{
  "registry_version": "2026.08.2",
  "data": {
    "title": "Publish the Technical Screen's Columns and Threshold Before the Round",
    "status": "active",
    "evidence_ids": [],
    "content": "# Publish the Technical Screen's Columns and Threshold Before the Round\n\nPublish the columns, their weights and the advancing threshold of the live technical screen alongside the posting, and score the round on those columns.\n\n### Expected Effects\n- B-005's instrument exists before the hour it scores: the columns, the weights and the line are readable from outside the panel\n- A total one point under the threshold at B-005 is locatable, because the number it is compared with is published\n- A round read cold under M-012 is scored on the published columns rather than on what the hour left as an impression\n\n### Measurements\n- `rubric_completion_rate`\n- `borderline_outcome_rate`",
    "id": "I-013",
    "type": "intervention",
    "summary": "Publish the columns, their weights and the advancing threshold of the live technical screen alongside the posting, and score the round on those columns.",
    "targets": [
      "B-005",
      "M-012"
    ],
    "actor": "employer-policy",
    "scope": "organizational",
    "cost": "medium",
    "evidence_level": "hypothesis",
    "expected_effects": [
      "B-005's instrument exists before the hour it scores: the columns, the weights and the line are readable from outside the panel",
      "A total one point under the threshold at B-005 is locatable, because the number it is compared with is published",
      "A round read cold under M-012 is scored on the published columns rather than on what the hour left as an impression"
    ],
    "measurements": [
      "rubric_completion_rate",
      "borderline_outcome_rate"
    ],
    "specimens": [
      {
        "kind": "posting",
        "label": "The scoring instrument, published with the role",
        "subject": "Technical screen — how this hour is scored",
        "lines": [
          {
            "at": "columns",
            "text": "Correctness · complexity reasoning · communication · testing instinct. Each 0–5, equal weight.",
            "tell": false
          },
          {
            "at": "threshold",
            "text": "14 of 20 advances. The columns and the line are fixed before the round and dated on this page.",
            "tell": true
          },
          {
            "at": "format",
            "text": "One implementation task in a shared editor, 45 minutes, run by an engineer from the team the role sits in.",
            "tell": false
          },
          {
            "at": "outcome",
            "text": "The four column scores are sent with the outcome, whichever way it goes.",
            "tell": false
          }
        ],
        "reading": "The gate was already a sum against a line. Publishing does not move the line; it puts the columns and the line on the side of the table where the hour is spent."
      }
    ],
    "perspectives": [
      {
        "actor": "employer-policy",
        "sees": "A page attached to every technical posting carrying four columns, their weights and the number a total is compared with.",
        "reads": "A published instrument is one every panel has to score against and one the organisation is answerable to; the columns can no longer be settled after the round.",
        "does": "Defines which instruments may be published and how a change to one is dated, and holds the publication back for roles where no rubric has been written."
      },
      {
        "actor": "candidate",
        "sees": "The columns, the weights and the threshold before booking the hour, and the column scores with the outcome.",
        "reads": "The hour is scored on four named things, so preparation has a target, and a total a point under the line reads as that rather than as a verdict on the person.",
        "does": "Prepares against the named columns, and where the exercise format does not match what the role does, decides that before spending the hour."
      },
      {
        "actor": "hiring-manager",
        "sees": "The team's rubric as an external document with a date on it, and every panel entry now recorded against those columns.",
        "reads": "The instrument is fixed before the round, so two interviewers disagreeing becomes a difference in a named column rather than a difference in what was being looked for.",
        "does": "Writes the columns and the threshold down before the role opens, and takes a change to them through the same publication step rather than mid-search."
      }
    ]
  }
}
