Long-Jump and Distant Defence in Gomoku: From Factor Labels to Computable Definitions and a Blind-Spot Challenge Suite
No. 05 adjudicated more than twenty experience-based assertions one by one. Its conclusion was that the several assertions pointing to real phenomena failed not because the observations were wrong, but because core predicates such as “power value”, “cooperation” and “distant defence” were never defined, and therefore could not be confirmed or refuted by any test. This paper takes up the debt left by that verdict: for two of the most frequently invoked factors — long jump and distant defence — it supplies computable definitions, replaces the untestable predicate “power value” with several enumerable structural quantities, and derives three new falsifiable hypotheses (H4–H6). The paper also specifies the construction protocol of an “engine blind-spot challenge suite”: sampling rules, engine-configuration freezing, decision criteria, and reporting fields. The protocol itself is reproducible; its execution has not yet been carried out, and this paper reports no measured data. All claims remain hypotheses; none has passed verification.
Authors
- 湖南耒阳 曾乐
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24356838
- Primary Topic
- Artificial Intelligence in Games
- Type
- article
- Field-Weighted Citation Impact
- 0.00