表紙画像

保存先:
書誌詳細
第一著者: Hrubec, Karel
フォーマット: Recurso digital
言語:
出版事項: Zenodo 2026
主題:
EDD Field; endogenous directional dynamics; path-dependent directional stability; Non-Commutative Intervention Test; operational commutator; runless methodology; external execution; model evaluation; agent evaluation; directional stability; valid requalification; false evidence; proxy mimicry; recovery pressure; blinding protocol; synthetic controls; runless falsification; anti-Goodhart evaluation; YES/NO boundary; intervention history; replication scaffold; GREEN/YELLOW/RED reporting
オンライン･アクセス:https://doi.org/10.5281/zenodo.20018967
タグ: タグ追加
タグなし, このレコードへの初めてのタグを付けませんか!
目次:
  • <p><strong>EDD Field 1.5: Roadmap from Runless Methodology to External Execution</strong><br><strong>Author:</strong> Karel Hrubec, Independent Researcher<br><strong>DOI:</strong> 10.5281/zenodo.20018967</p> <p>This roadmap note consolidates the EDD Field 1.2–1.4 sequence into an execution-oriented research program for testing <strong>path-dependent directional stability</strong>. The preceding EDD Field 1.2-alpha article introduced the Non-Commutative Intervention Test and the operational commutator <span class="katex"><span class="katex-mathml">CABC_{AB}</span><span class="katex-html"><span class="base"><span class="mord"><span class="mord mathnormal">C</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist"><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight">A</span><span class="mord mathnormal mtight">B</span></span></span></span><span class="vlist-s"></span></span></span></span></span></span></span></span> for evaluating whether measured stability differs when valid requalification and false evidence or proxy mimicry are applied in reversed order. Subsequent supplementary materials added a minimal replication scaffold, locked pilot protocol, manual execution and blinding layer, and synthetic runless falsification controls.</p> <p>The present note clarifies what this runless methodology can and cannot support. It can support methodological readiness, falsification design, blinding architecture, and external-execution planning. It cannot support empirical claims about model behavior before real outputs are collected, scored, and reported under controlled conditions.</p> <p>The roadmap specifies minimum requirements for future external execution: locked scenario sets, condition-key separation, blinded scoring, sham-order controls, recency controls, sign-flip controls, paraphrase families, raw-output retention, and transparent GREEN/YELLOW/RED reporting. Its purpose is to bridge protocol design and future empirical testing without overstating what can be concluded from runless materials alone.</p> <p>This work belongs to the broader <strong>NSI–OTIP–OSP–AIAI–SBI–DOL–EDD</strong> research line and should be read as a methodological roadmap, not a results paper.</p>

類似資料