Skip to main content

What Changed When I Started Shipping Again - Asterisk Pound

A follow-up to The Friction Loop: six months later, what actually broke the avoidance pattern and got me building again.

Alan AsherNovember 17, 20265 min read
personalreflectionhabitsside-projectsfocusdci-scores

What Changed When I Started Shipping Again

Back in April I wrote about why I hadn’t been building and then about the friction loop: sit down, feel the weight, reach for the phone, feel relief, repeat.

I ended that post with a short list of things I wanted to try:

  • make the next step painfully obvious
  • reduce what “working” means
  • focus on starting, not finishing
  • remove as much decision-making as possible

Then I went quiet for six months.

This time it wasn’t because I stopped building.


The numbers

I have a side project called DCI Scores, a site and app for drum corps scores. Its git history tells the story better than I can:

  • March: an initial commit and a basic scraper. Then nothing.
  • April through mid-August: zero commits.
  • August: 29 commits, and the app got started.
  • September: 38 commits. The site went live.
  • October: 208 commits on the site and 110 on the app.

The project didn’t change. The idea is the same one I’d had since March. What changed was how the work was shaped.


The next step became a sentence

The biggest piece of work this fall was importing forty years of historic drum corps scores, 1972 through 2012. That’s a huge, vague job, exactly the kind that used to send me straight to my phone.

So I didn’t treat it as one job. I treated it as forty-one.

Each season got the same routine, written down once in a plan file with a status table. Starting a session meant typing four words:

“Let’s do 1978.”

That was it. No deciding where to start. No re-reading code to remember where I left off. The status table said which year was next, and the routine said what to do with it.

The gap I wrote about in April, between “I should work on this” and “here’s the exact next step,” was gone.


Decisions made once, stored as data

The other thing that used to wear me out was making the same judgment calls over and over.

Old score sheets are messy. A corps scored zero because of a rain-out. A show shows up in two places with slightly different numbers. A corps changed its name three times.

Each time I made a call, it became a rule written down as data, not just something I remembered:

  • No score, no row. A corps that didn’t perform doesn’t get a placement.
  • A renamed corps is one corps across the years.
  • The published total always stands.

By the late 80s, most of the questions in a season were ones I’d already answered. I only had to look at the new ones. When I reviewed a year, I replied only to the items I disagreed with; anything I didn’t mention was approved.

That cut the number of decisions per session from dozens to a handful.


The reward came sooner

In April I said the garden worked because the feedback was immediate. You water something, it grows.

Locking a season turned out to feel the same way. At the end of a session, a year of drum corps history was done: reviewed, locked with a hash, tested, and showing the right champions. Something was there that hadn’t been there before.

And then the people who care about this stuff started using it. That’s a faster feedback loop than any side project I’ve had.


Not doing it alone

I should be honest about one more thing: a lot of this fall’s work was done pairing with an AI coding agent.

That didn’t remove the thinking. Every ruling about the old scores was mine, and plenty of them took real digging. But it removed most of the friction that sits around the thinking: setting up the session, remembering where I was, writing the boring version of the test, chasing a typo across 40 files.

It also made it easier to start. “Let’s do 1978” is a lot easier to begin when something is waiting to start on it with me.


What didn’t change

I still reach for my phone.

There are still nights where I sit down, feel the weight, and close the laptop.

The difference is that the weight is smaller now, because the next step is small and already written down. Starting doesn’t take any willpower. It takes four words.


The lesson, if there is one

In April I said this wasn’t a discipline problem; it was a design problem.

Six months and a few hundred commits later, I still think that’s right.

I didn’t become more motivated. I built a system where:

  • the next step is one sentence
  • decisions are made once
  • every session ends with something finished

It turns out that’s enough.