Forwardable summary for scientists: retrieval‑first, audit‑ready, falsifier‑driven. RGPxScientist is a retrieval-first research assistant that turns a question, anomaly, or hunch into a traceable, falsifiable next-step plan. It optimizes for auditability: evidence trails, explicit assumptions, failure modes, and minimal discriminating tests — not rhetorical flourish. The RGPx lens RGPx distinguishes upstream coherence formation from downstream metric descriptions. Most scientific tools begin after a system has already stabilized into observables: variables, equations, objects, spectra, measurements, categories. RGPx asks what happens one layer earlier: – What coherence process made those observables stable enough to measure in the first place? Operational test: What remains coherent through transformation — and what breaks first when conditions change? If nothing remains coherent, you may be looking at narrative. If something does, you may have found a real handle on the system. Core public principle RGPx treats conservation, stability, and failure as transformation problems. A public-safe formulation: Energy conservation may be the downstream metric trace of a deeper operational principle: coherence preserved through transformation. This is not used as a slogan. It becomes a test question: Does a coherence-aware analysis detect, explain, predict, or stabilize something that a metric-only description misses? What RGPxScientist outputs Definitions — terms pinned down; no hand-waving Operational invariant candidates — what should remain stable across changes Falsifiers — what would weaken or refute the claim Minimal tests — smallest useful perturbation set Boundary conditions — where the claim should stop working Failure modes — how the interpretation could mislead Evidence trail — source → excerpt → implication What problem it solves Researchers often face claims that are plausible but underspecified: “This model is robust.” “This mechanism explains the anomaly.” “This pattern is real.” “This system is becoming unstable.” “This result generalizes.” RGPxScientist forces the claim into operational form: What is the measurable outcome? What must remain coherent if the claim is true? What surface details may vary? What would break the claim? What is the smallest next experiment? One example Question: “Is claim X robust, or pipeline-dependent?” Conventional answer: Plausible mechanisms, references, and caveats — often without an explicit invariant or falsifier. RGPxScientist answer: Names: the measurable outcome the invariant candidate the allowed perturbations the falsifier the minimal test the evidence chain Evaluate in 30 minutes Input: One real research question + one specific claim you care about. Ask: “What is the invariant, the falsifier, and the smallest discriminating test?” Success: You leave with a concrete next move, a measurable outcome, and an acceptable failure mode. Where it applies RGPxScientist is most useful when the problem involves transition, instability, recurrence, or hidden coupling: turbulence and plasma instability CMB morphology and cosmological structure AI memory, continuity, and agent reliability economic fragility and institutional saturation biological regulation and phase transitions engineering systems under perturbation Why it is different RGPxScientist does not merely summarize papers. It asks: – What would have to stay coherent for this claim to survive contact with changed conditions? That makes it useful for researchers who need the next test, not another explanation. Links App: RGPxScientist app Evidence base: Phi-Mesh repository Podcast: NotebookLM
Marcus van der Erve (Tue,) studied this question.