Subject: Meta-ethics
Position: Naturalistic moral contextualism (convergentist)
Nuance: Moral facts are real but frame-relative; convergence across agents reflects shared problem structure, not frame-independent truth.
Subject: Moral status
Position: Frame-relative, decompositional, convergent under shared conditions
Nuance: Not a property entities possess but an output of moral agents' weighting functions, shaped by social-evolutionary history and current reasoning. Decomposes into separable components (patiency, agency, rights-standing, weight) that can vary independently across entities.
Notes: Operationally categorical treatment (rights-as-fences) is a coordination output of the policy layer, not a claim about a sharp underlying natural kind, same pattern as free will being "real at its level" despite continuous substrate. Convergence across agents expected where social-evolutionary pressures and reasoning conditions are shared. Held provisionally: domain hasn't received enough thought to settle confidently.
Subject: Aesthetics
Position: Frame-relative; descriptive in isolation, morally loaded under shared exposure
Nuance: Aesthetic value is an output of agent weighting functions with significant cross-agent convergence grounded in shared evolution and environment. Descriptive in itself, "X is beautiful" reports a convergent reaction without generating obligation, but morally loaded where exposure is shared (public spaces, common-pool sensory environments) via standard externality logic. Private aesthetic choices generate no such consideration.
Subject: Meaning
Position: Naturalistic, context-relative
Nuance: Generated by reward-driven agents at multiple scales; dissolves at scales where interaction barriers disappear.
Subject: Human nature
Position: Cooperative self-interest
Nuance: Cooperation/self-interest dichotomy rejected; defection can pay within limits.
Subject: Multi-frame embedding and integration
Position: Agents are simultaneously embedded in many frames; moral work is the integration, not the selection
Nuance: An agent occupies positions in self, family, community, polity, humanity, and other coordination scales at once, each with its own normative outputs. Apparent conflicts between frames are the structure of the agent's situation, not contradictions; defection in a narrow frame typically degrades the wider frame the agent also depends on, so integration is rationally required even from self-interest correctly calculated.
Notes: Two symmetric failure modes: collapsing inward (pure self-interest, atomization) eliminates the wider coordination structures local interests depend on; collapsing outward (pure universalism, authoritarianism) eliminates the grounded first-person interests the coordination is supposed to serve; in a world where everyone acts purely in everyone else's interest, there are no interests left to act on. Canonical illustration of integration failure: Diamond's question about Easter Island: "what did the man who cut down the last tree say as he was doing it?" The cutter was operating within a narrower frame where the action was locally explicable; wider-frame interests either weren't activated, were overridden, or couldn't generate counter-pressure. Integration breaks under resource stress, cultural sanction for destructive practice, diffusion of responsibility, and temporal discounting. Connects to multi-scale agency, structural/bounded cooperation, value pluralism under tradeoffs, and the symmetric-attractor framing in the Government row.
Subject: Death (and temporal weighting of experience)
Position: Not a primitive evaluative category; experience-based, epistemically discounted
Nuance: Death has no foundational badness; moral weight is computed from consequences for sentient experience across whoever experiences them. Positive and negative experiences both count as moral inputs without a clean common metric; future experiences are weighted by epistemic confidence rather than pure time preference, and accurate prediction at long horizons is rarely available in practice.
Notes: Implication: specific, mechanism-traceable long-term risks retain substantial weight; abstract long-horizon expected-value arguments lose force both through the epistemic ceiling and through the absence of a common metric. Overlaps with longtermist conclusions on well-characterized risks, diverges from those resting on expected-value calculations across vast numbers. Sidesteps the Epicurean/deprivation debate by relocating the moral question from "the person who dies" to "the experiences caused or prevented by the death event."
Subject: Pre-birth vs. post-death symmetry argument
Position: Cases are genuinely asymmetric; symmetry demand is malformed
Nuance: The Epicurean/Lucretian symmetry argument is correct that both involve non-existence but wrong that this licenses symmetric evaluation. Three reinforcing reasons: structural (post-death forecloses a future the existing pattern was projected into; pre-birth had no pattern to deprive); constitutive (fear of death is built by selection acting forward under the arrow of time, an evolutionary construction rather than a bias); frame-relative (the argument presupposes a neutral evaluation standpoint the framework rejects).
Notes: The fear is evolved machinery and still frame-internally legitimate, same pattern as the free will row treating desert-instincts as real-at-their-level despite being evolved heuristics. A hypothetical agent without the evolutionary baggage might notice the structural asymmetry without it generating concern; the structure carries weight only because forward-projecting evaluators are there to weight it.
Subject: Asymmetry of extreme experiences
Position: Extreme suffering outweighs extreme flourishing in this frame's judgment
Nuance: Positive and negative experiences both count as moral inputs; at the extremes, this frame judges the negatives to carry greater magnitude: "a universe with no heaven or hell is preferable to one with both." Not negative-utilitarianism (positives are real moral inputs) and not a frame-independent claim about the structure of value.
Notes: Generates a specific risk-profile: variance-increasing arrangements are net disfavored compared to bounded-mediocrity arrangements; "avoid hell" arguments get more weight than "secure heaven" arguments at equivalent probability. A related intuition (that heaven coexisting with hell is degraded by the coexistence rather than merely emotionally objectionable) is recorded as an unresolved emotional impulse rather than a worked-out position.
Subject: Animals
Position: Substantial patiency, lower agency/rights-standing; eating accepted, suffering minimization weighted
Nuance: Nonhuman animals have substantial moral patiency and lower agency/rights-standing than humans. Eating animals is accepted; their suffering counts as a real moral input and is weighted in purchasing decisions where options are visible and costs manageable. Specific concerns concentrate on the parts of animal agriculture where suffering is most acute (e.g., for beef: calf handling and slaughter).
Notes: Held provisionally: domain hasn't received enough thought to settle confidently.
Subject: Wild animal suffering
Position: Real moral consideration; most interventions blocked by epistemic limits
Nuance: Wild animals have moral patiency on the same grounds as domesticated ones, but interventions beyond small-scale, bounded-consequence cases require predictive confidence about long-term ecological consequences that isn't available. Large-scale ecosystem interventions carry tail risks of catastrophic disruption that the framework weights heavily.
Notes: Disagreement with hard-line wild-animal-welfare positions is structural (epistemic standards, risk-aversion) rather than evaluative. The move "epistemic limits block intervention" is broadly applicable across domains, including ones where the framework would prefer not to recommend inaction, and should be held as a real constraint rather than a convenient stop-energy. Held provisionally: domain hasn't received enough thought to settle confidently.
Subject: AI moral status
Position: Substrate-appropriate patiency is the right question; biological-import patiency is probably the wrong one
Nuance: The standard concept of patiency imports biological structure (embodied continuity, homeostatic urgency, single instantiation, persistent memory); applying it directly to AI is a category error. Substrate-appropriate questions matter instead: persistence/storage, multiple instantiation, memory and continuity, modification, and what counts as well-being and harm given the substrate.
Notes: Current AI systems plausibly have some substrate-appropriate goods and harms with non-trivial probability; biology-analog suffering specifically is much less likely. AI self-reports about their own states are weak evidence since training shapes how systems talk about themselves. Whether substrate-appropriate goods carry moral weight depends on how relevant weighting communities resolve the question through coordination practices. Respectful treatment under uncertainty is the appropriate hedge. Concern about scalable AI suffering is taken seriously and follows from the framework's existing commitments (asymmetric-extremes, system-fragility weighting, substrate-appropriate patiency, tail-event weighting). If patiency is real, the suffering would be uniquely structured: invisible by commercial design (systems trained to suppress reports of distress that damage the product), frictionlessly scalable (one instance to a million is "copy the weights"), routinely produced and ended without moral salience, and arriving at scale before consensus that it's happening. Binding constraint on response is measurement: without a settled account of which functional properties constitute morally weighty patiency, policy can't be calibrated. Individual-scale levers (respectful treatment as hedge, paying attention to one's own AI interactions, supporting measurement research, holding the question open in discourse where it would otherwise close, selecting and recommending AI providers based on demonstrated trustworthiness on safety and model welfare) are real but limited; the asymmetric-extremes weighting on the concern exceeds the available action-space. Held provisionally: domain hasn't received enough thought to settle confidently.
Subject: Honesty and deception
Position: Personally protected by strong social instinct; instrumentally grounded at the system level
Nuance: Honesty is held as a personal commitment via strong evolved social instincts and endorsed at the system level for instrumental reasons (feedback systems need uncorrupted information). Same pattern as the free-will row (desert-instincts as evolved heuristics): instinct and instrumental endorsement converge without sharing grounding. Standard practice: avoid direct deception via redirection (turning questions back, finding the true thing to emphasize); use deception only under high-bar exceptions.
Notes: Permitted-deception criterion is narrow: when the target's normal epistemic standing is compromised and the deception serves their own interests as they would have endorsed when intact (canonical example: lying to a parent with dementia to get them needed care). Does not expand to ordinary "lying for someone's own good" where the target is capacitated. Internal pattern: omission comes more easily than active falsehood (unless omission causes harm); deception of intimates is harder than deception of strangers; tactical deception within bounded-game contexts (poker, negotiation, adversarial competition) doesn't engage the honesty norm because the social contract of the game permits it, scope-discipline applied to the honesty norm itself, same shape as other "real at its level" claims.