DEV Community

Odinaka Robert Nnamani
Odinaka Robert Nnamani

Posted on

On to the next phase

On to the next phase

A project moves to its next phase when the question it has to answer changes. Phase one asks: does it work? Phase two asks: does it hold up when someone else touches it? The code does not announce the change. The work does.

I built rootmail alone since June, and phase one was mine: my machine, my inbox, my own tests. The SES piece I published in October was phase one thinking made public: an accepted call is not a delivered message, and the check is the work. Then I opened the beta, and the question changed. Eight invite-only seats. Twelve sends a day per organisation. Sandbox limits, confirmed inboxes only. The target is eight testers by November 26, and the work now is not adding features. It is proving the thing behaves when someone else touches it.

The same change happened with heypascal.me. I launched it in October as a menu, not a portfolio: eight services, prices next to them, a brief form that asks what you need. Phase one was building the site. Phase two is the briefs, and the briefs ask a different question: can someone who has never met me understand what I do well enough to hire me for it?

Here is what changes when the question changes.

The checks move first. A feature counts as done when the test that proves it passes, in an environment I do not control. A flag counts as finished when it is removed, not when the feature ships. A limit counts as real when it is written where the user can see it, in the headers and the 429, not when it lives in my head.

The record becomes the work. In phase one the record was for me: notes, a checklist, the next thing to build. In phase two the record is for the next person: what shipped, what broke, what is next, written where anyone can read it without asking me. The sendlit fix I suggested in an open-source thread landed as a merged pull request last week. I never met the maintainer. The thread was the whole working relationship, and the record was the proof.

And I stop adding. Phase one rewards adding: one more feature, one more endpoint, one more idea. Phase two punishes it, because every addition is another thing to prove. The discipline of the next phase is subtraction: the flag that comes out, the test that names its own setup, the limit said out loud before launch.

Nobody hands you a certificate for the next phase. There is no date on it. You decide the question changed, and then you do the work the new question asks for.

What question is your project answering right now: does it work, or does it hold up? Name the question, not the feature.
Next in this series: Seeking clarity.

Top comments (0)