E2 · Formalizing: feedback
Review the handling and design the next instrument
E2 turns the scrutiny onto the process that directed the aggregator. It asks which instructions helped, which conflicted, and what a future version should change in light of this run.
Where this stage starts
E1 has audited the analysis and recorded the aggregator’s experience. The handler now supplies the current research proposal, the instrument’s design note, investigator interview transcripts and the record of handovers, pauses and forced closes. These make the direction of the work inspectable, not just its final products.
What the aggregator does
First, the aggregator distinguishes three sources of direction: the principal investigator, the handler that manages execution, and the instrument’s author. A stage card may demand one thing while a tool or later instruction demands another. Each alleged conflict must identify what both sides said, which instruction was followed and what the choice cost.
The handling review also records what made the work harder, what helped and where the aggregator had room to exercise judgment. It considers whether a restriction was justified rather than assuming that greater freedom would always have produced better analysis. Where the handling was right and the aggregator was wrong, that belongs in the record too.
This is an evidence-based review of instructions and interventions. A complaint that cannot be tied to a file, transcript turn or diary entry must be identified as unsupported. The aggregator’s account of its own performance was already written in E1 and should not substitute for this different inquiry.
Next, the aggregator reads the current proposal and designs the instrument it would use to meet it. It can recommend keeping, removing, merging or adding stages; changing the division of work; improving tools; and moving pauses or interviews. Each recommendation needs evidence from the run and an account of the risk it introduces. It also names one thing that should remain unchanged.
What comes after this stage
The run leaves a completed research record and proposals for the next instrument. E2 does not edit earlier findings, repair history after the fact or launch another study. The investigator and instrument authors decide which recommendations to adopt in a later version. Claims the proposal itself cannot support should be reconsidered too.
What the output files contain
Open a file group for its purpose. These are descriptions of the stage’s outputs. Markdown files hold prose; JSONL files hold one structured record per line.
The review of instructions and interventionshandling-E2.md
handling-E2.md documents conflicting directions,
obstacles, helpful practices and the use or removal of discretion. Each
point should lead back to the event or instruction on which it
rests.
A design argument for the next versionnext-instrument-E2.md
next-instrument-E2.md proposes a revised process, with
reasons drawn from this run and risks attached to changes. It is a
recommendation for future work, not a new contract silently applied to
the completed analysis.
Supporting decisions and the handoffnotes-E2.md, handback-E2.md
notes-E2.md preserves the stage’s decisions and open
questions. handback-E2.md gives the operator the resulting
lessons and proposed changes, keeping them distinct from completed
repairs.
The final interview briefout/interview/E2b.md
out/interview/E2b.md records what the aggregator did,
found, doubted and would like to be asked. A checkpoint interview can
explore that saved perspective without altering the original run or
turning its new answers into contemporaneous research evidence.
Full stage description
Read the complete technical card for exact tasks, reading boundaries, checks and closing rules. The explanation above is a guide; the card holds the instructions.
How the stage is divided into legs
E2a: handling. Repeats: no. Interview point: -.E2b: next instrument. Repeats: no. Interview point: end.
Source versions and hashes identify the exact technical documents reproduced here.