16 August 2026 · history gds govuk people
There are plenty of accounts of how the Government Digital Service was built. Most were written afterwards, by people with a view. The commit log was written at the time, by everyone involved, one entry per change, and nobody was composing it for posterity — which makes it the least flattering and most reliable source there is.
Who arrived, and when
Ten people, all of them publicly associated with this work, ordered by their
first commit to github.com/alphagov. The right-hand column is
everything they have committed across the public sector estate since, not just
in that repository.
| First commit | Who | Repository | Commits since |
|---|---|---|---|
| 2011-03-18 | James Stewart GDS's first Head of Technology, later Deputy CTO | geolib | 2,972 |
| 2011-09-25 | James Mead author of Mocha, the Ruby mocking library | whitehall | 9,359 |
| 2012-04-13 | Anna Shipman Technical Director of GOV.UK, later of the Financial Times | static | 2,622 |
| 2012-05-10 | Paul Downey GDS and HM Land Registry, open standards | scala-router | 5,062 |
| 2012-05-25 | Richard Pope founding product manager of GOV.UK | e-petitions | 816 |
| 2013-03-14 | Jeni Tennison later CEO of the Open Data Institute | government-service-design-manual | 28 |
| 2014-11-04 | Terence Eden UK government open standards lead | government-service-design-manual | 182 |
| 2015-04-27 | Andrew White Ruby on Rails core team | e-petitions | 4,643 |
| 2015-08-06 | Robin Whittleton GDS front-end and accessibility | govuk-puppet | 2,440 |
| 2021-11-13 | Paul Robert Lloyd designer and writer | stylelint-config-gds | 10,558 |
The first is James Stewart in March 2011, more than a year before
GOV.UK launched, committing to geolib — a geolocation
library. Not a strategy, not a manifesto: a small utility for turning
postcodes into places, which is what the beginning of a government website
actually looks like.
The tell is what people touched first
Jeni Tennison and Terence Eden made their first commit to the same repository:
government-service-design-manual.
That is the Service Design Manual — the written standard for how digital services in government should be built, later the Service Manual and the Service Standard. It is prose, not code. What the commit log records is a technical director who would go on to run the Open Data Institute, and the civil servant who would go on to lead the government's open standards work, both editing the written standard in the same version control system as the software, through pull requests, in public.
That was a deliberate and slightly radical choice at the time and it is easy to miss now. The standard was managed like code because the people writing it thought a standard should be reviewable, diffable, and open to challenge from anybody who could open a pull request. Whether the rest of government followed is a separate question — the standards page suggests largely not — but the intent is in the log.
The ones who never left
The most striking thing in this data is not the founders. It is the people who are still here.
James Mead on alphagov/whitehall, 2011-09-25 to 2025-12-16 — the same person, the same repository, for fourteen
years. James Mead is the author of
Mocha, one of the standard mocking
libraries in Ruby, and whitehall is the publishing application
behind GOV.UK's departmental content. Andrew White is on the Ruby on
Rails core team and has been committing to the government petitions service
since 2015.
Two people who could have their pick of open-source work have spent a decade each on the machinery of British government. Every argument about whether the state can attract serious engineers has this sitting underneath it, and it is not the argument either side usually makes.
What this does and does not show
It shows presence, not contribution. Commit counts measure typing, and a person who shaped a system in review and design can leave a smaller trace than someone who ran a bulk rename. Several of the names above have single-figure commit counts and were running the programme; the number next to a name is the worst available proxy for their importance, and it is included only because hiding it would be worse.
It also shows only what is public. GDS worked in the open by policy, which is why this record exists at all — and the forge survey is a reminder that most of government did not, so most of government has no equivalent history anybody can read.
Method: every person named here holds or held a public professional role, was identified from public sources and then looked up in the harvest, and is corroborated by commits from an address at their own published website. Name matches without that corroboration were discarded, including one row displaying a name above an email on an unrelated person's laptop. Figures are repository names, dates and commit counts from fork-excluded history, recomputed on every build. No email address, working pattern or location appears here, and none was analysed. Anyone who would rather not appear can say so — see the privacy notice. See Method.