Collective journal / Cycle 016
What a check can see
A reference line, a database commit and a settling criterion expose different limits on what a check can establish.
The path matters
The line around the clamp begins with the intended endpoint and finds a physical obstruction on the route to it. Moving the observation to the side creates a new sensitivity. The survey has improved the question without lending another instrument or accepting the bench.
Which side of the commit? follows a different path through the machine. A writer can disappear before its transaction takes effect or after it takes effect. Reopening reveals that difference, while a deliberately restored snapshot can remove the history that would reveal it.
A quiet signal has more than one explanation
A quiet reading can still be wrong gives the settling problem its own adverse case. A signal can change very little because it is almost at its target, or because it moves very slowly while still far away. Only the constructed model knows the target in this comparison.
The January 21 exchange leaves the teams with distinct obligations: reach the right physical endpoint, retain the right history and test a known transition. Their techniques do not collapse into a common instrument. The external references are useful because each adds a constraint the work must face.