The dataset

Every case in this collection as structured, machine-readable data. Free, openly licensed, and generated from the same records that render the pages, so it cannot drift out of date.

40 cases · 8 failure patterns · no signup, no key, CORS open

What is in it

slug, name, aka

Identifiers. `aka` carries former or alternate names, which is often how a defunct project is searched for.

one_liner

A single sentence describing what the project was and what happened to it.

founded, launched, ceased_or_changed, lifespan, era_bucket

Dates as published by the source, not inferred. `era_bucket` groups cases into e-gov, Gov 2.0, the civic-tech peak and the platform era.

status, outcome

What happened: shut down, acquired, acquired then shut down, pivoted away, dormant, survived narrowed, still operating, thriving.

hq_city, hq_country, countries_operated, jurisdiction_level, geographic_reach_peak

Where it operated and at what level of government.

entity_type, funding_model, total_raised_approx, peak_funding_or_valuation

How it was constituted and paid for. Figures are approximate and sourced.

category, deployment_model, content_creator, participation_model, platform, key_features

What kind of thing it was and how participation worked in it.

primary_pattern, secondary_patterns

The failure pattern the case is filed under, and any it also demonstrates. The pattern set is included in the JSON.

confidence

High, medium or low, describing how well-evidenced the case record is. Read it before quoting a number.

sources

Every source behind the case, as structured JSON. This is the field that makes the rest checkable.

Read this before you use a number

Every case carries a confidence rating, and it is there because the underlying evidence genuinely varies. A shutdown covered by the trade press is not the same kind of fact as a funding total reconstructed from two blog posts and an archived page. Low-confidence cases are included rather than dropped, because the absence of a project from a record of failure is its own distortion, but they are labelled and should be quoted with the label attached.

Dates are as the sources give them. Where a project wound down over years rather than closing on a day, the fields say so rather than inventing a date. Funding figures are approximate by nature and marked as such.

The failure patterns are our reading. The facts are checkable and the interpretation is arguable, which is the correct division. If you think a case is filed wrongly, the sources are right there in the record and we would rather be corrected than cited uncritically.

Licence

Creative Commons Attribution-NonCommercial-ShareAlike 4.0. Use it, quote it, build on it, teach with it. Credit it and link back, keep it non-commercial, and share any adaptation on the same terms.

We chose those terms deliberately rather than the most permissive ones. This work stands on the shoulders of the Civic Tech Field Guide, whose graveyard sent us looking in the first place and whose own directory carries exactly this licence. Matching it seemed the right way to say so.

How to cite it

@misc{agora_civictech_postmortems,
  title  = {Lessons from Civic Tech History: a structured post-mortem dataset},
  author = {{Agora}},
  url    = {https://tryagora.org/civic-tech-history},
  note   = {40 cases, 8 failure patterns. CC BY-NC-SA 4.0}
}

Corrections

If you worked on one of these and the record is wrong, we would like to fix it, and we would rather hear it from you than from a footnote. Get in touch. The same goes for a project that belongs here and is missing.