# Instructor guide: Portable replication-package release gate

Course: Open replication laboratory

Suggested time: 80–120 minutes

## Learning target

Researchers turn an exploratory result into a portable, independently executable evidence package that preserves positive, null, and failed outcomes.

## Prepare

- Define the package schema and release license.
- Provision a clean rerun environment.
- Prepare one hidden missing dependency and one data-lineage break.

## Facilitation moves

- Do not let the original author operate the independent rerun.
- Require immutable raw inputs and checksums.
- Classify discrepancies before changing code or protocol.

## Misconception checks

- **Publishing analysis code is sufficient for replication.** Hardware/sample identity, calibration, raw data, metadata, environments, decisions, and failures are also required.
- **A failed independent run proves misconduct or falsity.** It triggers structured discrepancy localization across protocol, apparatus, environment, data, and analysis.

## Accessibility and participation

- Provide a machine-readable and human-readable manifest.
- Use explicit gate labels and remediation owners.
- Ensure documentation is screen-reader navigable and command examples are copyable.

## Evidence of learning

- A complete immutable manifest
- A clean-environment rerun
- A classified discrepancy and release decision

## Extension

Run a third-party reproduction using only the public package and publish a versioned discrepancy report.

## Evidence boundary

Assess the learner's reasoning only within the declared model and recorded observations. Do not upgrade a simulation result into a claim about an unmodeled physical system.