What I Would Do Differently If I Started This Year Over

Hindsight is a cheap teacher. Even so, naming the things I would do differently is useful for next year and for anyone who is about to start their own version of this journey. Here is my honest list of what I would change if I could restart 2025 from January.

Thing 1: Start the Daily Writing Habit Earlier

I started daily writing in April. I should have started in January. Three extra months of compounding would have produced roughly 90 more articles and three more months of sharpening the thinking. The return on daily writing is non-linear. Starting earlier matters more than writing more later.

Thing 2: Define the Category Taxonomy First

I invented categories as I went. Then I spent a full day retrofitting a taxonomy to 50 articles. Every new project should start with the taxonomy definition, not end with it.

Thing 3: Move to Fixed Fee Sooner

I ran hourly contracts through Q1 out of habit. Fixed fee was a better fit for most of that work. The switch happened in Q2. It should have happened in January.

Thing 4: Invest in Observability on Day One

I launched two services this year with minimal logging. Both later needed observability retrofits. Retrofitting logs and metrics is painful. Adding them from day one is easy. Never skip day-one observability.

# What I should have had from day one
log_event('api.request', request_id=req_id, method=method, path=path)

Thing 5: Say No Faster

I took on one project that did not fit my shape. It consumed disproportionate energy and taught me the same lesson I had been taught before: say no to bad-fit projects faster.

Thing 6: Build the Artifact-Script Habit Sooner

Hand-posting articles, hand-running migrations, hand-creating test data: all slower than a quick Python script. I should have reached for the script reflex earlier. Any task I will do more than three times is worth a script.

Thing 7: Tag Before Every Multi-File Edit

I lost hours to agent-assisted edits that went sideways. A simple git tag pre-session before each multi-file session costs nothing and enables zero-cost rollback. This habit saved me later in the year. I should have had it earlier.

Thing 8: Spend More Time on Pipeline Health

The months I was fully booked felt great. The month after a fully-booked period, when prospects were thin, did not feel great. Pipeline is a constant, not an occasional. I would invest more consistent hours in pipeline work across the whole year.

Thing 9: Keep a Running Decision Log

I made dozens of technical decisions this year. A running decision log (one line per decision, dated, with context) would have made this year-end review much easier. Next year I keep the log.

The Meta-Lesson

Most of these are about earlier adoption of habits I eventually adopted anyway. Habits have long payback periods. The best time to start was a year ago. The second-best time is today.

For habit-formation frameworks, see Atomic Habits as a practical reference.