Bring one claim: the human path

You can try MCRP with a page of notes and a colleague. No account, token, trust score, coding environment or new publication venue is needed for the exercise. Use a synthetic case from this packet, or material you are entitled to discuss. The proposed times below are facilitation budgets, not measured completion times.

You can start alone. Pick one row below and finish only its sentence; a colleague is needed for the later paired exercise, not for this first contribution.

These are worked starting answers, not endorsements. An equally useful response is “I cannot tell what this sentence permits me to conclude,” with the confusing phrase identified. No equation derivation, software installation or complete domain chapter is required for that response.

First contact: ten minutes

Pick a domain case and one intended use. Complete this card in ordinary language:

“Unknown” is an acceptable answer. It may prevent a particular use; it should not prevent the person from contributing a useful partial check. Do not treat a confidence slider as a substitute for these answers.

For example, an astronomy card could say: claim: the revised scalar estimate is 10.0; version/evidence: teaching calculation v2, raw 10.5 and gain 1.05; check: divide those numbers; not checked: physical calibration; decision maker: I may choose it for my own synthetic classroom calculation, but no real catalog authority has been supplied; revisit if: the raw value, gain, or intended use changes. A “version” simply identifies the particular copy being examined. A “dependency” is something that copy’s claim uses, such as the gain value. “Rely” means choosing a particular use, not declaring general truth.

First contribution: a forty-five-minute paired exercise

One person plays the author and introduces the amendment; the other checks and states a possible classroom use. Those combined teaching roles are not a model of independent professional review. Use fictional role names if that is more comfortable. A facilitator is optional; if nobody facilitates, record zero facilitator time rather than inventing a third participant.

1. Offer, 8 minutes. The author supplies the card and a frozen, shareable evidence excerpt. The checker restates the intended use in their own words. If they disagree about the target, repair the wording before checking.
2. Check, 12 minutes. The checker performs one feasible action from the domain case, saves what was examined, and records passed, failed, partial or unknown. A timeout is partial or unknown, never an implicit pass.
3. Rely, 8 minutes. The designated exercise decision maker states a permitted toy use and its limits, or records that coverage is unavailable. This exercise designation grants no real scientific, institutional or legal authority.
4. Amend, 10 minutes. Introduce the case’s dependency change. Both participants identify the exact prior uses that need reconsideration. Keep unaffected uses distinct and explain any uncertainty about the dependency map.
5. Reflect, 7 minutes. Count author, checker and facilitator time, missing evidence, unresolved disagreements, and unnecessary form fields. Compare with an ordinary referee paragraph on the same task.

The goal is a legible record and an honest limit. Correctly declining an unsupported use is a successful exercise outcome.

A facilitated domain clinic: seventy-five minutes

A facilitator prepares two comparable synthetic tasks, accessible materials, and a short answer guide. Alternate which task receives ordinary prose review and which receives the four-action card across pairs. Do not infer causal efficacy from this small convenience exercise.

Capture all participants offered a task, all who started, all who completed, refusals and reasons if voluntarily supplied. Include facilitator and repair labor. A cleaner result among completers can hide exclusion. Provide an asynchronous text option and permit withdrawal without penalty. Do not collect identifiable observations for research or make evaluative claims about people without the appropriate human-led ethics, consent and data-handling decisions.

Progress through useful responsibilities

These are suggested contribution paths, not status ranks or automatic promotion. Opportunity to learn requires supervision capacity. A newcomer reserve cannot produce unavailable expertise. A contributor’s reputation does not authorize them to decide outside their remit.

Contribution card

Send or file the following only in the destination the launch names. An issue on a public repository is public; it is inappropriate for personal allegations, patient data, confidential reviews or sensitive rights requests.

For a first contribution, the example/version, what you tried, and what confused or failed are enough. The fuller card below is optional; “unknown” and blank optional fields are acceptable. If no public feedback destination has been announced yet, keep the note locally or discuss it with a colleague; the exercise does not require submitting anything or finding a maintainer’s personal address.

Domain and intended use:
Example/claim version:
What I tried:
Expected behavior:
Observed behavior (include unknowns):
Smallest reproducing example, if shareable:
Suggested simplification or repair:
Time spent, including setup and follow-up:
Permission to quote this contribution: yes / no / ask

The launch must name an accountable intake owner, an alternate and the channels they can actually service. Acknowledgment is not acceptance. Publish realistic capacity and pause new discretionary commitments when capacity is exhausted. Keep appropriate reporting routes for existing obligations reachable.