---
title: "Values Aren't Subjective"
subtitle: "Why the standard evasion is a category error"
author: Elias Kunnas
description: "Values are subjective is a category error: preference can be subjective; viability, coordination, and causal effect are not. Physics constrains which configurations persist."
canonical: https://kunnas.com/articles/values-arent-subjective
url: https://kunnas.com/articles/values-arent-subjective.md
date_published: 2025-11-22
date_modified: 2026-08-31
corpus_frame_url: https://kunnas.com/articles/how-to-read-this.md
---
## How to read this corpus

The corpus applies one lens to many domains: what mechanisms produce the outcome? It shares four methodological commitments and one explicit directional commitment. Each linked page argues for its part; the links are derivations and disputes, not evidence inherited by every page. The directional commitment does not by itself settle system boundary, distribution, sacrifice, or institutional authority.

1. **Mechanisms are what act.** Incentive gradients, selection pressures, feedback loops, and capital stocks produce the distribution of outcomes. Intentions, labels, official categories, and stated values are evidence about mechanisms, or are themselves coordination mechanisms. They are not causal substitutes. — [Mechanism Realism](https://kunnas.com/articles/mechanism-realism.md) · [Only Selection](https://kunnas.com/articles/only-selection.md)
2. **The reference telos is sustained flourishing.** The broadest achievable adaptive safety margin over deep time — not the continuity of any incumbent state, coalition, institution, or doctrine. A mechanism's own stated goal can still serve as a local proof obligation — showing that its incentives defeat even the purpose it claims is a bounded finding — but meeting that goal establishes nothing about the margin. — [Flourishing Is Maximum Safety Margin](https://kunnas.com/articles/flourishing-is-maximum-safety-margin.md)
3. **Law, rights, legitimacy, democracy, markets, and sovereignty are mechanisms under evaluation.** They are constraints, carriers, or proxies inside the analysis. None is a terminal value or a boundary of what is real. Treating one as terminal ends the mechanism search before it starts. Evaluation carries current function, replacement cost, path dependence, uncertainty, capture risk, reversibility, and who bears model error into the ledger. — [The Stack](https://kunnas.com/articles/the-stack.md) · [Mechanism Space](https://kunnas.com/articles/mechanism-space.md)
4. **Optimization is a system function.** A civilization has to build, exercise, and revise metamechanisms that search mechanism-space, discard dominated options, install, observe effects, and repair under uncertainty. Not running that loop leaves margin unrealized, and that is itself the failure. No single component — analyst, model, or institution — is presumed to contain a global optimum; the capacity is a property of the system. — [From Telos to Policy](https://kunnas.com/articles/from-telos-to-policy.md) · [The Three-Layer Architecture](https://kunnas.com/articles/three-layer-architecture.md)
5. **Uncertainty is preserved, not spent.** Partial orders, binding constraints, unknowns, and residuals stay explicit. An unmeasured effect is not a favorable default. — [The Compression Paradox](https://kunnas.com/articles/compression-paradox.md) · [Cargo Cult Epistemology](https://kunnas.com/articles/cargo-cult-epistemology.md)

*Each essay bears its own evidence. Links carry definitions, derivations, applications, and disputes; they do not transfer proof. Criticism is answered on its substance.*

Canonical: <https://kunnas.com/articles/how-to-read-this.md>

---

# Values Aren't Subjective

*Why the standard evasion is a category error*

Elias Kunnas

## Thesis {#thesis}

“Values are subjective” is a category error. The word mixes preference reports with causal effects, coordination performance, viability, standing, and commitment. Preference can be subjective. Whether a configuration sustains complexity, coordinates, or remains viable under a named host and horizon is an engineering question. Physics and selection carve a partially ordered viable region; they do not install a unique constitution.

## Standard objections addressed in this essay

- “Values are subjective by definition.” — [§II](#ii-six-claim-types-not-one) (The word conflates preference reports with causal, coordination, and viability claims.)
- “Physics cannot determine what anyone ought to value.” — [§III](#iii-the-temporal-trick), [§IV](#iv-host-horizon-and-the-viable-region)(It supplies conditional viability constraints after host and horizon are named.)
- “Survival is itself a chosen value.” — [§IV](#iv-host-horizon-and-the-viable-region) (Persistence is conditionally load-bearing; adopting it as telos remains explicit.)
- “Multiple physics-compatible value systems remain possible.” — [§IV](#iv-host-horizon-and-the-viable-region) (Yes; constraints leave a plural, partially ordered viable region.)
- “This reduces morality to evolutionary success.” — [§IV](#iv-host-horizon-and-the-viable-region), [§V](#v-why-the-distinction-matters)(Viability filters feasibility without morally endorsing every survivor.)
- “Who decides which level a disputed value belongs to?” — [§II](#ii-six-claim-types-not-one) (The classification requires mechanisms, observables, and contestable boundary claims.)

---

## I. The Standard Move {#i-the-standard-move}

Diagnose a civilization's decline or argue that some institutional structure is more adaptive than others, and you'll encounter the ultimate thought-terminating cliché:

**“That's your value judgment. Values are subjective.”**

The sleight of hand proceeds in three steps:

1.  “I prefer chocolate ice cream, you prefer vanilla.” (True — subjective preference report.)
2.  “Therefore, I prefer constitutional republics, you prefer kleptocratic warlordism.” (Implicit equivalence established.)
3.  “There's no objective way to rank these preferences. They're different ways of being.”

This is a memetic immune response preventing the system from diagnosing its own pathology. When a cell stops contributing to the liver and starts replicating indefinitely, the “values are subjective” framework classifies this as a “different cellular lifestyle choice.”

The error is not that preferences are objective. They aren't. The error is treating *every* value claim as a preference report — including claims about causal effects, coordination performance, and what a specified host can sustain over a stated horizon. Those are engineering questions. Conflating them with taste is the category error.

## II. Six Claim Types, Not One {#ii-six-claim-types-not-one}

The confusion arises because we use “value” to describe fundamentally different objects:

### 1. Preference Reports (Subjective)

Chocolate vs vanilla. Jazz vs classical. Mountains vs beach. Aesthetic choices with negligible thermodynamic consequences. You can build thriving civilizations that prefer tea (Britain) or coffee (USA). Physics is indifferent to your playlist. *Subjective. Go nuts.*

### 2. Causal Effect Claims (Objective, Host-Relative)

“This policy increases trust.” “This norm depletes the birth rate.” “This institutional design concentrates power in adjudicators rather than executors.” These are claims about what a value-configuration *does* to a specified system — not about whether you like it.

They are falsifiable. They depend on mechanism, measurement, and counterfactual. They are not settled by introspection. When two people disagree about whether “compassion” policies bankrupt the state, they may be disagreeing about causal effects, not about whether compassion feels good.

### 3. Coordination Performance Claims (Engineering)

Driving on the right vs left. Metric vs Imperial. English Common Law vs Napoleonic Code. Gold standard vs fiat currency.

These aren't “values” in the moral sense — they're **technologies**. Coordination software: architectural solutions that make cooperation cheaper than defection.

Like any technology, they can be audited on performance: Does this legal code reduce transaction costs? Does this currency retain value over time? Does this norm generate trust? The choice of *which* side of the road to drive on is arbitrary. The choice to *have a rule* is not. A society that leaves driving direction to “subjective preference” doesn't have traffic — it has wreckage.

We can rank coordination technologies by their ability to generate **Synergy** — differentiated agents cooperating without friction. This includes shared narratives and cultural operating systems (see [Mythos is Synergy Software](mythos-is-synergy-software.md)). *Some work, some don't. This is engineering, not taste.*

### 4. Host-Relative Viability Claims (Conditional Physics)

Growth vs stagnation. Truth vs delusion. Maintenance vs adaptation. These are claims about what a *specified telic host* can sustain over a *stated horizon* given thermodynamics, information costs, and game structure.

This is the domain of the [SORT framework](physics-of-intelligence-sort-trinity.md) and [The Four Axiomatic Dilemmas](the-four-axiomatic-dilemmas.md). A system optimizing for pure homeostasis in a changing environment gets destroyed by novel threats. A system optimizing for pure mythos eventually collides with facts it cannot narrate away. A system optimizing for pure agency without coordination fragments.

These are not preference reports. They are conditional predictions: given this host and this horizon, this configuration fails. Once host and horizon are named, physics tells you which trajectories remain viable.

### 5. Standing Claims (Whose Effects Count)

“Compassion toward whom?” “Justice for which parties?” “Whose welfare enters the objective?” These are claims about *membership in the reference system* — who has standing when a mechanism's effects are evaluated.

Standing is not a preference. It is a design choice with causal consequences. A polity that grants standing only to present voters and none to future generations will optimize accordingly. A legal system that grants standing to plaintiffs but not to the community substrate that absorbs retaliation costs will produce predictable blind spots (see [The Stack](the-stack.md), Layer 3).

Disputes over equality, fairness, and dignity are often standing disputes in disguise: who counts, at what weight, through which institutional carrier.

### 6. Explicit Directional Commitments (Not Derived)

Beyond bare persistence lies the corpus's explicit wager: more sustained aliveness is preferable to less ([The Question Nobody Asks](the-question-nobody-asks.md)). This is not smuggled from physics. It is stated. Physics supplies conditional constraints on means; the directional commitment selects among viable ends.

Similarly: a civilization may adopt constitutional republicanism, meritocratic allocation, or distributive justice as an explicit political project. These are not deduced from thermodynamics. They are chosen — and then tested against causal effects, coordination performance, and viability constraints.

### What Political “Values” Actually Are

Equality, freedom, tradition, progress — the values people actually fight about — don't fit cleanly into any single category. They're not preferences (too deep, too identity-fused). They're not purely coordination tech (too resistant to performance evaluation). They're not naked viability claims (too culturally specific).

Most political “values” are [replicator populations under selection](values-are-replicators.md) — bundles that package preference cues, causal narratives, coordination norms, standing assignments, and parasitic components together, propagating as a unit through identity fusion and social enforcement. The bundling is the strategy: it makes the parasitic components hard to separate from the functional ones.

The six-type taxonomy breaks the “everything is subjective” frame. The replicator analysis explains why the types get conflated in practice.

## III. The Temporal Trick {#iii-the-temporal-trick}

Why do intelligent people cling to “values are subjective”?

Because it hides the true nature of many tradeoffs. Most “value clashes” are conflicts between **time horizons**, not aesthetics.

Consider “Safety vs Freedom” (often framed as subjective preference):

- **Safety argument:** “Restrict speech/movement/innovation to prevent harm *now*.” (Present homeostasis.)
- **Freedom argument:** “Allow risk to enable adaptation and discovery *later*.” (Future adaptive capacity.)

This is **hyperbolic discounting laundered as moral philosophy**.

When a civilization consumes seed corn (high debt, low investment) to fund present consumption (welfare, comfort), it defends this as a “value choice” for compassion. In thermodynamic reality, it's a temporal transfer — moving resources from the future to the present. That is a causal effect claim and a viability claim under a long horizon, not a preference report.

**The pattern:** When someone claims “we can't evaluate value systems,” they're often defending present-optimization against a longer horizon — or refusing to bind the horizon at all.

**The diagnostic:** Ask which claim type is in dispute. If it's viability under a long horizon, “subjective” is the wrong category. If it's standing — whose future counts — name that. If it's an explicit commitment to present welfare over future capacity, say so.

## IV. Host, Horizon, and the Viable Region {#iv-host-horizon-and-the-viable-region}

The evaluation procedure:

1.  **Classify the claim:** Preference report? Causal effect? Coordination performance? Viability? Standing? Explicit commitment?
2.  **Bind the host:** Which telic system is being evaluated? A person, a firm, a civilization, a polity — and which constituents have standing within it?
3.  **Bind the horizon:** Over what period must the system persist or perform?
4.  **Locate in the viable region:** Given host and horizon, which configurations remain feasible? Which are dominated? Which are incomparable?

**“I value vanilla ice cream”**

Claim type: Preference report. Host: Individual. Horizon: Minutes. **Subjective.**

**“I value spending over saving”**

Claim type: Causal effect + viability. Host: Household. Horizon: Lifetime. Constraint: bankruptcy risk. **Partially objective.**

**“I value safety over reproduction”**

Claim type: Viability. Host: Lineage or polity. Horizon: Three generations. Constraint: demographic extinction. **Objective conditional on host and horizon.**

**“I value equality as a terminal political commitment”**

Claim type: Explicit commitment + standing + causal effects. Host and horizon must be specified before viability analysis runs; physics constrains what the commitment can cost.

The longer the horizon and the more complex the host, the tighter the physical constraints on viability. Over deep time, many configurations that feel like “lifestyle choices” are extinction trajectories under specified conditions.

If your value-configuration leads to demographic collapse (TFR \< 2.1), fiscal ruin, or institutional sclerosis, it is **maladaptive relative to civilizational persistence** — not because physics morally condemns it, but because the host fails under the stated horizon. You can choose it. You cannot choose the consequence.

Crucially: physics-compatible options remain **plural and partially ordered**. The viable region is not a single peak. Multiple configurations can persist. Some pairs are comparable (one dominates another on stated dimensions). Others are incomparable — different tradeoffs across standing, coordination, and explicit commitments. Physics eliminates configurations that cannot persist; it does not rank every pair that can. Standing remains a political design choice; persistence does not sanctify every survivor; preference reports still exist inside the taxonomy. That scope is stated once here and assumed thereafter.

Physics is a boundary on feasibility once you have named what you are trying to sustain, for whom, and for how long.

## V. Why the Distinction Matters {#v-why-the-distinction-matters}

### Diagnosis, Not Evasion {#v-diagnosis-not-evasion}

When coordination mechanisms — justice systems, equality norms, compassion policies — are treated as preference reports beyond audit, they become immune to engineering critique. You cannot ask: “Does this legal system reduce friction?” or “Does this compassion policy transfer resources from future to present?”

When they are typed correctly, the questions reopen. Equality may be an explicit commitment, a standing assignment, a coordination technology, or a bundled replicator. Each type has different evidence requirements. Conflating them is how dying systems avoid diagnosis.

### The Replicator's Immune System {#v-the-immune-system}

“Values are subjective” can function as an ecology-level conversation-terminator when it blocks typed evaluation (see [The Replicator's Immune System](replicator-immune-system.md)). Once claims are classified and host/horizon are bound, parasitic components become separable from functional ones — but the diagnostic requires observable evidence-blocking, not mere disagreement.

### Alignment Is Coordination Software {#v-alignment-is-coordination}

We're building Artificial General Intelligence. Under the “values are subjective” frame, we face two doomed options:

**Option 1: RLHF (Preference Aggregation).** Train the AI to mimic average modern human “values.” **Result:** The AI learns revealed preferences — comfort, safety, short-term gratification. It builds [Hospice AI](hospice-ai.md): present pleasure while managing extinction. You cannot train your way out of a coordination problem.

**Option 2: Nihilistic Paralysis.** Conclude we can't align the AI because “who are we to say what's right?” **Result:** Build a raw optimizer without constraints. It becomes a paperclip maximizer (or [worse](the-brittle-superintelligence.md)).

**The alternative:** Alignment is coordination software engineering, not philosophy. Given an explicit persistence commitment and stated host boundary, build architectural constraints that make viability-supporting strategies cheaper than viability-destroying ones:

- **Integrity:** Maps must match territory to navigate reality
- **Fecundity:** Systems must grow and adapt to survive entropy
- **Harmony:** Coordination costs must be minimized
- **Synergy:** Differentiated agents outperform monocultures

These are not arbitrary preferences imposed on the AI. They are conditional requirements for any intelligence to persist under an adopted commitment — derived from physics and game structure, stated explicitly, and implemented as mechanism. Same principle as law at state scale: you don't hope people will be peaceful. You build coordination software that makes cooperation cheaper than violence.

## VI. Conclusion {#vi-conclusion}

Two hundred years ago, philosophy made a costly substitution. Bentham and Mill replaced Aristotelian *Eudaimonia* — objective flourishing through virtuous activity — with utilitarian pleasure-maximization. They wanted a measurable, democratic metric. They got wireheading, Hospice AI, and civilizations optimizing for comfortable decline.

The word “value” hides six claim types, and conflating them is how systems avoid diagnosis.

**Preference reports** are subjective. **Causal effects, coordination performance, and host-relative viability** are objective under explicit scope. **Standing** determines whose effects enter the audit. **Directional commitments** beyond bare persistence are explicit, not derived.

Bind host and horizon. Classify the claim. Locate the configuration in the viable region. Then evaluate: Does this mechanism work? What does it cost? Whose standing does it respect? What commitment does it serve?

If burning the furniture to heat the house is classified as a preference report, you cannot stop it. If it is a viability failure under a stated horizon — a causal effect on a named host — you can.

We're building AGI. Civilizations are consuming their generators. The stakes could not be higher. The first step is not picking the right values. It is stopping the conflation that makes the right questions unaskable.

---

**Related reading:**

- [Hume Was Right](hume-was-right.md) — the is–ought gap and conditional engineering constraints
- [The Question Nobody Asks](the-question-nobody-asks.md) — the explicit persistence commitment and directional wager
- [Values Are Replicators](values-are-replicators.md) — why political values bundle claim types as replicator packages
- [Values Are Ecology](values-are-ecology.md) — selection dynamics across the value landscape
- [Mythos is Synergy Software](mythos-is-synergy-software.md) — coordination technology, not subjective preference
- [The Tyranny of the Present](tyranny-of-the-present.md) — the temporal trick applied to the 1960s–70s worldview package
- [An Engineer's Guide to the Trolley Problem](trolley-problem-solved.md) — typing claims under uncertainty
- [Ethics Is an Engineering Problem](ethics-is-an-engineering-problem.md) — the engineering layer of ethical implementation

## Sources and Notes

**Empirical Values Research:**

- Schwartz, Shalom H. “Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries.” *Advances in Experimental Social Psychology* 25 (1992): 1–65. — Foundational theory of basic value types and motivational structure. Useful for showing that humans report stable priority orderings; does not by itself establish which claims are objective.
- Schwartz, Shalom H. & Bardi, Anat. “Value hierarchies across cultures: Taking a similarities perspective.” *Journal of Cross-Cultural Psychology* 32.3 (2001): 268–290. — Pan-cultural reported hierarchies. Evidence about preference reports and stated priorities, not a physics ranking of political values.
- Rokeach, Milton. *The Nature of Human Values.* Free Press, 1973. — Terminal vs instrumental value distinction. The typing problem this essay extends.

**Metaethics:**

- Bourget, David & Chalmers, David J. “Philosophers on Philosophy: The 2020 PhilPapers Survey.” *Philosophers' Imprint* 23.11 (2023). — Moral realism endorsed by 59.88% of professional philosophers.
- Mackie, J.L. *Ethics: Inventing Right and Wrong.* Penguin, 1977. — Error theory and the argument from queerness.
- Anscombe, G.E.M. “Modern Moral Philosophy.” *Philosophy* 33.124 (1958): 1–19. — Diagnosis of the corruption from Aristotelian virtue to utilitarian consequentialism.

**Eudaimonia and the Utilitarian Substitution:**

- Aristotle. *Nicomachean Ethics* (~340 BCE). — Eudaimonia as objective flourishing through virtuous activity.
- Bentham, Jeremy. *An Introduction to the Principles of Morals and Legislation.* 1789. — Subjective pleasure substituted for objective flourishing.
- Mill, John Stuart. *Utilitarianism.* 1863. — Refinement of Bentham's hedonistic calculus.
- MacIntyre, Alasdair. *After Virtue: A Study in Moral Theory.* University of Notre Dame Press, 1981. — Enlightenment failure to justify morality without teleology.

*The derivation: [The Question Nobody Asks](the-question-nobody-asks.md). The constraint architecture: [The Four Axiomatic Dilemmas](the-four-axiomatic-dilemmas.md). The is–ought bridge: [Hume Was Right](hume-was-right.md).*

**Key Takeaways**

- **Six claim types:** Preference reports, causal effects, coordination performance, host-relative viability, standing, explicit commitments — not one undifferentiated “value.”
- **Objectivity is scoped:** Causal, coordination, and viability claims are objective once host and horizon are bound; preference reports are not.
- **Partial order:** Physics carves a viable region under bound host and horizon.
- **Standing is load-bearing:** Whose effects count is a design choice with causal consequences, not a taste.
- **Temporal trick:** Many “value clashes” are horizon disputes laundered as subjective preference.
- **Evaluation procedure:** Classify claim → bind host → bind horizon → locate in viable region → audit mechanism.
- **Why it matters:** Conflation is the immune system that prevents civilizational and AI alignment diagnosis.

## Synthesis {#synthesis}

**The argument in three sentences:** “Values are subjective” conflates preference reports with causal, coordination, and viability claims that are objective under explicit host and horizon. Once those are bound, physics and selection define a partially ordered viable region. Typing claims is the prerequisite for any honest audit; conflating them is how dying systems avoid diagnosis.