The Missing Model

TAM-078 · The Approximate Mind

Ananya Desai works at a policy research institute in Delhi. She has been asked to model the consequences of restructuring India’s public distribution system, the network of ration shops that provides subsidized grain to roughly 800 million people. The restructuring is sensible on paper. Direct benefit transfers to bank accounts rather than in-kind distribution through physical shops. More efficient. Less leakage. Better targeting. The fiscal model is clean.

She has the fiscal model. She has the nutritional model. She has the political model, the one that estimates electoral consequences across states. She has a logistics model for the transition period. Each is well-built. Each is defensible.

She does not have a model for what happens to the neighborhoods where the ration shop was the last remaining institutional gathering point. Where the monthly trip was not just a transaction but an encounter: the woman who told you about the new clinic hours, the uncle who mentioned that the school was hiring, the queue where you stood next to someone who knew someone who might know about a job. She does not have a model for what happens to compound stress in households that relied on the ration shop not just for calories but for the encounter, the information exchange, the periodic proof that someone in the institutional structure knew their name.

She keeps a plant on her desk. A money plant in a cracked ceramic pot, the kind you find at any roadside nursery for thirty rupees. She has been trying to keep it alive for two years. It is not thriving. It is not dying. It persists in a state of marginal viability that she finds, for reasons she has not examined, impossible to give up on.

The Five Boundaries

#

Every social contract produces consequences across dimensions that the research tradition treats as separate. Economic effects. Social effects. Psychological effects. Physiological effects. Intergenerational effects. Each dimension has its own field, its own methods, its own journals, its own funding streams, its own career structures.

The consequences do not respect these boundaries. They never have. But the instruments we use to study them do.

The economic-social boundary. When the ration shop closes, the fiscal model registers a gain. The social model, if anyone built one, would register a loss. The gain and the loss are not independent. The economic restructuring that produces the fiscal gain is the same event as the social restructuring that produces the social loss. They are one event observed from two disciplinary positions. No model combines them, because the economists who build fiscal models and the sociologists who study social cohesion publish in different journals, attend different conferences, and are evaluated by different tenure committees.

The social-psychological boundary. When the gathering point dissolves, the social structure changes. When the social structure changes, the individuals embedded in it experience the change as something that happens inside them: loneliness, disorientation, the ambient sense that the world has become less legible. This series has given it a name, cognitive indifference, the flatness of a world where nothing requires full attention. The social change and the psychological change are not parallel processes. The social change produces the psychological change through mechanisms that neither sociology nor psychology, practiced separately, can trace.

The psychological-physiological boundary. When the psyche absorbs the consequences of institutional restructuring, the body follows. Not metaphorically. Literally. Chronic stress produces elevated cortisol, inflammatory markers, allostatic load. Over years, the load produces organ system degradation. The process Arline Geronimus named weathering. Psychological models do not model organ systems. Medical models do not model structural stress. The gap between them is where compound harm accumulates.

The physiological-intergenerational boundary. When the current generation’s bodies carry the consequences of the current social contract, the next generation inherits a starting position the contract did not intend. Epigenetic transmission of stress markers. Developmental consequences of parental compound burden. Formation effects: the child raised in a household saturated with stress develops differently from the child raised in a household with margin. No model crosses this boundary because the data required spans decades and the funding spans budget cycles.

The intergenerational-economic boundary. When the generation formed inside the current arrangement enters the economy, it enters with the cognitive, psychological, and physiological profile the arrangement produced. The blocked generation from Part 64. The apprenticeship crisis from The Transformed. The developmental consequences of the current arrangement become the economic inputs of the next one. The loop closes.

Each boundary is policed by an institutional architecture that rewards depth within the boundary and penalizes work that crosses it. The integration is not technically impossible. It is institutionally prevented. Not by decree. By incentive.

The Interdisciplinary Objection

#

The obvious response: this is what interdisciplinary studies is for. The field exists. It has departments, journals, conferences, funding programs. It has existed for decades.

And it has not produced the integration.

Here is why, and it is not because the people in the field have failed. It is because the institution that houses the field cannot sustain what the field requires.

Interdisciplinary studies became its own discipline. It developed its own methodological commitments, its own career ladder, its own tenure criteria, its own journals. The people in interdisciplinary departments are evaluated by interdisciplinary standards, publish in interdisciplinary journals, hire against interdisciplinary criteria. They crossed the old boundaries and built new ones around the crossing itself.

The word “interdisciplinary” contains the problem. “Inter” means between. It assumes the disciplines exist as stable entities and the work happens in the gaps. But the integration the social contract model requires is not between disciplines. It is beneath them. The mechanisms operate at a stratum the disciplines were not built to reach, not in the gaps between the strata they occupy.

What interdisciplinary studies produces in practice is multidisciplinary work. A team of specialists, each contributing their disciplinary perspective, writing chapters of the same report. The economist produces economic analysis with sociological awareness. The sociologist produces social analysis with economic context. Each chapter is competent. Nobody writes the connective tissue. Nobody models how the economic output becomes the social input becomes the psychological input becomes the physiological input. The transition paragraph between chapters says “these dimensions interact in complex ways” and the policy-maker reads the chapter whose discipline matches their institutional position and skips the rest.

The research university spent 150 years selecting against the cognitive profile the integration requires. The generalist who moves across domains, who follows the problem rather than the field, who sees the mechanisms at the level where they interact rather than at the level where they are separately observed, has been systematically disadvantaged by every institutional force that acts on a researcher: departmental hiring, journal publication, grant evaluation, tenure review. Interdisciplinary studies was the university’s attempt to correct for this without changing the selection mechanism. The correction got absorbed by the mechanism.

The farmer in Vidarbha whose polyculture manages risk, soil health, dietary diversity, and seed preservation simultaneously is not doing interdisciplinary agriculture. She is doing agriculture. The disciplines that decomposed her practice into separate analytical domains are the ones that created the boundaries. She never had them.

What the Model Would Need to Be

#

Not an optimizer. Not a predictor. A consequence modeler with retroductive capacity.

Given a proposed social contract, or a proposed restructuring of an existing one, the model would trace consequences across all five boundaries, not sequentially but interactively. Not “first the economic effects, then the social effects” but “the economic and social effects co-produce each other, and their interaction produces psychological effects that are not the sum of the economic and social effects considered separately, and the psychological effects produce physiological effects through mechanisms the disciplinary models do not contain.”

The model produces distributions, not predictions. “If you restructure the public distribution system in this way, in this geography, for this population with this compound condition profile, here is the distribution of consequences across economic, social, psychological, and physiological dimensions over five years, ten years, a generation.” The distributions are anchored in historical analogy: documented cases where similar restructurings occurred and the consequences were tracked, however incompletely, across dimensions.

The model flags the gaps. “The historical record contains no case where a food distribution system was restructured in a population with this specific compound condition profile. The consequence projection for this population is extrapolated, not observed. The confidence interval for this segment is wide enough to contain both significant benefit and significant harm. The model cannot distinguish between these outcomes because the mechanism has never been documented.”

The gap is the finding. The model’s most valuable output is not the consequence projection. It is the map of where the projection is insufficient, where the empirical record does not contain the mechanisms that will determine whether the restructuring helps or harms. The gap tells the policy-maker: for this population, you are making a bet without adequate information. The model cannot give you the information. It can tell you exactly where the information is missing and what the cost of proceeding without it might be.

What Does Not Exist Yet

#

The integration of economic, social, psychological, and physiological consequence modeling in a single framework. The pieces exist in isolation. Economic models simulate markets. Social network models simulate connection. Psychological models simulate individual response. Physiological models simulate biological systems. Each is sophisticated within its domain.

The integration is a research program, not a model training exercise. Someone needs to build the connective layer: the formal specification of how an economic output becomes a social input, how a social output becomes a psychological input, how a psychological output becomes a physiological input. Each connection is a modeling problem in its own right, and each modeling problem crosses a disciplinary boundary that the current research infrastructure polices.

Estimated timeline: three to five years to build the integrative framework. Then model training on top of it. Then calibration against historical cases. Then pilot deployment alongside actual policy deliberation. A decade, minimum, from concept to a functioning instrument.

The question is not whether the model can be built. The question is whether anyone commissions it before the consequences of uninformed social contract design make the commissioning unavoidable.

What Ananya Actually Does

#

Ananya knows all of this. Not in these terms. In the terms of a person who has spent eight years producing policy models that decision-makers use to make decisions, watching the decisions produce consequences her models did not predict, and knowing exactly why the models did not predict them: because the consequences unfolded in dimensions the models were not built to include.

She has started keeping a separate file. Not the official models. A personal document where she writes down the consequences she can see coming that her models cannot show. The social cohesion effects. The compound stress effects. The intergenerational formation effects. She writes them in prose because she cannot write them in equations, not because they are not real but because the formalism that would represent them does not exist.

The limitations section of her official report mentions “non-economic dimensions of welfare” in a paragraph on page forty-three. She knows that nobody who makes the decision will read page forty-three. She knows that the decision will be made on the fiscal model. She knows that the consequences will unfold across all the dimensions she could not model. She knows that by the time the consequences are visible, the decision will be difficult to reverse because the institutional infrastructure, the ration shops, the supply chains, the personnel, will have been dismantled.

I wonder whether there is a version of her separate file that could be formalized, not into the integrated model this essay describes, which is years away, but into something intermediate: a structured consequence map that names the dimensions the fiscal model cannot see, identifies the populations for whom the gap between the model’s predictions and reality is likely to be largest, and presents the map alongside the fiscal model as a companion document the decision-maker must acknowledge before proceeding.

This is not the model. It is the shadow of the model. The outline of what the model would show if it existed. It is insufficient. But it is less insufficient than the fiscal model alone, and it might be buildable now, by someone like Ananya, with the knowledge she already carries and the tools she already has.

The Plant

#

Ananya’s money plant has been in marginal viability for two years. She waters it inconsistently. She forgets for days and then overcompensates. She has moved it twice, once to get more light, once to get less. It has lost leaves and grown new ones. It has never flourished and has never died.

She thinks about it sometimes when she is working on the models. The plant persists in conditions that are not adequate for thriving. It does not collapse. It subsides into a stable state of managed insufficiency. It survives because money plants are remarkably tolerant, because they can lose what they need and still metabolize what they have, because the gap between what they require to thrive and what they require to persist is wide.

The populations her models cannot see are persisting in the same way. Not thriving. Not collapsing. Absorbing the consequences of policy decisions made on fiscal models, metabolizing what they have, losing what they need in increments too small for any single metric to register.

The plant is still on her desk. The file labeled “Integration” is still on her computer. She opens it sometimes and looks at the blankness. She does not know what the first line should be. She knows that someone needs to write it.

The ration shop on the corner near her apartment closed last month. She noticed because the queue used to be visible from her window on distribution days. The queue is gone. The people who stood in it are elsewhere, purchasing their entitlement through a digital transfer that is more efficient and less leaky and does not require them to stand next to each other once a month.

She does not know what the queue was carrying. She suspects nobody measured it. She suspects that by the time someone does measure it, the thing the queue was carrying will have been absent long enough that its absence will have become normal, and normal absences are the hardest to see.

The plant persists. The file is empty. The queue is gone.

This is Part 78 of The Approximate Mind, a series exploring how artificial intelligence reshapes what it means to think, create, work, and live. The epistemic cluster that began with Part 74 (The Interrogator) and continued through The Epistemic Framework (75), The Amplitude Problem (76), and The Injected Center (77) asked what AI systems cannot see and what systems designed to see it would need to be. This essay asks why we cannot simulate the consequences of a social contract across economic, social, psychological, physiological, and intergenerational dimensions. The answer is not technical. It is institutional: the research architecture that produces knowledge within each dimension prevents the integration across dimensions where the actual consequences unfold. The companion essay, Part 79, asks what research itself would look like if it started from the premise that decomposing what should not be decomposed is the methodological error, not the methodological standard.

References

#

Critical Realism and Ontological Depth

Bhaskar, Roy. A Realist Theory of Science. Verso, 1975.

Bhaskar, Roy. The Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences. Harvester Press, 1979.

Systems Dynamics and Modeling

Meadows, Donella H., et al. Limits to Growth. Universe Books, 1972.

Forrester, Jay W. World Dynamics. Wright-Allen Press, 1971.

Sterman, John D. Business Dynamics: Systems Thinking and Modeling for a Complex World. McGraw-Hill, 2000.

The Institutional Architecture of Knowledge

Foucault, Michel. The Order of Things: An Archaeology of the Human Sciences. Vintage Books, 1970.

Abbott, Andrew. Chaos of Disciplines. University of Chicago Press, 2001.

Flyvbjerg, Bent. Making Social Science Matter: Why Social Inquiry Fails and How It Can Succeed Again. Cambridge University Press, 2001.

Social Contracts and Their Consequences

Rawls, John. A Theory of Justice. Harvard University Press, 1971.

Sen, Amartya. Development as Freedom. Anchor Books, 1999.

Nussbaum, Martha. Creating Capabilities: The Human Development Approach. Harvard University Press, 2011.

Weathering and Compound Stress

Geronimus, Arline T. Weathering: The Extraordinary Stress of Ordinary Life in an Unjust Society. Little, Brown Spark, 2023.

Marmot, Michael. The Health Gap: The Challenge of an Unequal World. Bloomsbury, 2015.

Case, Anne, and Angus Deaton. Deaths of Despair and the Future of Capitalism. Princeton University Press, 2020.

India’s Public Distribution System

Drèze, Jean, and Amartya Sen. An Uncertain Glory: India and Its Contradictions. Princeton University Press, 2013.

Khera, Reetika, ed. The Battle for Employment Guarantee. Oxford University Press, 2011.

The Generalist Mind

Epstein, David. Range: Why Generalists Triumph in a Specialized World. Riverhead Books, 2019.