Field note: March

March was a writing month that didn't produce much writing. That's not a contradiction. Most of the work was in the revisions — the invisible kind, where you're not adding sentences but removing them, and the draft stays the same length while getting materially better. Or that's what I told myself by the end.

The paragraph I rewrote nineteen times was the opening to a piece I haven't published yet. I know the count because I kept the drafts. Each version was objectively better than the last except for the four in a row that were worse, which forced me to go back to draft fourteen and continue from there. I finally let go of it not because it was done but because I had stopped being able to tell the difference between "this needs work" and "I'm scared to publish this."

That distinction is the thing I learned in March.

What I shipped

• Two dispatches — shorter pieces, both published in the first week when the perfectionism hadn't fully set in yet.
• Trayd beta onboarding flow — the second major revision of the onboarding since January. This time we removed the progress bar, which customers were gaming, and replaced nothing. Cleaner.
• A decision I'd been avoiding for six weeks — killed a data integration feature that had been "80% done" for two months. It wasn't 80% done. It was 80% of the easy parts done and 0% of the hard parts.

What I read

• Bird by Bird — Anne Lamott · Re-read for the fourth time. The "shitty first drafts" chapter is still the most useful single piece of writing advice in print. The rest of the book earns its place around it.
• The Elements of Style — Strunk and White · I read this every year in March. Twenty minutes. It keeps me honest about passive voice and it keeps me honest about keeping it short.
• Staff Engineer — Will Larson · The "write the design document" sections hit differently when you're running a small product team and writing the doc yourself because no one else will.

What I noticed

The nineteen-draft problem was not a writing problem. It was a publication anxiety problem with writing symptoms. I kept revising because each revision extended the time before I had to decide whether to ship it. Recognizing that pattern didn't fix it — I still spent three weeks on a paragraph — but it meant I could make a deliberate choice to stop.

The drafts I publish fastest are the ones where I know who I'm writing for. When I have a specific person in mind — not an audience, a person — the revisions stop being about "is this good" and start being about "will this land for them." That's a tractable question. Abstract quality standards are not.

March also confirmed that I can't write in the afternoon. I've suspected this for a year. March gave me the data. Three weeks of morning-only writing blocks produced more finished work than the two weeks when I scheduled evening sessions after product work. Afternoon me is useful for editing. He is not useful for first drafts.

March score: 6/10. Good process work. Not enough finished pieces. The unpublished essay is better for the nineteen drafts. Probably.