Research and engineering for judgment, verifiable outcomes, and accountable collaboration across humans, organizations, and AI systems.

We study how limited observation supports conclusions, how intended changes become explicit completion criteria, and how actions and outcomes can be examined across trust boundaries.

We investigate target-relative determinability, causal observability, and the structures needed for consequential action across systems.

Accountable Reality Transition (ART) explores these questions through explicit hypotheses, counterexamples, and adversarial tests.

Target State Transition Object (TSTO) defines an immutable object for expressing what state change counts as completion.

The repository also contains JEP–TSTO bindings connecting completion criteria with judgment-related event records.

We develop software for outcome specification, evidence review, and completion records, alongside experiments that make research claims inspectable.

Public demonstrations include:

• Causal Observability Demo — exploring whether available observations determine a target fact within a finite model.
• COE Core Demo — an exploratory simulation of shared-state formation among agents.

• Explicit boundaries — Keep observations, claims, evidence, verification, and acceptance decisions distinguishable.
• Focused specifications — Place domain-specific policy and workflow in profiles and applications.
• Inspectable results — Make assumptions, examples, counterexamples, and validation methods visible.
• Clear maturity — Identify research hypotheses, experimental drafts, and implementations by their actual status.
• Revisability — Let evidence, failed tests, and implementation experience drive corrections.

Licensing and reuse terms are specified in each repository.

Judgment Event Protocol (JEP) provides a related event-recording layer for Judgment, Delegation, Termination, and Verification.

Research frameworks, protocol specifications, and applications retain their own scope and validation requirements.

• Email: yuqiang@humanjudgment.org
• Research questions, counterexamples, and implementation feedback: use Issues in the relevant repository.