Choosing My Licenses

The previous post made the case: the license is not a formality — it is what separates a sustainable open model from one that quietly gives everything away. Now comes the harder part: actually choosing.

Code and writing serve different purposes, attract different kinds of use, and need different protections.

My Wishes

The goal is not to restrict who can use the work. It is to make sure that anyone who profits from it has to engage — not take and disappear.

Concretely, what I want from any license I choose:

• Anyone can use, learn from, and build on the work. No permission needed. No barriers for individuals, researchers, non-profits, or companies building something of their own.
• Commercial operators who profit from the work must engage. The open terms are one path: keep changes open. A commercial license is another: pay for the right to keep them private. Both paths are available; neither is hidden.
• The line is compensation, not permission. Use the work freely. Build with it freely. The moment the work itself becomes a commercial product someone else is selling without engaging with me, that is where the license draws the line.

These are not restrictions on what the work can become. They are conditions on who gets to profit from it.

Options

Code

For code, the specific risk is a commercial operator running the code as a hosted service — charging customers, keeping modifications private, contributing nothing back. ✅ meets the criterion, ⚠️ partial or conditional, ❌ does not.

• Free to use — anyone can use, modify, and build without restriction; ⚠️ if free use is conditional (e.g. revenue or size threshold applies).
• Covers SaaS — copyleft or restrictions trigger on network access, not just on distributing copies; ❌ if only distribution triggers the license or there is no restriction at all.
• Compensation path — the license channels commercial exploitation toward payment rather than simply blocking it; ❌ if commercial use is only restricted or only freely permitted with no middle path.
• OSI-approved — meets the Open Source Initiative definition; ❌ for source-available or proprietary-addendum licenses.
• Dual licensing — copyleft leverage creates a meaningful paid commercial license option for operators who cannot accept the open terms; ⚠️ if dual licensing is technically possible but the leverage is weak.

Writing

For writing — blog posts, essays, and long-form pieces — the risk is commercial repackaging rather than SaaS deployment: someone compiling that work into a paid book or sold course without asking. ✅ meets the criterion, ⚠️ partial or conditional, ❌ does not.

• Free to share — anyone can redistribute the work without asking; ⚠️ if sharing is permitted but under specific conditions only.
• Allows adaptation — the work can be translated, remixed, or built upon; ⚠️ if adaptations are permitted but must carry the same license (share-alike).
• NC protection — commercial use requires explicit permission from the author; ⚠️ if commercial protection is present but ambiguous or conditional.
• Widely recognised — the license is broadly understood by readers, publishers, and platforms without explanation; ⚠️ if recognition is limited to specific communities.

My Decision

Across eighteen code licenses and eleven writing licenses, only one in each category satisfies every criterion. Not the most well-known, not the most permissive, not the easiest to explain at a dinner party — but the only one where none of the criteria came back as a cross.

Code

GNU Affero General Public License v3 (AGPL)

AGPL closes the SaaS loophole: most copyleft licenses only trigger on distribution, so a company can run the code as a commercial hosted service without ever sharing changes. AGPL treats network access the same as distribution — modifications must be published, or a commercial license must be purchased. That commercial license is the compensation path. Exception: for utilities, libraries, or tooling where SaaS deployment is not the risk, MIT or Apache 2.0 remains the right choice.

Writing

CC Attribution-NonCommercial (CC BY-NC) 4.0

CC BY-NC keeps the ideas in open circulation while reserving commercial rights. Anyone can read, quote, translate, adapt, and share freely — as long as the use stays non-commercial. What requires permission is commercial repackaging: a paid book, a sold course, a licensed content product. The individual pieces are not the product; the curation, sequencing, and assembly are — and the license does not give that away for free.

Alignment

Both licenses express the same position: the work is open for anyone building in good faith, and commercially closed to anyone who wants to profit from it without engaging.

That is not a restriction on commerce. It is a rejection of extraction — the model where a company takes what someone else built, packages it, charges for it, and contributes nothing back. Both licenses draw the same line, just in different domains.

AGPL draws it in code: use it, fork it, self-host it freely — but run it as a commercial service and you must publish your changes or buy a license. CC BY-NC draws it in writing: share it, quote it, adapt it freely — but compile it into a commercial product and you must ask first.

Two artifacts. Two license families. One principle. The terms are public and the reasoning is here. Anyone who wants to build a commercial relationship on top of this work knows where to start.