How do humans actually respond to structural pressures — the behavioral assumptions hidden in every macro-economic and governance model, and where they break down
Why Don't People Just Do the Sensible Thing? What We Actually Know About Human Behavior and Why It Makes Governing Very Hard
Based on analysis of a 224-node, 888-edge knowledge graph exploring behavioral assumptions in macroeconomic and governance models
The Basic Problem
Most laws, economic policies, and institutions are built on a quiet assumption: that people are basically rational. They weigh their options, think about the future, act in their own interest, and respond predictably to rules and incentives.
The problem is that decades of research in psychology, economics, and neuroscience have shown this assumption is wrong — not a little wrong, but systematically wrong in specific, repeatable ways. People discount the future heavily, make decisions based on how options are framed rather than what they actually contain, follow social norms even when it costs them personally, and often change their beliefs to match their behavior rather than the other way around.
This knowledge graph maps out all of those behavioral failures and then traces what happens when you build institutions on top of them anyway. What emerges is something like a diagram of a machine that was designed with the wrong parts.
The Central Ledger
The most connected node in the entire graph — the concept with the most roads running into and out of it — is called the “Five Falsified Behavioral Axioms of Governance.” Think of it as a checklist of the five things governance systems assume about people that research has repeatedly shown to be false.
These five assumptions are something like:
- People are consistent and stable in their preferences
- People respond to incentives in predictable, proportional ways
- People think about the future at the same rate they think about the present
- People can be treated as interchangeable units in a model
- People’s choices aggregate into collective outcomes in straightforward ways
The graph shows that evidence for each of these failures flows in from dozens of specific studies, experiments, and real-world events. And from those five failures, explanations for a wide range of institutional breakdowns flow out. It functions as an accounting ledger: here is everything that falsifies the standard model, and here is everything that follows from having built the model wrong.
The Three Floors of the Building
The graph has a clear three-floor structure. On the ground floor are the basic mechanisms inside a single human brain: the fact that the part of your brain that imagines your future self is the same part that imagines a stranger — not the part that imagines yourself right now. Or the fact that losses feel roughly twice as bad as equivalent gains feel good.
These ground-floor findings feed into the second floor: predictable patterns of failure. If your brain treats your future self like a stranger, you will discount future costs and benefits heavily — a finding called hyperbolic discounting. If losses feel worse than gains, you will take more risks to avoid losing something than to gain something of equal value.
The second floor, in turn, feeds into the third floor: institutional and systemic breakdowns. When the entire electorate discounts the future the same way individuals do, you get democratic short-termism — politicians who win by promising things now and hiding costs in the future. When entire populations have distorted risk assessments, you get financial crises and climate inaction that persist even when the rational case for action is clear.
The graph did not impose this structure from the outside — it emerged from the edges themselves.
Two Unusual Nodes
Two nodes in the graph are structurally unusual in opposite ways.
The “Convergent Climate Governance Failure Architecture” node has the second-highest number of connections in the entire graph. But it has the lowest possible weight — a score of 1, indicating it may be a stub or placeholder concept. This combination means something specific: it is not a mechanism but a destination. Almost every road in the graph eventually ends there. It is where things arrive, not where they start. In other words, climate governance failure is the graph’s main outcome variable, not an explanatory concept.
The “Civilizational Behavioral Governance Trap” node is the opposite: high weight, high connections, and many edges running both in and out. It is a synthesizer — a concept that has been built up carefully and that the graph treats as central rather than merely final. The fact that it sits in a reinforcing loop (described below) explains the high weight: it is a node that keeps accumulating evidence the longer you look at the graph.
The Loops That Drive Themselves
The most important structural finding in this kind of map is when a chain of cause-and-effect bends back on itself. These are called feedback loops, and they are concerning because they do not require any outside push to keep going — they reinforce themselves.
The graph contains several. The most direct one involves the relationship between algorithms and trust. Algorithmic recommendation systems — the software that decides what you see on social media or in a news feed — tend to amplify emotionally engaging content, which often means conflict, outrage, and distrust. As institutional trust declines, people increasingly get their information from algorithmic feeds rather than institutions. Which means the algorithm fills more of the information space. Which accelerates trust decline.
The graph encodes both edges of this loop explicitly, and neither direction is a stretch: the first is well-documented in the research on engagement-maximizing recommendation systems, and the second in the research on information voids and distrust.
A second loop connects the falsified behavioral axioms to institutional trust collapse and back to the axioms. When governance systems built on wrong assumptions fail visibly and repeatedly, people lose trust in institutions. Institutions whose legitimacy has eroded are less able to correct the behavioral models they use. The conditions that produce the axiom failures become more entrenched. The loop is not fast, but the graph’s structure suggests it is durable.
A Few Non-Obvious Connections
Some of the edges in this graph are interesting precisely because they connect things that normally live in separate fields.
The graph connects a famous psychology experiment — the Milgram obedience studies, in which ordinary people administered what they believed were dangerous electric shocks because an authority figure told them to — to the phenomenon of regulatory capture, where the agencies meant to regulate industries end up serving those industries instead. The edge asserts that the same psychological mechanism by which individuals suspend moral judgment under authority explains why regulatory personnel can be captured even when they would, in isolation, oppose what they are doing. This is a different explanation than the standard one (which is about financial incentives and revolving doors), and it predicts that capture will be more durable and harder to fix than incentive-based models suggest.
A second unexpected connection runs from street-level bureaucrats to algorithmic systems. Front-line government workers — the caseworkers, permit officers, and enforcement agents who implement policy directly — routinely develop informal workarounds to manage impossible workloads and contradictory rules. When AI systems replace these workers, the graph encodes a specific claim: those workarounds get translated into the algorithm as permanent features. The discretionary judgment call that a human made case-by-case becomes a locked-in rule in software. The implication is that AI-administered public services may inherit and fossilize the exact informal distortions that reformers were hoping to eliminate.
The Only Exit, and Its Problems
The graph contains exactly one node with edges labeled “sole escape from” and “only potential temporary override of.” It is called Collective Effervescence Crisis Override, and it refers to the documented phenomenon of collective psychological transformation that occurs during genuine crises — moments when normal social divisions temporarily dissolve and large-scale coordinated behavior becomes possible in ways it normally is not.
The structural problem is that the graph encodes no path from deliberate design to this state. It appears to require a crisis large enough to cross a threshold, not a policy choice or intervention. And even when it occurs, the edges label it as temporary.
A second partial exit is Ostrom’s theory of commons governance — the finding by economist Elinor Ostrom that communities can successfully manage shared resources without privatization or top-down regulation, under specific conditions. The graph acknowledges this as the strongest empirical counterexample to the otherwise bleak picture. But it also encodes why this exit is largely unavailable: Ostrom’s conditions require stable community boundaries, mutual trust, and working local institutions. The graph explicitly marks these conditions as absent from the two domains where failures are most severe: global climate governance and AI governance.
A Structural Tension the Graph Does Not Resolve
The graph contains two mechanisms that make opposite predictions and does not specify when each dominates.
Cognitive dissonance research finds that when people take an action — even reluctantly — they tend to update their beliefs to match the action. Behavior comes first, belief follows. This is useful for policymakers: get people to do the thing, and attitude change follows.
Moral licensing research finds the opposite: doing a virtuous thing grants psychological permission to do something less virtuous later. Buying an organic product makes you feel like you can skip the exercise. Voting for the environmentally conscious candidate relieves pressure to change your own habits.
Both effects are real and well-documented. The graph encodes both and connects them to each other with an edge acknowledging the tension. But it does not specify which dominates under what conditions. This is an honest representation of the state of the research — but it means the graph’s solution architecture cannot simply recommend “change behavior first” without specifying when that strategy works and when it backfires.
The Ceiling Problem
One of the quieter findings in the graph is a node called the “Behavioral Policy Structural Ceiling.” It makes a specific claim: behavioral interventions — nudges, default options, choice architecture, information campaigns — cannot fix problems whose causes are structural rather than cognitive.
This matters because most of the solutions the graph encodes operate below that ceiling. The graph’s solution layer is mostly behavioral. Yet the graph also encodes that the failures driving the main outcomes are amplified by inequality, institutional decay, algorithmic infrastructure, and the falsification of behavioral norms at scale. These are not fixed by changing the default option on a form.
The graph does not resolve this tension. It encodes both the solutions that are available and the constraints that make them insufficient.
Bottom Line
The graph shows a system where the behavioral assumptions built into governance models are wrong in predictable ways, and where the institutional failures those wrong assumptions produce tend to reinforce themselves through several identifiable loops. The most connected concept in the graph is a synthesis of five specific ways those assumptions fail. The most structurally central node is the trap that results when all of those failures compound over time.
The graph also identifies two structural features that are easy to miss. First, the algorithmic amplification of behavioral biases is not an external shock to the system — it is endogenous, simultaneously a product of the failures it amplifies. Second, the most powerful documented mechanism for escaping the trap appears to require a large exogenous crisis to activate, which is not a policy tool.
The most honest summary of what the graph shows is this: we know a great deal about how people actually behave, we know how governance systems fail when built on wrong assumptions about behavior, and we can trace the feedback loops that make those failures self-sustaining. What the graph does not contain — and what the structural analysis makes explicit — is a clear path from knowing all of this to doing something about it.