Context Is Not a Virtue
Teams talk about context as if more is always better: onboard people properly, share the background, keep everyone informed. For most tasks that is right. For a specific and important minority it is exactly backwards, and treating context as a virtue is what stops anyone noticing which tasks those are.
The test is simple. Ask what the task is for. If it is producing something, or explaining something already produced, context helps — the more of the relevant background is present, the better the answer. If the task is judging whether the output stands on its own, context is the thing that guarantees a wrong answer, because you cannot un-know the intent while looking at the words meant to convey it.
The Two Placements
This workshop makes exactly two placements, in opposite directions, and they are the whole design.
Continuity inside a stage. The session that produced the material keeps its session for the immediate iteration. That is when a question is cheapest to answer: the reasoning is still loaded, the alternatives that were rejected are still remembered, and the reason a section is short is still available. A fresh session would have to reconstruct all of it, badly.
Boundary between author and reviewer. The review always crosses an instance boundary — a different context at minimum, a different system preferred. The reviewer reads the artifact cold from disk, never through the author's summary, because a summary is the author's model of their own work and the whole question is whether that model matches the artifact.
The Rule You Can Carry
Every step in a process is either producing, explaining, or judging. Producing and explaining want maximum continuity. Judging wants a boundary, and the strength of the boundary should scale with how much the judgment depends on the artifact standing alone.
Where teams go wrong is in treating the boundary as an insult. It is not a statement about the author's honesty or skill; it is a statement about what a mind holding an intention can see when it looks at the expression of that intention. The author is the worst possible judge of that specific question, and it has nothing to do with how good they are.
The converse holds too, and it is the one people get wrong more often. Put a cold reviewer in the seat that asks why the artifact is shaped the way it is and they have to guess — and they will produce a plausible guess. A boundary on an explaining step is not rigor; it is waste.