Independent research manifesto · 2026
We will not make a patient’s experience smaller than the diagnosis.
Diagnosis is essential for finding risk and choosing treatment, but it does not always describe every change a patient experiences, every loss of function, or every response to care. We will not rush to close that difference with another label. While continuing to guard against missed disease, we will observe more carefully, ask where historical medical records and contemporary research genuinely meet, and formulate questions that can be examined in the next consultation.
While there is value in the items on the right, we value the items on the left more.
problems worth solving over over papers worth counting
new distinctions and discoveries over over fluent synthesis
traceability to editions and passages over over inherited authority
connections that preserve difference over over premature equivalence
effect, uncertainty, patient importance, and local reality over over verdicts by p-value alone
counterexamples, failure, and null results over over confirmation theatre
publish, review, and revise in public over over disclosure only after permission
a shared graph of problems, hypotheses, methods, evidence, and critique over over the monopoly of finished papers
explicit boundaries of who, when, and how far over over universal extrapolation
AI-augmented, human-accountable inquiry over over autonomous publication
Our commitments
What we commit to
- Form bold hypotheses while keeping observation, source text, inference, and efficacy claims distinct.
- Do not harden what diagnosis leaves under-described into a new diagnosis; keep diagnostic openness and symptom-focused therapeutic research active together.
- Connect texts, commentaries, cases, medical history, Chinese medicine, Kampo, and modern research as mutually critical evidence planes—not one hierarchy.
- Attach sources, passages, relations, uncertainty, and disconfirmation conditions to every consequential claim.
- Do not count expert knowledge paraphrased by a model as a research contribution.
- Require major studies to state both what local practice can newly observe and what it must not yet change.
- Publish adverse evidence and failed connections as starting points for the next investigator.
- Rapidly publish problems, hypotheses, methods, data, analyses, and critiques as independently citable objects, then mature them through versions and public assessment.
- Open our writing, metadata, and code for downstream inquiry, and connect others’ revisions and refutations to the same research record.
- Place patient and clinician safety, consent, and privacy ahead of discovery speed.
- Never overstate journal identity, DOI, review, indexing, or AI use beyond its actual state.