Evidence appendix — Kestrel Energy

Grouped by evidence-confidence class, strongest first. Each class states the wording it licenses and the conclusion it does not, so a sentence quoted from this assessment can be checked against what produced it.

Sources are named, never quoted: script filenames, dictionary pairs and registry names. No IBM documentation text appears in this assessment or in the product that produced it.

A1 — IBM upgrade-script stated (2)

May be stated as: State it as what the release does, and name the script.

May not be concluded: Do not state it as an outcome without checking the target dictionary agrees. A script that drops an attribute and adds it back has redefined it, not removed it.

A2 — Bytecode/API confirmed (4)

May be stated as: State that the type or member is gone at the target release.

May not be concluded: Do not call a member removed when it is still reachable on a supertype. Do not count IBM's own *.unittest.* classes as API changes — that is packaging.

B1 — Dictionary observed and independently corroborated (2)

May be stated as: State it as a release change, naming both the dictionary difference and the corroborating script.

May not be concluded: Do not claim the script's specific values. Corroboration says the release touched the attribute; the values still come from the dictionary.

B2 — Dictionary observed only (14)

May be stated as: Say "between these two dictionaries", and say which two.

May not be concluded: Never state it as a release change, and never as something IBM did. A dump is an environment: the difference may be the installed industry solutions, not the release. Say so rather than hiding it.