Living Research

A notebook runs. Its results publish to a page. The repo is the research. The page is a view of it.

The problem

Papers freeze knowledge into static PDFs that can't be verified, updated, or rerun.
Results sit behind paywalls while the code and data to reproduce them are missing.
Peer review happens behind closed doors between two anonymous reviewers.
By the time a paper is published, the field has already moved on.
In a 2016 Nature survey, over 70% of 1,576 scientists reported failing to reproduce another researcher's experiments.

How it works

Notebooks are the primary artifact. Code, analysis, and explanation live together. Click a badge, run the full pipeline in Colab.
Results publish back to the repo. A publish cell commits outputs and findings data. GitHub Pages deploys the updated page.
CI validates, it doesn't produce. Workflows confirm notebooks still execute. The notebook is the source of truth, not CI.
Git history is the intellectual record. Every commit traces how thinking evolved. Every contributor is attributed.
Contributions happen through issues and PRs. Transparent, attributable, open.

living-research/template
├── notebooks/         # Jupyter notebooks — run end-to-end
├── scripts/           # utilities (Colab publish flow)
├── data/
│   ├── raw/           # immutable source data
│   └── processed/     # derived data, regenerated
├── results/           # outputs: plots, models
├── docs/              # findings page → GitHub Pages
│   └── data/          # data the page reads at runtime
└── .github/workflows/ # CI validates notebooks

Principles

Reproducibility by default. If it can't rerun, it's not research.
Living, not frozen. New data? Update the notebook. Found a flaw? Open an issue. Knowledge evolves.
Show the work. Dead ends, pivots, negative results. The history is as valuable as the conclusion.
Credit through contribution. Git tracks who did what. No ambiguous author lists.

The published page is a view of the research. The repo is the research.