The Contributor Ladder

Most maintainer effort lands on the tier that returns the least. This is the model I use to decide where it should go instead.

It does not rank people

Nobody is a Surface contributor. Someone is currently at Surface, on this project, today. The question a maintainer or a founder should be asking is never “is this person worth my time.” It is: what moves them up one level, and can I automate everything below it?

How this differs from the CNCF ladder

The CNCF contributor ladder is a governance document: it defines formal roles, the requirements to earn each, and the privileges that come with them, from contributor to reviewer to approver to maintainer. It answers “what roles exist here, and how does someone get one.”

This model sits underneath that. It answers a different question: who is ready to climb, and where should a maintainer spend the hours to get them there. A project can have an excellent contributor ladder written down and still watch nobody climb it, because nothing in the document transfers the context that a promotion assumes. Use both.

The three tiers

Reading the tiers without bias

Read each row as what the project owes someone at that point, not as a verdict on them. Start from what you can observe, not from who they are, and the next move follows from what they need.

The long-running pattern

None of this is new in shape. For more than twenty years, studies have found a small group doing most of the work and a large group arriving once and leaving.

None of these sources uses my three tiers, and none measures maintainer hours directly. They count contributors, merges, rejected pull requests and survey answers. The two 2026 papers are preprints, not yet peer reviewed. Read all of it as evidence that the pattern is real and getting steeper, not as proof of the model.

What moves Builder → Steward

Give authority rather than volume. The most common way to lose a good Builder is to reward them with more tickets, which reads as a penalty for being useful. Stewards emerge when someone is handed a decision that matters and is trusted to be wrong occasionally: review rights, a say in a real tradeoff, ownership of something that outlives their attention.

Designing for different paths in

People reach each tier by routes that rarely resemble each other. Culture, time zone, first language, career stage, and cognitive style all shape the path in, ADHD and autism among them. That makes it a design constraint on the conversion mechanics above. If your only path from Surface to Builder runs through a synchronous call, a real-time chat culture, or an unwritten social norm, you have capped who can climb. Written, asynchronous, low-social-cost paths convert more people, and they convert them from more places.

Why this matters more in the AI era

AI mass-produces tier 1 artefacts. The flood lands on the exact tier with the worst effort-to-outcome ratio a maintainer has, and starves the Builder-to-Steward pipeline that keeps projects alive. That is the barrier to human contribution I keep pointing at. The answer is to let AI absorb tier 1 entirely, so that scarce human attention survives for tier 2, where it compounds.

I track how Bluesky and Hacker News are talking about AI-written code

If you’re building commercially on Open Source

The same ladder is how you keep a clear divide between volunteer and commercial contribution without pretending the divide doesn’t exist. Paid contributors typically enter at Builder and are pushed toward velocity. Volunteers climb slower and carry longevity. A project that routes both through identical expectations loses the volunteers first, and then, a release or two later, the institutional memory that made the velocity possible.