SoftwarePoliticsTracker: a hand-curated index of political statements by software projects

The repository is a README and nothing else

The whole of SoftwarePoliticsTracker is a table in a README, and the top level of the repository contains exactly two files, a LICENSE and that README.

There is no code, no configuration, no data file, no schema and no build. The primary language of the repository is not even recorded, because there is no source code to record a language for. The licence is BSD-2-Clause.

That is worth saying plainly at the start, because the name and the description, tracking the political stances of software projects, foundations and corporations, suggests a tool with a database behind it. There is no database. The table in the readme is the artifact.

The format of an entry is consistent throughout. Each row has a marker column, a bolded name of the organisation, and a notes cell containing a comma separated list of claims, each one a markdown link to a post. So Microsoft, IBM, Red Hat, Mozilla, GNOME, GTK, KDE, the Linux Foundation, the Python Software Foundation, NixOS, Fedora, Ubuntu, Bazaar, the Software Freedom Conservancy, elementaryOS, Debian, Arch Linux, Bazzite and Chimera Linux each appear with their own row, and the marker column uses a left arrow glyph for the category the readme calls Woke, Leftist or DEI.

The readme says it is a work in progress and that there is a lot of data left to add, and the visible table is a small fraction of the software world either way. There are no GitHub releases, and the last push was on 2026-09-06.

So the honest description of the project is a published, human-maintained list with a link on every claim, which is a different and more modest thing than a tracker.

Two scope notes that do most of the work

The readme opens with two notes, and they are the most important content in the document because they define what the table is not claiming.

The first note says it does not include statements made by users and average contributors, or by employees of software organisations. Only official actions and statements from the organisation itself, or from leadership or official representatives, count.

That is a defensible and fairly demanding inclusion rule, and it is also a rule with no stated enforcement mechanism. Nothing in the repository shows how a claim was checked against that criterion, or who checked it. Whether a project lead's remark on a personal account counts, or whether a conference talk by a maintainer counts, or whether a foundation's published statement counts, are all judgement calls the readme does not document.

The second note is the one that matters most for anyone planning to cite this. The readme is a work in progress with a lot of data left to add, which means the absence of an organisation from the table says nothing about that organisation, and the presence of one says only that a claim was found and linked.

There is a third structural point that is not stated as a note but is obvious from the links. Almost every citation in the table points at posts on the Lunduke Journal account on X, or at articles on the same author's own Substack and locals site.

So the sourcing model is: one author finds posts, one author decides what counts, and the table records the link. Every claim is checkable, and checkability is the one real strength of the format.

What a row actually contains

Reading a few rows closely shows what the table records and what kind of claim it is making.

The Microsoft row lists encouraging and paying for employees to gender transition their children, conducting what it calls Pronouns in Software research, providing Trans themes for Office, and promoting Trans activism in games. The IBM row lists a DOJ settlement over discriminatory DEI policies. The Red Hat row lists anti-White training and DEI hiring quotas, a spokesperson promoting Antifa, prominent employees attacking what the author calls MAGA and Linux, and a representative assaulting Framework Computer and DHH over politics.

The Mozilla row lists a non-binary Firefox mascot with They and Them pronouns and a rebranding as a global crew of activists alongside Marxism and LGBTQIA+ activism coverage. The GNOME row is the longest, with a dozen or more entries covering a Trans Day of Visibility, a non-binary people's day, a Crush Fascism and Free Palestine message, an Antifa assertion, a release manager described as a registered sex offender, an accusation against Framework Computer, a pledge of fealty to the UN, an extended Pride month, Pride Month framing of open source.

The pattern across all of them is the same. Each item is an action, a statement, or a piece of policy that the organisation took, and the framing is the author's characterisation rather than a neutral description. A pledge of fealty to the UN and a DEI hiring quota are recorded in the same register, and the arrow marker indicates that the organisation is placed in a category rather than that the organisation self-identified that way.

The Python Software Foundation row is worth noting because it is the one case where the linked claim is about restraint rather than action: turning down a $1.5 million US government grant rather than dropping discriminatory DEI policies, followed by launching a DEI marketing campaign after rejecting the grant. The Software Freedom Conservancy row, by contrast, records an attack on the author as Fox News of FOSS.

So the table's unit of analysis is the organisation's conduct, filtered through one person's threshold for what counts, and published with links so a reader can disagree with the framing and still find the underlying fact.

The mechanics of a claim, and what a reader can and cannot infer

Every row follows the same shape, and understanding it is what makes the table usable rather than merely quotable.

The first column is a marker, a left arrow glyph, and every visible row carries it. The readme explains what it means: the listed organisations have made statements or taken actions that the author places in a category he describes as Woke, Leftist or DEI. The heading above the table uses the same vocabulary.

The second column is the organisation name in bold. The third is the substance, a set of individual claims separated by commas, each wrapped in a markdown link. The links are the point, because they turn each entry from an assertion into a reference. A reader who wants to check the Red Hat claim about anti-White training does not have to take the table on trust, they follow the link.

What a reader cannot infer is anything about organisations not listed, and the readme says so directly. A partial list that presents itself as partial is a different artifact from a partial list that presents itself as a ranking, and this one is explicit that there is a lot of data left to add.

Two further limits follow from the structure rather than from the readme. First, the unit of analysis is the organisation, not the individual, so a row aggregates several separate events into one entry, and there is no way to tell from the table alone which of those events the author considers most significant. Second, the claims are of very different kinds. A DOJ settlement over policies, a mascot, a hiring quota, a pledge to an organisation and a talk at a conference are all recorded in the same register, and they are not equally weighted as evidence of anything.

None of that is criticism of the project as an index. It is a description of what a list with one author, one set of citations and no stated methodology can and cannot support, and it is why the format makes verification possible rather than making verification unnecessary.

Why the format matters more than the content

There is a defensible argument for a project in this shape, independent of whether you agree with the categorisation.

The starting point is that political conduct by software projects is scattered across a large surface. A foundation's blog, a project's release notes, a conference talk, a GitHub issue, a mailing list post. Collecting the official statements of a dozen organisations and putting a link on each is real work, and the work of deciding which statements count as official is the interesting part.

The other thing the format does is make the selection inspectable. Every claim in the table is one click from the claim. A reader who thinks a row is miscategorised can open the link, see the statement in its original context, and decide for themselves. That is a better property than a summary list with no citations, and it is better than a list of uncited assertions.

What the format does not do is balance the picture. There is no column for the other side, no set of entries describing conduct that would place an organisation in a different category, and no methodology describing how claims were found. The readme says the data is incomplete, which is honest, and that is a different thing from saying the selection is representative.

The category vocabulary is also doing unstated work. The readme describes the marked entries as statements or actions best categorised as Woke, Leftist or DEI, and the arrow glyph is a left arrow, which visually points the reader toward the left. That is a design decision in the presentation rather than a claim about any organisation, and it is a decision the author has made.

For a reader, the sensible use is the obvious one. It is a list of places to look. Open the links, read the primary statements, and form your own view, because a table with a thesis built into its layout is a starting point for research rather than a conclusion from it.

Editorial conclusion

Adopt SoftwarePoliticsTracker if you want a starting index of where to look for official positions taken by particular projects, and you intend to check each source yourself. Do not adopt it as a reference, a dataset for analysis, or a balanced account, because it is one person's selection with one author's citations and no stated methodology, and the readme itself says there is a lot of data left to add. Before you rely on any row, open the linked source and read the primary statement, and treat the categories as the author's framing rather than a classification anyone has reviewed.

Frequently asked questions

Official sources

1. BryanLunduke/SoftwarePoliticsTracker on GitHub
2. Issues
3. License: BSD-2-Clause
4. README