of two companion papers, the framework defines the architecture, scoring engine and data plumbing of a tool meant for routine, semi-automated use by farmers, advisers, policy makers and consumers. Its author argues that after nearly four decades of sustainability science, the field has too often measured the wrong thing, in the wrong place, for the wrong audience.

The scale of the problem is sobering. Since the 1987 Brundtland Report defined sustainable development as progress that meets present needs without compromising the ability of future generations to meet their own, the idea has been written into the United Nations’ 2030 Agenda and policy frameworks worldwide. Yet within ten years of that founding definition, more than 300 competing definitions of sustainability were circulating, and the assessment field that grew alongside them remains fragmented. Carbon and water footprints are precise but narrow. Life cycle assessment, the workhorse of environmental accounting, is built around products and functional units and struggles to capture the multifunctional character of a farm—its landscape work, its ecosystem services, its ties to a rural community. Multi-criteria decision analysis folds in stakeholder preferences but inherits their subjectivity through weighting and aggregation, while life cycle sustainability assessment stays data-hungry and hard to run routinely. The result is a landscape of methods designed mainly for researchers, often locked behind specialist expertise, proprietary software or subscription databases, and delivering answers too slowly to steer a working farm.

The method’s central premise is that farms are complex systems in the strict sense: assemblies of many deeply interacting components, in which biological cycles intertwine with technological ones and consequences ripple outward into a territory. A feeding strategy that curbs nitrogen excretion matters far more inside a nitrate-vulnerable zone than outside one. A livestock operation that keeps a depopulating valley inhabited performs a social function no yield statistic can register. Conventional tools, the paper argues, either ignore these context-dependent interactions or confine them to the environmental pillar, while product-based frameworks miss the interplay among on-farm activities—the manure that feeds the fields, the fields that feed the herd, the on-farm dairy that turns milk into a product tourists will drive hours to buy. I-sustain instead evaluates the farm and its place as one integrated whole, asking what its functions do to the environmental, territorial and socio-economic fabric around them, and how strongly that fabric amplifies or mutes each function. Sustainability, in this view, is not a property of a farm alone.

Beneath the concept sits a deliberately compact equation borrowed from safety engineering, where risk equals frequency times magnitude under frameworks such as EN ISO 12100:2010 and CEI EN IEC 31010:2019. I-sustain rewrites this as H = e × a, where e is the frequency with which a farm performs the functions behind a given sustainability cluster, and a is the multiplying effect that the natural, social and economic context imposes on those functions. Both factors are scored on four-level ordinal scales: frequency runs from “lacking” through “rare” and “probable” to “highly probable”; magnitude from “light” through “medium” and “heavy” to “very heavy”. Crossing them yields sixteen combinations, collapsed into four bands of importance—negligible, medium, relevant and very relevant—on an exponential rather than linear scale, so chronic, small pressures such as nitrogen leaching cannot quietly accumulate into invisibility. Farm indicators, quantitative and binary, are normalised to a zero-to-one scale, averaged and rescaled to the four frequency levels with thresholds at 0.25, 0.50 and 0.75; territorial indicators from public databases are treated identically to fix the magnitude score. The matrix, the paper stresses, is a semi-quantitative scoring device, not a precision instrument.

Where most frameworks sort everything into the classic environmental, social and economic pillars, I-sustain redraws the map into three domains of its own: environmental, territorial and social-economic. The environmental domain covers the familiar pressures—greenhouse gas emissions, nutrient losses, water use, soil degradation, pollution and biodiversity effects. The territorial domain is the distinctive move, separating functions that are less about consuming resources than about holding a landscape together: maintaining rural roads and hydro-geological defences, preventing erosion and inundation, preserving landscape value, biodiversity and the biotic balance of the agro-ecosystem. These are public goods that markets rarely remunerate, and the paper anchors them in the ecosystem-services literature, citing the UN-linked Ecosystem Services Valuation Database. The social-economic domain pointedly excludes a farm’s private profitability—income per worker and gross saleable production are set aside as too market-dependent—focusing instead on what farming does to a locality: short supply chains, rural tourism, cultural traditions, local employment, distributed income and migration flows. Each domain is scored independently, so no dimension can be traded off invisibly, and roughly eight to ten clusters per domain keep the contributions balanced.

The method also flips the sign of the entire exercise. Because what most indices actually capture is the distance between a system and an ideal state, they are, the author notes, measuring unsustainability—a score that grows as things get worse, telling producers what to avoid. I-sustain inverts the reference point. The baseline becomes a hypothetical farm doing nothing beyond ordinary practice, judged by the “principle of reasonable care”, and every function performed in favour of sustainability earns positive credit. Producers are told, constructively, what they could do next rather than what they are doing wrong. Cluster scores are summed into a Sustainability Index and translated into four letter classes, A through D, each with a plus, zero or minus modifier: twelve levels designed to function as a kind of environmental label for supply-chain operators and shoppers. Optional policy weights can tilt the index toward politically prioritised functions, but the baseline version assigns none, and the paper caps recommended weightings at thirty percent after worked sensitivity calculations, always reporting weighted and unweighted results together.

In a worked example, the paper walks through a hypothetical family farm: ten hectares of hilly ground worked by four family members and a part-time employee, carrying forty livestock units of a little-diffused native cattle breed, with pasture reached along five kilometres of unpaved state roads the family maintains. Milk is processed into cheese on-site in a plant powered by self-produced hydroelectric and solar energy and sold to visiting tourists. For the nitrates cluster, farm indicators—a nitrogen efficiency of twenty percent against a dairy reference near twenty-five percent, no multiphase feeding, no amino-acid supplementation, a stocking rate of four livestock units per hectare, but solid rather than slurry manure—average 0.2667, which rescales to a frequency score of two, “rare”. The territory supplies the other half of the story: regional livestock density, nitrate levels in natural water reserves and the groundwater-protective character of local soils average 0.2444, giving a magnitude score of one, “light”. Multiplying them yields H = 2, classed “medium”. The same effort inside a nitrate-vulnerable zone with leaching-prone sandy soils would climb the matrix steeply.

What separates I-sustain from a well-argued spreadsheet is the data plumbing it is designed for. Farms increasingly keep dossiers, field records, livestock registers and nutrient plans in digital form, so the method is built around automatic or semi-automatic annual updates: a future software layer would query public and private repositories—environmental agencies, national statistical institutes, livestock registries, traceability platforms—and merge them with a standardised record card covering the farm’s agronomic, livestock, management, ecosystemic, certification, cultural and socio-economic characteristics. The ambition is a dynamic, continuously refreshed appraisal rather than a one-off certificate. The framing is deliberately gate-to-gate and system-based rather than product-based, tracing material and energy flows across on-farm processes to expose opportunities for circularity: closing nutrient loops, integrating enterprises, monetising ecosystem services. The author positions the tool within the Open Innovation paradigm, a bridge among researchers, producers, institutions and society, aimed at the circular-economy transition that European policy keeps demanding of the food sector.

The method’s political machinery is explicit. Because clusters and weights are fixed transparently by panels of scientists, technicians, policy makers and supply-chain stakeholders—ideally through structured Delphi elicitation—governments can steer farm behaviour by re-weighting what counts. The paper calculates that weighting one cluster in ten by plus ten percent lifts a farm’s final index by between 0.33 and 3.6 percent depending on the other scores; plus twenty percent yields 0.66 to 7.11 percent; plus thirty percent, the recommended ceiling, up to 10.67 percent. Conditional funding under instruments such as the EU’s Common Agricultural Policy could thus reward hydro-geological defence in landslide-prone regions or short supply chains in depopulating areas, while the unweighted score preserves an objective benchmark underneath. The author adds that the tool can compare intensive and extensive systems fairly, scoring every function each performs, and thereby operationalise what the literature calls sustainable intensification: output judged not on technical efficiency alone but on environmental and territorial performance measured in context.

The author is candid that this first paper is theoretical; no statistical validation is offered, and the empirical debut comes in the companion study, which applies the method to real dairy and beef farms. Arithmetic aggregation, chosen for transparency, risks letting strong clusters mask weak ones, so results are always reported at cluster and domain level to keep weaknesses visible. Four-level discretisation discards information, identical scores can conceal different realities, and the value assigned to territorial functions is ultimately a normative, policy-laden judgement that varies across regions and eras. Transfer beyond agriculture—to other bio-based systems, territorial supply chains or rural development projects—is possible only where functions are identifiable, effects are context-dependent and reliable data exist on both sides. I-sustain, the study concludes, is not a definitive measure of sustainability. It is a transparent, adaptable lens through which farm multifunctionality, territorial context and political priorities can be judged together—and, perhaps soon, printed on a label.