Pythology Research Architecture
Pythology researches how intelligence can move from raw evidence to mechanisms, forecasts, alternative decisions, human challenge and learning without hiding uncertainty behind a convincing answer.
Domain research
A shared architecture does not mean shared certainty. Every domain requires its own ontology, evidence standards, mechanisms, validation and calibration history.
Planetary Intelligence
EarthNet exposes the architecture to live, noisy, independently observable planetary events. It is where causal reasoning, Prometheus and Decision Futures can be tested prospectively rather than merely demonstrated retrospectively.
Environmental hazards, infrastructure pressure, satellite evidence and wider operating context in one evidence state.
Explore EarthNet →Commit forecasts before outcomes are known, then score calibration when later evidence resolves the question.
Explore Prometheus →Compare a baseline with a small number of pre-registered human decisions when the consequences are measurable and genuinely matter.
Explore Decision Futures →Interrogate system health, evidence, reasoning, uncertainty and branch differences through a controlled human command interface.
Biological Intelligence
Biological Intelligence is one connected research architecture operating at different scales: evidence → biological state → mechanism → molecular interrogation → experiment → outcome → learning.
Reviewed claims, provenance, contradictions, precursor states and evidence eligibility.
Explore Bio-Symbology →Genomic, transcriptomic, epigenomic, proteomic, metabolic, cellular, spatial and temporal biological state.
Explore Omni Genomics →Targets, variants, structures, pathways and evidence-visible Mechanism of Action interrogation.
Explore Proteus →Turn unresolved causal questions into experiments and feed supplied outcomes back into the evidence loop.
Explore the Bioreactor →Environmental and population-scale biological surveillance under explicit evidence and governance boundaries.
Explore Bio Mesh →Compare testable perturbation or intervention hypotheses while preserving the distinction between observed experiments and alternatives that were never performed.
Target ranking and MoA auditing based on supported mechanisms, contradictions and uncertainty rather than association alone.
Research into restoration of function, with the eye as the first proving problem and structural, functional, durability and safety evidence kept separate.
Defensive detection, mechanism understanding, countermeasure evidence and public-health resilience without contributing to biological harm.
Trace variant → molecular consequence → biological state → phenotype while showing exactly where evidence weakens.
Investigate compensatory pathways, off-target effects and mechanisms that could undermine an intervention.
Freeze evidence before known outcomes, commit a judgement, then compare with what experiments and clinical programmes later revealed.
Physical Intelligence
Physical Intelligence explores world models for engineered systems where geometry, materials, forces and operating conditions impose hard constraints on what an intelligent system may responsibly infer.
Combine observations and simulation without allowing a learned model to override known physical limits without evidence.
A specialist simulation component for energy, transport, ports, logistics and other engineered dependencies when Decision Futures branches require physical constraints.
Explore Technosphere →Compare maintenance, loading, routing, operating or design choices against a frozen physical state, then validate the enacted branch with tests or field evidence.
Explore Physical Intelligence →Research horizons
Pythology maintains controlled long-horizon work where the architecture can be tested without turning speculative possibilities into public claims.
Observable agent behaviour, adversarial evaluation, permissions, recovery and human accountability.
Explore programme →Explore early signals and difficult research questions while keeping speculation, evidence and validation status visibly separate.
Explore horizons →Compare how different choices could alter downstream outcomes without confusing an unchosen scenario with observed evidence.
Explore Decision Futures →Research discipline
The architecture is useful only if it can expose where it is weak. A persuasive explanation is not evidence; a probability is not causation; and a model is not validated because it agrees with itself.
Preserve source, timestamp, context and limits.
Keep plausible alternative explanations alive long enough to test them.
Freeze forecasts and decision branches before the outcome is known.
Define what evidence would weaken or overturn the belief.
Use resolved outcomes to update future confidence without rewriting the original record.
.jpeg)