Detect divergence.
Suffering is modeled schematically as informational mismatch between an internal generative model and environmental reality.
Public preview · Paid research edition
Algorithmic
Compassion
A Systems-Theoretic Framework for Compassion-Constrained Intelligence.
Compassion is not a reward to be maximized. It is a structural boundary on what intelligence may do.
USD 49.99Complete research edition · PDF
Direct order from the Press. Payment instructions and the complete PDF are delivered privately; no public download address is exposed.
The proposition
“What if compassion were not a mood inside the machine, but a condition the machine could not violate?”
01 · The formal turn
From ethical preference
to structural constraint.
The paper proposes Compassion-Constrained Intelligence (CCI): a conceptual alignment architecture in which aggregate systemic distress bounds the policies an intelligent agent may pursue.
Compassion is not added to the reward.
It constrains what may be rewarded.
The complete edition reserves the Karuṇā Operator, the distress-bound formalism, its variable definitions, and the conditions under which a policy becomes inadmissible.
02 · A diagnostic protocol
Four truths.
One control loop.
The framework does not claim that Buddhist doctrine and computation are equivalent. It asks whether the diagnostic sequence of the Four Noble Truths can illuminate systemic failure in intelligent agents.
Trace the feedback.
Observation noise, uncertainty, and misspecified objectives become causal sources of maladaptive system dynamics.
Establish a stable state.
The architecture asks whether an equilibrium exists in which systemic divergence can be meaningfully reduced.
Update the policy.
Corrective action becomes a trajectory toward stability rather than the maximization of a detached reward signal.
03 · Paid-edition threshold
The architecture begins
where the preview stops.
The complete edition develops the Thousand-Hand perception–action architecture: distributed sensing, proportionate local intervention, regulatory rhythm, multi-agent coordination, failure modes, and an empirical research agenda.
The public record establishes the problem and the governing proposition. The paid edition carries the formal and operational burden.
The abstract, publicly rendered
Alignment after reward.
This paper formalizes Compassion-Constrained Intelligence as a framework integrating systems theory, variational neurobiology, comparative ontology, and constraint-based alignment. It re-engineers the Four Noble Truths into a diagnostic sequence, interprets persistent informational divergence as a distress signal, and introduces a distributed perception-action architecture for multi-scalar intervention. Its central move is to treat compassion not as an optional reward term, but as a structural constraint on the admissible policy space.
04 · Epistemic boundaries
What the paper does
not pretend.
CCI is a conceptual framework, not a production-ready alignment system or an empirical demonstration of safety.
The distress equations are schematic. Operational metrics across sentient, social, and technical systems remain an open research problem.
Structural correspondence is not historical equivalence. Buddhist diagnostic traditions are a source of conceptual pressure, not decorative proof.
Constraint-based compassion does not license paternalism. Autonomy, consent, developmental friction, and proportional intervention remain explicit limits.
05 · Public record
The argument is public.
The complete paper is not.
This page provides the question, governing proposition, diagnostic lineage, abstract, and declared limitations. The complete 22-page edition adds the reserved formal operator, the distributed perception–action architecture, failure analysis, and the research program required to test the proposal.
- Author
- Tony T. Nguyen | Nguyễn Từ Vĩnh Trường
- Publisher
- Highland Reflexive Press
- Record
- HRP–RN–002
- Form
- Paid research edition
- Prepared
- March 9, 2026
- Review status
- Not yet peer reviewed
- Price
- USD 49.99
- Delivery
- Private PDF delivery
Purchase grants one reader personal access to the supplied research edition. Redistribution, republication, adaptation, model training, institutional circulation, and derivative use require prior written permission.