{
  "registry_version": "1.1.0",
  "data": {
    "title": "Critique injected on script in the collaborative exercise",
    "status": "active",
    "evidence_ids": [
      "evidence.employment_interview_reliability_new_meta_analytic_estimates_by_structure_and_format",
      "evidence.structured_selection_procedures_and_evaluator_preparation_29_c_f_r_1607_4_eu_ai_act_annex_iii"
    ],
    "content": "# Critique injected on script in the collaborative exercise\n\nThe collaborative round carries a written script: the same objection is raised at the same point for every candidate, and the panel records what was said in response rather than how the room felt.\n\n### Expected Effects\n- mech.communication_or_working_style_friction is measured against a common prompt rather than inferred from an interviewer's chosen one\n- Interviewer variance stops entering the verdict as if it were candidate behaviour\n\n### Measurements\n- `scripted_prompt_coverage`\n- `panel_inter_rater_agreement`",
    "aliases": [],
    "id": "int.critique_injected_on_script_in_the_collaborative_exercise",
    "type": "intervention",
    "summary": "The collaborative round carries a written script: the same objection is raised at the same point for every candidate, and the panel records what was said in response rather than how the room felt.",
    "targets": [
      "mech.communication_or_working_style_friction",
      "bar.recruiter_screening_call",
      "bar.team_cross_functional_panel"
    ],
    "actor": "hiring-manager",
    "scope": "team",
    "cost": "low",
    "evidence_level": "supported",
    "expected_effects": [
      "mech.communication_or_working_style_friction is measured against a common prompt rather than inferred from an interviewer's chosen one",
      "Interviewer variance stops entering the verdict as if it were candidate behaviour"
    ],
    "measurements": [
      "scripted_prompt_coverage",
      "panel_inter_rater_agreement"
    ],
    "specimens": [
      {
        "kind": "transcript",
        "label": "The script, and the line it produces",
        "lines": [
          {
            "at": "00:18",
            "text": "Scripted at minute 18, verbatim for every candidate: \"I think this design is wrong — you have put the queue in the wrong place.\"",
            "tell": true
          },
          {
            "at": "00:19",
            "text": "Recorded: asked which failure the objection was about; restated the trade-off; proposed moving it and named what that would cost.",
            "tell": false
          },
          {
            "at": "00:31",
            "text": "Not recorded: tone, warmth, whether the room enjoyed it. The rubric has no column for it, so no column carries it.",
            "tell": false
          }
        ],
        "reading": "Friction in a collaborative exercise is a real signal, and this does not deny it — it makes it comparable. Unscripted, the objection arrives when an interviewer happens to feel like raising it, in words chosen for that candidate, and what comes back is scored against a memory of how other rounds went. Scripted, the same prompt produces answers that can be set side by side."
      }
    ],
    "perspectives": [
      {
        "actor": "actor.hiring_manager",
        "sees": "A one-page script with the objection, the minute it lands, and the three things the answer is scored on.",
        "reads": "What the panel disagreed about was never the candidate — it was which objection each interviewer raised and how hard. Fixing the prompt leaves only the answer to disagree about.",
        "does": "Writes the script once per role, and holds the panel to scoring the response rather than the room."
      },
      {
        "actor": "actor.candidate",
        "sees": "An objection that arrives as part of the exercise, and a rubric that names what the answer is being read for.",
        "reads": "The pushback is the test, not a reaction to something already gone wrong — so it can be answered on the merits instead of managed.",
        "does": "Answers the objection rather than reading the room for what caused it."
      },
      {
        "actor": "actor.recruiter",
        "sees": "Panel write-ups that name a scripted moment and quote the response, instead of arriving as impressions.",
        "reads": "A write-up quoting a common prompt is the only kind that can be compared between candidates or defended after the fact.",
        "does": "Returns the quoted exchange as the feedback, and refuses write-ups that score the room."
      }
    ]
  }
}