The one-command rerun

The final test of reproducibility: can a stranger go from raw data to every figure and table with one command?

What it looks like

A stranger clones your repo and runs one line, such as make all or python run_all.py. Twenty minutes later, results/ contains every figure and table in your paper.

How to get there

1. Every step is a script: clean, analyse, plot. No manual clicks.
2. A single entry point runs them in order: a Makefile, a shell script, or a workflow tool like Snakemake or targets (in R).
3. Paths are relative, so it runs anywhere.
4. Long steps can be skipped with saved intermediate results, but the full rerun must still work.
5. The README says the command, how long it takes, and what it needs.

Prove it

Run it on a fresh machine: a new cloud notebook, a friend's laptop, or GitHub Actions. Delete results/ first, so you know everything really is rebuilt.

The payoff

When a reviewer asks for a change, you edit one line and rerun. When you find a bug, every figure updates. When someone builds on your work, they start in minutes, not weeks.

RememberFrom raw data to every result, with one command, on a fresh machine. That is reproducible.