Obsidian Election Research Workspace

A shareable starter vault for building interactive, source-aware election research dashboards in Obsidian.

This project extracts the reusable UI and data contract from a larger 2026 midterm research system. The original workspace covers 176 races; this public template ships three fictional races so it can demonstrate the complete interface without redistributing third-party datasets or research notes.

What it demonstrates

• One interactive dashboard per race, rendered from a small JSON sidecar.
• Candidate, forecast, polling, campaign-finance, historical-result, and model views.
• Keyboard-accessible view switching and responsive layouts.
• A clean boundary between handwritten research notes and generated data.
• A validator that checks every race before the workspace is treated as healthy.
• Synthetic fixtures suitable for screenshots, workshops, and automated tests.

Try the demo

1. Copy demo-vault into a new or existing Obsidian vault.
2. Install and enable the community plugin Dataview.
3. Open Elections/Election Research Demo.md.
4. Follow the links to the three example race notes.

See INSTALL.md for the complete starter-vault and existing-vault paths.

Each race note embeds the same view:

await dv.view("Resources/Views/election-race", { raceId: "demo-north-senate" });

The renderer loads Resources/Data/Elections/races/<race-id>/dashboard.json. Pass dataRoot in the view input if your vault uses a different layout.

Repository layout

demo-vault/
  Elections/                         Example notes
  Resources/Views/election-race/     Dataview JavaScript and CSS
  Resources/Data/Elections/races/    Synthetic JSON sidecars
scripts/validate_workspace.py        Schema and note-link validation
tests/                               Contract tests

Why synthetic data?

The interface is reusable; a current election dataset is not automatically redistributable. Public political facts may still require attribution, while forecasts, polling aggregations, and model outputs can carry separate terms. The starter vault therefore uses invented jurisdictions, candidates, pollsters, and sources. Replace them with data you are licensed to use and cite every real source in the sidecar.

Built with AI

The workspace was developed through human–AI collaboration across interface design, data normalization, validation, and accessibility. The original system was validated across all 176 race notes, while this public extraction includes synthetic fixtures and its own contract tests.

See ARCHITECTURE.md for the data flow and extension points.

Related AI Experiment

MacWiki applies the same human–AI collaboration method to a native macOS Wikipedia client with an extensive Swift test suite.

License

Apache-2.0. The demo data is fictional and provided only as a software fixture.