The truncated gradient: correct diagnosis, wrong prescription
The libertarian diagnosis of state dysfunction is substantially correct.
Hayek identified the knowledge problem: central planners cannot aggregate the dispersed information that markets process automatically. No committee can replicate what prices compute for free.
Public choice theory identified the incentive problem: politicians optimize for re-election, bureaucrats optimize for budget expansion, interest groups optimize for rent extraction. The state doesn't serve "the public interest" because no mechanism ensures that it does.
Regulatory capture theory identified the predator problem: the entities that regulation is supposed to constrain end up writing the regulation. The referee joins the team with the most money.
These aren't ideological claims. They're derived from the mechanism structure. The libertarian analytical tradition—from Mises through Hayek through Buchanan through Friedman—didn't merely observe that states malfunction. They derived, from the architecture of available governance mechanisms, why states necessarily malfunction. Given self-interested agents + electoral incentives + information asymmetry, capture is deducible. Given distributed knowledge + central planning, failure is deducible. These are deductions from mechanism structure, not anecdotes.
And the deductions are correct—within the mechanism vocabulary they operate on. The diagnosis is real. What follows from it is where things go wrong.
When you correctly diagnose that the state's feedback architecture is broken, three options exist:
The mainstream political debate oscillates between Options 1 and 2. The left argues for more state. The right argues for less state. Libertarians are the right's most intellectually rigorous wing, following Option 2 to its logical conclusion: minimize the state to the smallest possible footprint.
Option 3 is almost entirely absent from the discourse.
This is the truncated option set. When Option 3 doesn't exist in your conceptual vocabulary, Option 2 is the locally correct gradient. If the engine is broken and you can't redesign the control system, removing the engine is better than letting it run haywire. Libertarianism is the rational response to a world where telocracy — governance with an explicit objective function — hasn't been articulated.
The libertarian proofs operate on a closed set of governance primitives: {central planning, markets, elections, bureaucracy, regulation}. Within that set, the deductions hold. You genuinely cannot make the state efficient using those primitives alone. The proofs are valid. But valid proofs within a bounded vocabulary don't prove impossibility in an expanded vocabulary.
It's like proving that no polynomial solves the general quintic (Abel-Ruffini theorem). The proof is airtight—within polynomials. Then Galois theory arrives and the problem transforms entirely. The theorem remains true. The problem space expanded around it.
Option 3 requires conceptual primitives that weren't in the governance vocabulary until recently:
Mechanism design theory (Hurwicz, Myerson, Maskin — Nobel 2007) formalized how to engineer incentive structures that produce desired outcomes. Before mechanism design, "fix the incentives" was a wish, not a discipline.
Computational governance — the idea that governance is an information-processing problem solvable by better algorithms, not just better people — only became thinkable with the computational paradigm. You can't propose "the state as search function" before the concept of a search function exists.
Selection pressure analysis applied to institutions — understanding that selection pressure, not moral failure, produces institutional dysfunction — requires evolutionary thinking applied outside biology. This cross-pollination is recent.
AI alignment formalized the problem of aligning a powerful optimization process with goals it wasn't trained to pursue. The governance alignment problem is structurally identical, but the parallel only became visible once AI alignment named the pattern.
Hayek had half the toolkit. He understood distributed information processing (markets). He understood the knowledge problem (central planning fails). But he stopped at markets. He didn't ask: what if you could build institutional architectures with the information-processing advantages of markets AND the coordination capacity of states?
He didn't ask because the question was premature. The engineering tools to answer it didn't exist yet.
In optimization, gradient descent means: look at where you are, determine which direction improves the objective function, take a step in that direction. Repeat.
Gradient descent on a truncated landscape produces predictable failure. If the landscape only has two dimensions — more state, less state — then the gradient correctly points toward less state whenever the current position is "too much broken state." The algorithm is working. The landscape is wrong.
This is what happened to the libertarian intellectual tradition. The gradient is locally correct. Given {more state, less state}, less state produces better outcomes at the margin when the state is already dysfunctional. Fewer regulations means less capture. Less redistribution means less moral hazard. Smaller bureaucracies means less Parkinsonian expansion.
But the gradient leads to a local optimum, not the global optimum. Minimum viable state is better than maximum dysfunctional state. But a state with properly designed feedback architecture is better than both.
The core claim: Libertarianism is gradient descent on a 2D landscape. The actual landscape is 3D. The third dimension — feedback architecture quality — is where the global optimum lives. Libertarians found the best point on a plane. The mountain is above them, in a direction they can't see.
Telocracy is governance with an explicit, measurable objective function — and institutional architecture to pursue it.
The key features that distinguish it from both "more state" and "less state":
Explicit telos. The state has a defined purpose: maximize civilizational flourishing over infinite time. This isn't chosen by ideology — it's derived from physics. A system that doesn't maintain organized complexity against entropy dies. The telos is "don't die" extended to civilizational timescales.
Measurement infrastructure. A Fourth Branch that continuously measures whether institutions produce their stated outcomes. Not process compliance — actual results. The mechanism that current governance entirely lacks: feedback from reality to policy.
Automatic constraints. Switzerland's debt brake is a prototype: a constitutional rule that fires automatically when fiscal parameters exceed thresholds. Politicians decide what to cut, not whether to cut. Extend this to demographics, institutional health, infrastructure. Hard rules that ground the democratic ratchet — the tendency of electoral systems to expand obligations faster than capacity.
Sunset clauses. Every institution expires unless it demonstrates continued function. The default is death. Survival requires proof. This inverts the current architecture where institutions are immortal regardless of performance — which is the failure mode that libertarians correctly identify.
Full accounting. Every policy decision accompanied by its full cost across all capital types and time horizons. No complexity laundering. No temporal displacement. The information asymmetry that enables dysfunction — the information asymmetry that Hayek identified — is attacked directly.
Note what this is NOT:
A car's engine is misfiring. The mechanic says: "The control system is sending wrong signals to the fuel injectors." Two responses:
Response A: "Remove the engine. We'll push the car."
Response B: "Replace the control system."
Response A is libertarianism. It correctly identifies that the engine is malfunctioning. It correctly identifies that the malfunction comes from the control system. It incorrectly concludes that the solution is to remove the engine rather than fix the control system.
The problem with Response A is that it destroys the substrate that the actual fix requires.
A state — even a dysfunctional one — provides coordination infrastructure that no market can replicate. Markets require the state to define and enforce property rights — libertarians know this (it's the "night-watchman state" concession). But the same logic extends: pollution, commons depletion, and intergenerational cost-shifting are coordination failures that fall outside any individual transaction — Hayek acknowledged this but didn't develop it. Infrastructure, demographic sustainability, and civilizational defense require coordination across timeframes longer than any market actor's discount horizon. And you need institutional infrastructure to build the Fourth Branch — dismantling the state before building its replacement is demolishing the house before the new one is ready.
The libertarian who removes the state to fix governance is like the surgeon who removes the patient to fix the disease. The disease is real. The treatment kills the host.
These aren't just functions the state performs. They're capital stocks that require active maintenance: demographic capital (population sustainability), social capital (trust, cooperation norms), institutional capital (functioning legal and administrative structures), cognitive capital (accumulated knowledge, education systems), infrastructure capital (physical networks), and environmental commons. Each depreciates. Each requires investment. Each is invisible to markets because none carries a price signal.
A civilization that maintains only its financial and physical capital while letting demographic, social, and institutional capital erode is drawing down the balance sheet. The numbers look fine until they don't — the same way a company looks profitable right up until the deferred maintenance bankrupts it. Full accounting — across all capital types, across all time horizons — is the minimum condition for knowing whether you're building or consuming.
The libertarian tradition could afford this blind spot because these stocks were abundant when the tradition developed. Hayek could take social trust for granted. Friedman could assume functional institutions. The well was full. It isn't anymore. The libertarian who budgets for defense alone is maintaining one line on a balance sheet with a dozen entries. The other lines are depreciating unmonitored. Not because markets failed — markets were never designed to maintain capital stocks without price signals. This is the class of problem that requires institutional architecture, and it's the class of problem that "less state" cannot address.
The surprising conclusion: libertarianism and telocracy share more diagnostic ground than either shares with the conventional left-right spectrum.
| Insight | Libertarian Version | Telocratic Version |
|---|---|---|
| State dysfunction is structural, not personal | Yes (public choice theory) | Yes (alignment problem) |
| Incentives determine outcomes, not intentions | Yes (Hayek, Friedman) | Yes (architecture over disposition) |
| Information problems are central | Yes (knowledge problem) | Yes (feedback architecture) |
| Current governance is structurally captured | Yes (regulatory capture) | Yes (alignment problem) |
| Democratic ratchet is real | Yes (Buchanan) | Yes (obligations > capacity) |
| Reform from within is almost impossible | Yes (government failure) | Yes (misaligned agents control reform) |
The divergence is narrow but critical:
| Question | Libertarian Answer | Telocratic Answer |
|---|---|---|
| Can the state's architecture be fixed? | No → minimize it | Yes → rewire it |
| Can markets handle coordination? | Mostly yes | Within horizons; not across them |
| Is there a third option beyond more/less state? | Not considered | Yes: better feedback architecture |
A libertarian who encounters telocracy faces a genuine update. The diagnosis they trust — public choice, knowledge problem, regulatory capture, democratic ratchet — all survive intact. What changes is the prescription: not "remove the engine" but "install a better control system."
In fact, telocracy delivers libertarian goals more completely than libertarianism does — by libertarian criteria:
| Libertarian Goal | Libertarianism Delivers | Telocracy Delivers |
|---|---|---|
| Minimize discretionary state power | Remove state functions | Replace discretion with automatic constitutional constraints — a debt brake removes more political discretion than defunding a ministry |
| Maximize feedback quality | Market prices | Market prices + Fourth Branch measurement of capital stocks that markets can't price |
| Institutional accountability | Competition (Tiebout sorting) | Competition + sunset clauses + mandatory outcome measurement — institutions that can't demonstrate function die automatically |
| Prevent capture | Minimize what can be captured | Make capture structurally unproductive — automatic constraints fire regardless of who controls the institution |
An automatic constitutional constraint is less discretionary state power than a politician making the same decision. A sunset clause that kills unproductive institutions is more libertarian than a deregulation bill that has to survive lobbying. The Fourth Branch that makes all costs public is more Hayekian than abolishing the department that hides them — because it fixes the information asymmetry rather than removing one source of it.
The obvious counter-example: the EU's fiscal rules were automatic on paper and suspended in practice. The difference is architectural. The EU's Stability Pact was default-allow — spending continued unless someone actively punished the violator, and the violators voted on their own punishment. The Swiss debt brake is default-deny — spending stops automatically unless revenue covers it. Politicians don't decide whether to cut; they decide what to cut. When inaction produces discipline rather than drift, the constraint doesn't need political will to hold.
Even the Swiss prototype has a blind spot: the debt brake constrains fiscal capital but doesn't distinguish consumption from investment. Under pressure, politicians cut infrastructure (deferred maintenance) rather than entitlements — depleting one capital stock to preserve another. This is why a single automatic constraint isn't enough. Full accounting across all capital types is the generalization: the same default-deny architecture, extended to every line on the civilizational balance sheet.
The update is: the option set was incomplete. "Less state" was the best available answer. It is no longer the best available answer. A better one exists — one that preserves every libertarian diagnostic insight while avoiding the substrate-destruction problem.
Libertarianism has been the most intellectually serious critique of state dysfunction for 80 years. Its prescriptions have been partially adopted — deregulation, privatization, monetary discipline — but as a complete program, no polity has achieved the minimal state. The reason is the engine problem: people intuitively understand that removing the state removes coordination capacity. "Who will build the roads?" is a meme, but it encodes a real objection. The libertarian response — "the market will" — is unconvincing for functions with externalities, long horizons, or coordination requirements beyond bilateral exchange.
Telocracy resolves this impasse. It takes the libertarian diagnosis seriously — more seriously than the mainstream left or right, which deny or minimize state dysfunction. And it provides a prescription that doesn't require people to accept the implausible claim that markets can handle everything.
The obvious libertarian counter: who builds it, and what prevents the builders from being captured? These are engineering problems with engineering answers — constitutional independence, international oversight, capital funding instead of annual budgets, mandatory transparency, automatic triggers that bypass political discretion. The Fourth Branch details the structural requirements. A concrete implementation proposal for Finland demonstrates that capture-resistance can be designed into institutional architecture, not assumed.
The libertarian who discovers telocracy doesn't abandon their analysis. They complete it. The truncated gradient was pointing in the right direction — away from dysfunctional state architecture. It just couldn't see the peak that requires going through the state to reach: a state with feedback architecture that actually works.
The thesis: Libertarianism is the most important intellectual tradition that almost nobody follows, because it diagnosed the disease correctly but prescribed amputation when surgery was needed. The surgery — telocracy — requires tools that only recently became available. The libertarian who updates on the expanded option set becomes the most valuable ally the actual fix has.
This draws from Aliveness, a framework for understanding what sustains organized complexity over time.
Governance series: Diagnosis → Telocracy → Institution → Full Accounting → Libertarianism Is an Incomplete Solution
Related: