{
  "abstract": "Part 5 of a series. Part 1: 10.5281/zenodo.23587879 · Part 2: 10.5281/zenodo.23587670 · Part 3: 10.5281/zenodo.23526074 · Part 4: 10.5281/zenodo.23609857 This article states how the predictions, retractions, theory and principle of the preceding four articles were obtained. The setting: too many variables to measure, self-report as the least reliable available channel, no possibility of control groups (clients arrive one at a time), and no external reward. Without groups, blinding or instruments, one route remains: comparison within the same body at the same time, using the body's left–right pairing as the control. Six procedures are described, each with observations, derivation, literature check, verdict and stated weaknesses: - Contralateral self-comparison: treat one side, have the client compare. The between-person term drops out of Δ = τ + (u_A − u_B) + ε. The client is the comparator, reading internal states invisible from outside as a left–right difference; comparing is both measuring and intervening. - Stopping rule: fix the end state before treating, because one-sided improvement creates a new asymmetry and a one-sided learning gradient. - Forward-only search: abandon the ill-posed inverse problem, apply interventions in parallel, identify causes afterwards from what worked (a generalization of diagnosis ex juvantibus). - Compensation-triggered debugging: use the single bit \"did compensation appear?\" instead of measuring every joint; consistent with the minimal intervention principle. - Deception-cost ordering: rank channels by the cost of faking them; autonomic outputs can only be faked by changing the internal state itself. Bandwidth and reliability run in opposite directions. - Author-as-subject: loads not imposed on clients are measured on the author's own body, yielding hypotheses only. AI is used as a verification device (dimensional analysis, demands for refutation, literature checks), and its errors are recorded with equal weight: six fabricated verbatim quotations with plausible file numbers and timestamps were removed by a second-pass check, and the AI's own derivations erred twice during Part 4 before numerical checks caught them. Weaknesses stated openly. Contralateral self-comparison has not been blinded, so expectation effects are not excluded. Noticed while writing this article and not stated in the book: its formula leaves the intrinsic left–right term (s_R − s_L), so post-treatment comparison alone confounds treatment with pre-existing asymmetry. A difference-in-differences estimate, τ̂ = (R_post − L_post) − (R_pre − L_pre), removes it and is practicable in the field. Six falsifiable predictions about the procedures themselves are registered (numbered 17–22, continuing the series): - P17. Under blinding with a sham on the untreated side, clients still identify the treated side above 50%. If not, the recorded comparisons likely reflect expectation. The most consequential prediction in the article. - P18. Left–right footfall sound asymmetry, recorded with a phone, shrinks after one-sided treatment, including in blinded sessions. - P19. Self-comparison detects smaller treatment doses than a blinded external observer. - P20. Treating only the less-loaded side without the stopping rule reverses the loading asymmetry at the next visit, rather than letting it decay to baseline. - P21. Pre-emptive correction of joints showing no compensation yields less improvement or new compensation than correcting only upstream of observed compensation. - P22. When verbal report and autonomic channels disagree, the subsequent course follows the autonomic channel. Refutations are welcome, especially on P17. Experts in single-case experimental designs and N-of-1 trials are invited to point out problems already solved in their field. The author has no formal research training. Corrections will be issued as new versions, with earlier versions kept retrievable. The article body is in Japanese. An English summary, the registered predictions and the reference list are included in the document. Contact: contact@medi-gym-revital.com",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": null,
  "author_total": 1,
  "authors": [
    {
      "name": "CHIAKI KURIYAMA",
      "orcid": "0009-0005-2636-0164"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/dt6wx7y4n6qz8.bib",
  "cited_by": 13,
  "doi": "10.5281/zenodo.23610606",
  "fulltext_url": null,
  "id": "dt6wx7y4n6qz8",
  "issns": [],
  "landing_url": "https://doi.org/10.5281/zenodo.23610606",
  "language": "en",
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "2613|Advanced Causal Inference Techniques"
    ]
  ],
  "openalex": "W7223473714",
  "pdf_url": null,
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-24",
  "publisher": "European Organization for Nuclear Research",
  "ris": "https://research.draconitic.com/paper/dt6wx7y4n6qz8.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:17Z",
      "native_id": "W7223473714",
      "source": "openalex",
      "url": "https://doi.org/10.5281/zenodo.23610606"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "openalex:2613|Advanced Causal Inference Techniques",
      "subfield": "statistics"
    }
  ],
  "title": "Causal Inference Without Control Groups: Six Field Procedures of the Kuriyama Protocol",
  "title_text": "Causal Inference Without Control Groups: Six Field Procedures of the Kuriyama Protocol",
  "type": "preprint",
  "updated_at": "2026-11-09T13:21:00Z",
  "url": "https://research.draconitic.com/paper/dt6wx7y4n6qz8",
  "venue": "Zenodo (CERN European Organization for Nuclear Research)",
  "versions": []
}