Research manifesto

The bottom rung is gone. Here's what replaces it.

The first software job is being compressed. The replacement has to be public, reviewable, and real.

The old bargain said: get hired, learn on bounded work, earn trust, and grow. The new market increasingly says: prove you can work before anyone gives you the work.

Argument 01

The apprenticeship layer was never automatic. It was maintained.

The junior role used to be a rough but reliable social contract. A company hired potential, handed it bounded work, and let new developers learn judgment under review. The work was not glamorous. It was tickets, bugs, tests, migrations, docs, support threads, and code review.

That contract is cracking. The tasks that made early careers legible are the same tasks companies are most tempted to automate, compress, or hand to senior people with AI copilots.

Argument 02

The experience paradox has become infrastructure.

The first casualty is not only a job posting. It is the loss of a place where beginners can safely make small mistakes, learn the local standard, and build a record that someone else inspected.

Without that ladder, the market asks new developers for experience while removing the low-risk work that used to create it.

Argument 03

When output is cheap, proof has to move closer to reality.

AI makes private output cheaper. It can produce a polished demo, a plausible explanation, and a respectable resume bullet in seconds. That is useful for learning, but it weakens the hiring signal when every applicant can manufacture similar artifacts.

The replacement signal is not louder self-promotion. It is public work: issue context, pull request diffs, maintainer feedback, revisions, and merged changes.

Argument 04

Open source turns learning into a public trail.

Open source is not a perfect apprenticeship system. Maintainers are busy, issues vary in quality, and labels are only hints. But it has one advantage the new labor market badly needs: the evidence is public and reviewable.

A small accepted pull request shows more than code. It shows reading comprehension, constraint handling, patience with feedback, and the ability to ship inside someone else's system.

Argument 05

AI belongs in the workflow, not in place of accountability.

The new first rung should not pretend AI disappears. Developers should use it. But the unit of proof must be the reviewed change, not the generated artifact.

The useful question is not whether a beginner used AI. It is whether they understood the problem, validated the change, responded to review, and left the project better than they found it.

Argument 06

The replacement rung is contribution, not credential theater.

Good First Issue exists for that narrower, more durable path. It helps new developers find work small enough to start and real enough to matter.

The answer to a vanishing bottom rung is not nostalgia. It is a better first step: public, scoped, maintainable work that gives beginners a fair way to prove they can ship.

Start here

Build the record the market can inspect.

Choose a focused issue, work in public, respond to review, and leave a durable trail of real contribution.

Browse issues

Sources