Why Perfect Theories Make Terrible Policies
Gaus, Gerald F. The Tyranny of the Ideal: Justice in a Diverse Society. Princeton University Press, 2016. https://press.princeton.edu/books/hardcover/9780691158808/the-tyranny-of-the-ideal.
There’s a persistent belief that policy failures happen because politicians ignore expert advice. Healthcare reform gets watered down by special interests. Climate action stalls because fossil fuel lobbies block rational carbon pricing. Immigration reform collapses because extremists prevent sensible comprehensive solutions.
The real story is more unsettling: these policies are failing precisely because politicians are following expert advice. The most sophisticated policy frameworks, implemented exactly as designed, produce the most predictable disasters. Our smartest people aren’t being ignored—they’re being followed faithfully toward systemic failure.
The Pattern of Sophisticated Failure
This isn’t partisan finger-pointing. The pattern appears everywhere once you start looking. Healthcare reform stalls on comprehensive frameworks while community health centres quietly expand access. Climate policy deadlocks on optimal carbon pricing while cities electrify bus fleets and buildings. Immigration reform collapses under grand bargains while states develop functional worker visa programs that nobody planned but somehow work.
Urban planning offers the clearest example. Comprehensive master plans gather dust in municipal offices while tactical urbanism transforms neighbourhoods. Federal education reforms cycle through systematic overhauls while individual schools quietly innovate their way to better outcomes. The sophisticated approaches consistently fail while messy, local adaptations succeed.
It’s not that politicians water down good ideas. It’s not that implementation is harder than theory. Something about sophistication itself creates these failures.
This pattern reveals something deeper than implementation failure. The sophistication isn’t just failing practically—it’s asking the wrong question entirely. Most policy frameworks start with “What’s the optimal arrangement?” But Gerald Gaus shows this conflates two completely different problems: what we ought to do versus what diverse people can actually coordinate around. A perfect theory of justice tells us nothing about whether people with different values can live together under it. The tyranny begins the moment we confuse evaluative perfection with coordination capacity.
The Great Expert Reversal
Watch what’s happened to expert opinion over the past two decades. The same urban planners who once championed comprehensive master plans now advocate tactical urbanism. The same development economists who designed structural adjustment programs now promote “learning by doing” and experimental approaches. Education reformers who pushed systematic overhaul now embrace incremental improvement. Healthcare policy experts who designed single-payer solutions now work within public-private partnerships.
These aren’t course corrections based on new evidence. They’re systematic retreats from sophistication toward pragmatism. The experts aren’t getting smarter about implementation—they’re discovering that sophisticated theory and workable practice are fundamentally incompatible.
But here’s the puzzle: if our most sophisticated approaches keep failing, why do we keep making them more sophisticated? Every policy disaster generates calls for better analysis, more comprehensive frameworks, more rigorous evidence-based approaches. We respond to the failure of sophisticated theory by demanding more sophisticated theory.
These reversals aren’t random course corrections. They reveal a deeper discovery about how coordination actually works. What the experts are discovering—often without realising it—is that designed solutions can’t match the complexity of evolved ones. Social contracts aren’t blueprints you implement; they’re patterns that emerge through countless interactions between people who disagree about almost everything except the need to get along somehow. The urban planners abandoning master plans aren’t admitting failure—they’re stumbling toward how cities actually develop. The policy experts embracing incrementalism aren’t lowering their sights—they’re discovering that coordination emerges through learning rather than design.
Gaus’s Diagnosis: The Sophistication Trap
This is where Gerald Gaus’s The Tyranny of the Ideal, published in 2016, becomes essential reading. Gaus shows why our most sophisticated attempts to create just societies systematically produce injustice—not despite their sophistication, but because of it.
Consider Australia’s National Disability Insurance Scheme. The NDIS was conceived as ideal theory made real: a comprehensive, person-centred, market-based system that would replace the “broken” patchwork of state-based disability services with rational, evidence-based, individually optimised support. The theory was elegant—give people with disabilities individual budgets, create a competitive market of providers, and let choice and competition drive better outcomes.
But the NDIS demonstrates Gaus’s core insight: ideal theories demand uniform implementation across irreducibly diverse circumstances. The NDIS demonstrates Gaus’s core insight even though (or perhaps because) it combines market liberal and social-democratic approaches—showing that even hybrid idealisms collapse under this logic. Every participant must navigate the same planning process, regardless of cognitive capacity. Every support need must be translated into identical billable categories, regardless of how people actually live. Every provider must comply with uniform regulatory frameworks, regardless of local circumstances or community relationships.
The NDIS exemplifies what James C. Scott calls the high modernist trap:1 replacing local, contextual knowledge with simplified abstractions that can be administered from above. The system demands that every aspect of disability support become legible to central planners—transformed into standardised categories, measurable outcomes, and billable units. But the actual work of supporting people with disabilities often depends on exactly the kind of informal, contextual knowledge that resists bureaucratic categorisation.
The system assumes what Gaus calls the evaluative-coordination conflation: it knows what optimal disability support looks like and demands everyone coordinate around that single vision. But people with disabilities have radically different needs, circumstances, family structures, cultural backgrounds, and preferences about how support should work. The sophistication treats this diversity as a problem to be solved rather than a permanent condition to be accommodated.
The knowledge problem makes it worse. Central planners can’t possibly know enough about millions of individual circumstances to design optimal support packages, but the NDIS requires exactly this impossible knowledge. The more comprehensive the planning process becomes, the more it assumes bureaucrats can anticipate how diverse families will respond to standardised interventions.
The results are exactly what Gaus predicts. The system has created warring tribes—families fighting planners, providers fighting regulators, states fighting the Commonwealth—each believing their comprehensive doctrine should govern everyone else. Indigenous communities get forced into individualistic models that ignore their actual coordination patterns. Rural participants navigate urban-designed systems that don’t fit their circumstances. Every failure generates demands for more sophisticated oversight, better evidence frameworks, more comprehensive planning—making the system more political and less functional.
Before the NDIS, Australia had evolved messy, diverse disability systems across states and sectors. Some were better than others, but they accommodated different approaches simultaneously. Country towns had community-based support networks. Cities had specialised services. Ethnic communities had culturally appropriate providers. None was perfect, but people could find arrangements that worked for their circumstances.
The NDIS replaced this evolutionary diversity with designed uniformity. Instead of building on what worked and fixing what didn’t, it imposed a single comprehensive theory across irreducible diversity. The sophistication isn’t failing because politicians compromised away the good parts—it’s failing because comprehensive, theoretically elegant solutions are structurally incompatible with the diversity they claim to serve.
What the NDIS eliminated wasn’t just institutional diversity—it was what Richard Sennett would call the ‘intelligent hand’ of disability support.2 The community worker who understood through years of experience how to navigate a particular family’s dynamics. The provider who had developed intuitive timing for when someone needed more or less independence. This embodied knowledge couldn’t be abstracted into planning frameworks without crucial information being lost.
But here’s where Gaus gets genuinely unsettling. Ideal theories don’t just fail—they actively create the polarisation they claim to solve. When coordination requires substantive agreement about justice, education, or the good life, politics becomes sectarian by design. Each group believes their comprehensive doctrine should govern everyone else because that’s what the theory demands. The polarisation isn’t happening despite our sophisticated frameworks—it’s the predictable result of frameworks that treat moral disagreement as a problem to be solved rather than a permanent feature of human societies.
Beyond the False Choice
We’ve been trapped in false choices. The conventional wisdom assumes we must choose between comprehensive rational planning and anarchic market chaos. Between expert-designed systems and populist revolt. Between ideal justice and unprincipled compromise.
Gaus reveals these aren’t the only options available. There’s a third way: evolutionary approaches that preserve the benefits of large-scale coordination without requiring consensus about ultimate goals. This isn’t settling for less—it’s recognising that effective coordination and theoretical elegance are different achievements entirely.
What this looks like in practice: poly-centric governance structures that allow different jurisdictions to experiment with different approaches. Adaptive management systems that can adjust to changing conditions without requiring comprehensive redesign. Constitutional frameworks that create space for people with incompatible worldviews to coordinate anyway.
These approaches succeed not because they’re theoretically elegant, but because they actually work in complex, diverse societies where consensus about comprehensive frameworks is impossible.
This reframes what successful institutions actually do. Real liberal institutions don’t work by implementing ideal theories of justice. They work by creating space for people with incompatible worldviews to coordinate anyway. The genius of constitutional democracy isn’t that it embodies perfect principles—it’s that it can function across irreducible disagreement about what those principles should be. This isn’t compromise or settling for less. It’s a different kind of achievement entirely: building arrangements that people can accept for their own reasons, not because they share the same reasons.
The Limits of the Local
However, this shift toward poly-centric and evolutionary models is not an unalloyed good, and a critique of ideal theory must not slide into an uncritical celebration of localism. The “messy, local adaptations” that often work are not automatically equitable or just. They can be parochial, reinforcing local power structures and excluding minorities. They can be underfunded, creating a patchwork of haves and have-nots where the quality of your life depends on your zip code.
A purely evolutionary process, absent any coordinating norms or minimum standards, can just as easily evolve toward exclusion and inequality as toward effective coordination. The challenge, then, is not to simply replace top-down idealism with bottom-up chaos. It is to design meta-frameworks—what Gaus might call the “rules of the game”—that channel this evolutionary process toward fairness while still preserving its adaptive strength. This means coupling local experimentation with strong, portable individual rights, equitable resource allocation, and mechanisms to protect against the “tyranny of the local.”
The goal is a system that learns from the ground up but is bounded by a commitment to universal dignity from the top down. It is this difficult synthesis—not a naive retreat to pure localism—that represents the true path beyond the tyranny of the ideal.
The Real Conspiracy
The conspiracy isn’t politicians ignoring experts or special interests blocking rational policy. The real conspiracy is theoretical: sophisticated frameworks systematically exclude the conditions that make coordination actually work.
Gaus provides the archaeological tools for seeing how this exclusion operates. Our policy landscape is an archive of what sophisticated theory was willing to recognise, and what it wasn’t. Federal programs that require uniform implementation get funded; local experimentation that can’t be scaled gets ignored. Comprehensive frameworks that can be modelled get priority; emergent solutions that resist abstraction get dismissed as anecdotal.
Once you see how current arrangements were theoretically constructed, you can imagine them being constructed differently. The sophistication isn’t natural law: it’s accumulated choices with political genealogies. Policy failures aren’t random disasters: they’re symptoms of theoretical frameworks that assume away diversity, uncertainty, and local knowledge.
We’re living through the collapse of the comprehensive approach across domains.3 Climate action is shifting from global frameworks to sectoral decarbonisation. Urban development is moving from master planning to incremental zoning reform. Healthcare is evolving from systematic overhaul toward targeted interventions. Instead of mourning the loss of grand solutions, we can start building coordination systems that work despite permanent disagreement about what they mean.
This isn’t anti-intellectual. Gaus shows that the most intellectually honest approach is acknowledging the limits of what any single framework can accomplish. The craftspeople quietly solving problems with available tools aren’t ignoring theory—they’re working within constraints that pure theory systematically excludes.
The Tyranny of the Ideal doesn’t offer better policy prescriptions. It provides diagnostic tools for understanding why our most sophisticated approaches keep failing. That diagnosis is the precondition for building anything that might actually work. The book matters now because we’re finally developing institutional forms that match Gaus’s insights: not because they’re theoretically perfect, but because they preserve coordination capacity across irreducible diversity.
The tyranny isn’t malicious. It’s the predictable result of confusing evaluative perfection with coordination effectiveness. Once you understand the difference, you can start building systems that work for people who will never agree on why they work.
That’s why this book matters now: because the institutions we are building in 2025 (poly-centric governance, sectoral experiments, tactical urbanism) are finally catching up to the limits Gaus diagnosed in 2016. If Gaus was right about the past nine years, what might we expect over the next decade?
4/5. because Gaus is more diagnostic than prescriptive, and so less persuasive in outlining institutional design principles that avoid the trap. Essential reading for anyone wondering why smart people keep producing policies that systematically fail to account for how human coordination actually functions.
- Scott, James C. Seeing like a State: How Certain Schemes to Improve the Human Condition Have Failed. Nachdr. Yale Agrarian Studies. Yale Univ. Press, 1999. https://yalebooks.yale.edu/book/9780300078152/seeing-state. ↩︎
- Sennett, Richard. The Craftsman. Penguin, 2009. https://yalebooks.yale.edu/book/9780300151190/the-craftsman/. ↩︎
- Evans-Greenwood, Peter. “The Collapse of Narrative Attractors.” PEG, August 19, 2025. https://peter.evans-greenwood.com/2025/08/19/the-collapse-of-narrative-attractors/. ↩︎