T — Tracked

Version information must be recorded for all components, ideally using the same content-addressed version control system. The primary value is not version numbering (“v1” vs “v2”) but content-addressed identification – two datasets with identical content hashes are provably identical.

Tracking encompasses not only version history but also provenance: what actions produced or modified each component, what inputs were consumed, and what versions of code and environment were involved. For code-driven modifications, provenance should be captured programmatically rather than by manual annotation.

See the STAMPED paper for the full treatment.

Walkthrough: stellar distances from Gaia parallax

12 March 2026·14 mins

Reproducibility Rigor Transparency Reusable Accessible Workflow S — Self-contained T — Tracked A — Actionable M — Modular P — Portable E — Ephemeral D — Distributable STAMPED-intro gaia parallax walkthrough python datalad make

Incrementally builds a research object from a bare script to a tracked, portable, reproducible pipeline — motivated by real problems, not by acronym order.

From Script to STAMPED Research Object

20 February 2026·9 mins

Reproducibility Rigor Transparency Reusable Accessible Pattern S — Self-contained T — Tracked A — Actionable M — Modular P — Portable E — Ephemeral D — Distributable STAMPED-intro shell POSIX awk make git singularity containers

Four scenarios showing how to incrementally add STAMPED properties to a shell-based analysis using git, make, and singularity.

Git Submodules for Modular Dataset Composition

19 February 2026·6 mins

Reproducibility Rigor Interoperable Reusable Tool M — Modular T — Tracked git submodules dataset-organization

Demonstrates how git submodules enable independent versioning and composition of dataset components.

Recording Computational Provenance with 'datalad run'

19 February 2026·6 mins

Reproducibility Transparency Reusable Tool T — Tracked A — Actionable datalad provenance reproducibility

Shows how ‘datalad run’ wraps arbitrary commands to record inputs, outputs, and the exact command in machine-reexecutable form.