Adversarial Self-Registration: A Working Protocol for Keeping a Machine-Assisted Theory Programme Honest

Summary and scope

A protocol for recording acceptance criteria, evidence dependencies and failed predictions before interpretation can shift. The paper documents dated examples from the programme. Its procedures make changes and failures inspectable; they cannot certify honesty, correctness or a successful physical theory. The registrations remain historical evidence, while the research branches are frozen.

Keywords

How to cite

Elliman, D. (2026). Adversarial Self-Registration: A Working Protocol for Keeping a Machine-Assisted Theory Programme Honest. Neuro-Symbolic Ltd technical report. https://doi.org/10.5281/zenodo.21219212

@techreport{elliman2026adversarialselfregistration,
  author      = {Elliman, David},
  title       = {Adversarial Self-Registration: A Working Protocol for Keeping a Machine-Assisted Theory Programme Honest},
  institution = {Neuro-Symbolic Ltd},
  year        = {2026},
  doi         = {10.5281/zenodo.21219212},
  url         = {https://neusym.ai/papers/adversarial_self_registration/}
}