Skip to content

Crowd sourcing

github-actions[bot] edited this page Nov 17, 2026 · 1 revision

The open database and the API

LibreLifts is an app and a crowd-sourced open database of lifting reference data. This page is the plan for the database: what it is modelled on, the order it is built in, how it grows and what can go wrong. The decisions themselves are in the records:

  • ADR 0013: what the database is and what it holds.
  • ADR 0014: the licences, and what contributors agree to.
  • ADR 0015: how to contribute.
  • ADR 0016: the hosted API, which is the paid product.

The model

The data is open: ODbL for the database, CC BY-SA for media, with free dumps. The app is free and bundles a snapshot of the data. The project earns money from a hosted API to the database, which has a free tier and paid tiers.

What is sold is the service, not the data. A customer pays for higher limits, fresher data, media delivery, uptime and support.

The limit of this model. Anyone can build a competing app or API from the dumps. The API's terms only stop a product from running on LibreLifts' servers without paying. This is the price of open data, and the reason contributors can trust the project.

What other projects do

Figures were checked on each project's own pages on 2026-11-17 and will go out of date.

Project Data licence How people contribute What is charged for Lesson
TheSportsDB None stated; no dumps Accounts, with editors API tiers at $9 and $20 a month; free at 30 requests a minute Cheap tiers sell to hobby developers. Grew through Kodi and Plex.
MusicBrainz CC0 core; CC BY-NC-SA for the rest Every edit is voted on; history is public Commercial supporters, from $100 a month Companies pay for a current feed and support, although the data is free.
Open Food Facts ODbL; images CC BY-SA Open editing; moderators revert Nothing; donations and grants A phone app that adds data while it is used drives contribution.
OpenStreetMap ODbL Open editing; changesets are public Nothing; others sell hosting on top The foundation is bound to keep the data open.
wger CC BY-SA, per record Accounts with a minimum age Nothing; donations Without review, quality is uneven and duplicates pile up.

Projects that went the other way are the warning. CDDB and IMDb closed data that volunteers had built. TheTVDB started charging the open-source apps that sent it contributors, and Kodi moved to TMDB.

What the database can be that others are not

  • Clean provenance. Every record and every media file has an author and a licence.
  • Stable ids, with aliases ("RDL" and "Romanian deadlift" are one exercise) and variations.
  • Lifting semantics: how load is counted (per hand, bar included, share of body weight), movement pattern, one or two sides.
  • Media with cleared rights, in one consistent style.
  • Program templates in a format an app can run.
  • Open strength standards, derived from OpenPowerlifting's public-domain results.
  • Ids that link out to wger and Wikidata, so other projects can adopt the data.

Phases

Each phase is usable without the next one.

Phase What is built What it costs to run
0. Decisions The four records and this page. Nothing
1. Seed A schema, and the six lifts of the 5x5 program, growing to the common barbell and dumbbell exercises. The app uses the bundled snapshot. Contributions by pull request. Nothing
2. Publish Free dumps with each release, and a static page per exercise on the website. Nothing
3. API A hosted, read-only API with keys and a free tier, then paid tiers. A server, billing
4. Editor Accounts, review of edits, media uploads, translations. Moderation
5. Breadth Program templates and strength standards. Review

Phase 3 is the first that needs a server, a privacy policy, terms and a legal entity. Each of those gets its own decision record before it is built.

Prices

Not decided. As a starting point for that decision, comparable APIs charge between $9 and $100 a month, and a ladder such as free, $9, $29 and $99 fits that range. The free tier has to stay generous for open-source and personal projects, because they are where contributors come from.

How it grows

  1. The app is the first user. Every exercise the app needs is a reason to add a record.
  2. Other open-source apps are the second. TheSportsDB and TMDB grew because media centres used them by default. The aim is to be the dataset an open-source fitness app reaches for: a documented schema, stable ids, dumps, and a mapping to openweight (ADR 0006).
  3. Exercise pages bring search traffic, and each page asks for corrections.
  4. Announce when there is something to use: after phase 2, not before.

Risks

  • Media rights. The largest legal risk in fitness data. One file of unknown origin can taint a dump. The rule in ADR 0014 is checked in CI.
  • Quality. Open datasets fill with duplicates and half-finished records. Review before merge is the defence, and it does not scale past one reviewer without tooling.
  • Scrapers. Free APIs are scraped heavily. Keys, per-key limits and free dumps answer that.
  • One maintainer. Review, moderation and support all land on one person until there is a community.
  • Slow income. Projects on this model earn little until other products depend on them.

Clone this wiki locally