MAYAN_ALFA P0.7 — Public Compare Layer Note
===========================================

This folder contains the public-facing compare layer generated or collected by
01F_RUN_COMPARE_ALL.sh.

The purpose of this layer is to provide release evidence that MAYAN_ALFA
benchmark outputs were reviewed against independent reference paths such as
Miller-Rabin and primesieve where available.

How to read this layer
----------------------

The compare layer should be interpreted as benchmark validation evidence within
a specific documented workflow. It supports a careful statement such as:

  "The release includes structured comparison artifacts between MAYAN_ALFA and
  independent reference paths, supporting the reproducibility and auditability of
  the reported benchmark outputs."

It should not be interpreted as:

  - an absolute mathematical proof,
  - a cryptographic security claim,
  - a universal performance guarantee,
  - an unsupported extrapolation beyond the benchmark scope.

Public release boundary
-----------------------

Public GitHub and Zenodo packages may include public-safe comparison summaries,
QA notes and benchmark artifacts within the approved public boundary.

Protected MAYAN_ALFA core logic, internal heuristics, private Observation
Registry material and uncontrolled [CONTROLLED_SCALE_REDACTED] controlled layer data must remain outside public,
BRONZE and SILVER packages.

Product boundary
----------------

Public:
  Public-safe comparison summaries and benchmark evidence.

BRONZE:
  Research toolkit access without protected core source and without real [CONTROLLED_SCALE_REDACTED]
  controlled layer data.

SILVER:
  Publication archive access with paper/review structure, still without
  protected core source and without real [CONTROLLED_SCALE_REDACTED] controlled layer data.

[CONTROLLED_ARCHIVE_REDACTED]:
  The only layer intended to contain the real [CONTROLLED_SCALE_REDACTED] validation archive.

Publication language
--------------------

Recommended language:
  observations, measurements, validation evidence, benchmark comparison,
  scaling characteristics, documented workflow context.

Avoid:
  sensational claims, unsupported superiority claims, absolute mathematical
  declarations and statements that exceed the validated release context.
