Signal-vs-Noise Drill

Module 6, Lesson 6.4 — rate 12 candidate watchers before you ever write the prompt.
The response test. A watcher is only worth building if you can name the specific action you will take within the hour when it fires. “I’ll see what’s up” is not a response.
Audience = only you. Every watcher in this course writes to a destination where you are the sole reader — file, private log, your own inbox.

How the drill works

  1. Twelve candidate watchers follow. Each is a one-paragraph idea the way a student would actually pitch it — the event to watch, the proposed response, and the rough qualifier.
  2. Rate each candidate using the three-way rubric:
    Signal Passes the response test; condition-and-threshold is specific; the watcher is worth building roughly as stated. Maybe tighten wording, but no redesign needed. Adjustable The underlying idea could be signal, but the condition, qualifier, non-event, or bias is off. A targeted edit can rescue it. Noise Fails the response test, or the event cannot be defined tightly enough to avoid alert fatigue. Retire before build.
  3. For adjustable candidates, write the one-sentence sharpening that would move the candidate to signal.
  4. For noise candidates, name which part of the response test the candidate failed.
  5. After you submit all twelve, you will see the reference rubric and how your ratings compared.

There is no perfect score. The drill is calibration. What matters is which candidates you and the rubric disagreed on — those are the patterns most likely to bite you when you design your own watcher in Part B.