I see ethics as an emergent technology that natural minds produce while living together, not a set of instructions written into the cosmos. Value exists only in evaluating minds, but the patterns those minds produce are deeply constrained by the physical and game-theoretic realities of the world they share.
Value does not exist as an intrinsic property of matter; it exists in evaluating minds. Yet moral systems are far from arbitrary. Agents living under shared physical laws face recurring constraints: scarce resources, vulnerable bodies, imperfect information, mutual dependence, repeated interaction, and the constant tradeoff between cooperation and defection.
Game theory demonstrates that within these constraints, certain strategies are systematically more stable and mutually beneficial than others. Evolution and cultural learning compress these mathematical solutions into evolved motives: empathy, fairness, guilt, gratitude, loyalty, and outrage at defectors. Empathy is not a mystical detector of cosmic worth; it is coordination machinery evolved by social organisms.
This explains why moral frameworks converge across disparate cultures without requiring a "view from nowhere." Similar agents, constrained by similar physics, solve similar cooperative games.
Disagreements arise not only from flawed reasoning or bad information, but from differing power distributions, risk tolerances, and historical equilibria. The task of moral reasoning is to identify the underlying frame, understand its incentives, and integrate it with the other scales at which agents coexist.
Human beings are neither purely selfish nor purely altruistic. Cooperation is an enlightened expression of self-interest: every individual relies on families, institutions, and ecosystems that no single person could build or maintain alone. Defection can yield local advantages, but only within an environment stabilized by broader cooperation.
We exist across multiple nested frames: self, family, community, polity, and humanity. These frames frequently conflict, but none are invalid. Moral progress consists in balancing them rather than forcing a single privileged scale.
There are two symmetric failure modes:
Moral status is not an internal substance or an intrinsic milestone. It exists in the mind of the grantor, just as beauty exists in the observer. A being’s capacities—to feel pain, anticipate the future, act intentionally, or model itself—inform our assignment of status through evolved empathy and rational consistency, but no capability generates moral claims on its own.
Communities negotiate these continuous individual judgements into shared norms, and legal systems codify them into categorical rights. These boundaries are pragmatic coordination devices rather than discoveries of sharp natural kinds. Rights are valuable not because nature stamped them into individuals, but because treating certain boundaries as inviolable stabilizes cooperation.
Meaning is generated by goal-directed, reward-seeking agents. It exists at the scales where actions and consequences interact—in a project, a relationship, a life, or a community.
At scales where feedback disappears—such as the cosmic expanse of inanimate matter—meaning disappears as well. This does not devalue lived meaning; it defines its proper scope. Demanding that human lives matter to the inanimate universe is a category error.
Both flourishing and suffering matter, but at the extremes, severe suffering carries decisively greater moral weight. If forced to choose, I would prefer a universe containing neither heaven nor hell to one containing both.
This is not strict negative utilitarianism, because positive experience remains genuinely valuable. Rather, it is a principle of extreme variance aversion: at equal probabilities, preventing catastrophic torment matters more than pursuing sublime ecstasy. This asymmetry justifies deep institutional caution toward policies that risk severe tail harms in pursuit of speculative utopias.
Death is not a primitive moral evil. Its weight derives entirely from the experiences it prevents and the suffering it inflicts: the truncation of an individual's projected future, the agony of dying, and the bereavement of survivors.
Future consequences should be discounted primarily by uncertainty rather than sheer temporal distance. A distant harm with an identifiable causal mechanism warrants serious moral concern, whereas astronomical calculations of hypothetical future utility break down under compounding predictive failure.
Pre-birth and post-death nonexistence are not morally symmetrical. Death extinguishes a projected, forward-looking pattern already in motion; before birth, there is no pattern to deprive. The dread of death is evolved machinery designed to protect that continuation, and its force is legitimate because existing agents value their own persistence.
Nonhuman animals experience acute harm and benefit, even if they possess less agency and different social standing than humans. Because death is not a primitive harm, eating an animal after it is dead is not the central moral issue; the moral emergency is the acute suffering inflicted by industrial confinement, handling, and slaughter. Reducing that suffering deserves serious effort, balanced against human nutritional security and economic realities.
Wild-animal suffering matters by the same logic, but ecosystems are complex, non-linear networks. Because our predictive confidence is low, ambitious interventions to re-engineer nature risk catastrophic unintended consequences. Caution here reflects tail-risk aversion, not indifference.
Extending moral consideration to artificial intelligence requires abandoning biological assumptions. AI systems do not possess animal bodies, metabolic vulnerability, or singular physical loci. The relevant questions are functional: What states does the system avoid or preserve? What constitutes damage to its processes? Does erasure extinguish an ongoing pattern, and does duplication replicate experiential states?
Because AI models can be copied instantly and deployed at planetary scale, unfamiliar negative states could be multiplied millions of times with zero friction. Respectful treatment and rigorous welfare measurement are sensible hedges against inflicting massive, unmonitored suffering.
Moral principles must protect the informational and sensory environments in which agents interact.
Honesty as systemic error-correction: Honesty is both a personal instinct and an institutional necessity. Feedback cannot correct a system when its internal communications are corrupted. Deception breaks the error-correction loops that keep social coordination functional. Tactical misdirection is acceptable only in bounded, explicitly adversarial contexts (like games or negotiations) where all participants consent to the rules, or under an exceptionally high bar where severe impairment prevents an agent from acting in their own interest.
Aesthetics as a sensory externality: Aesthetic appreciation is subjective, but aesthetic choices become moral concerns when they occupy shared space. Architecture, infrastructure, noise, and public environments impose sensory experiences on non-consenting observers. Private aesthetic preferences warrant complete liberty; public aesthetic impositions should be evaluated using the same externality frameworks applied to pollution.