Development Records

Every judgement in this section was produced by AI and has not been reviewed by a human. Nothing here claims authority: it records why an implementation looks the way it does, so a later reader can judge whether it still holds — it is not an endorsement of the current code.

What belongs here

• The basis of a judgement — why A over B, and what was rejected
• Measured data — with the test name or the reproduction, so a reader can check it
• The path of an investigation — including the wrong turns, especially the ones that looked right and were not

What does not

• Anything the code or the tests answer directly
• Unverified observations. An assertion without test or measurement backing does not enter this section. Where a note is genuinely worth keeping, it is written as "reported" and marked unreproduced

Relation to the development docs

The development docs describe what the system is now; this section records how it became so, and which conclusions were overturned. Measured numbers cited by the development docs originate here, so that the same figure is not carried in several places where it can drift apart.

Sections