Why Habits Should Be Experiments

The problem with "I'll do X forever"

Most habit advice sounds the same: pick a habit, commit to it, and never break the chain. The implication is that failure means weakness. Miss a day? Start over. Fall off after two weeks? You clearly didn't want it enough.

This framing is toxic. It turns self-improvement into a pass/fail exam where the only grade that counts is perfection. And perfection, as a strategy, has a terrible track record.

What experiments look like

An experiment has a different DNA. It starts with a hypothesis: "I think stretching for 5 minutes after waking up will reduce my back pain." It runs for a fixed period — say, two weeks. And it ends with a review: did it work? Was it sustainable? Should I keep it, tweak it, or kill it?

This reframe changes everything. A missed day isn't a failure — it's a data point. Maybe Monday mornings are too hectic. Maybe 5 minutes is too long before coffee. The experiment tells you what to adjust instead of what to feel guilty about.

The weekly review is the secret weapon

Experiments only work if you actually review them. That's why Serein asks you to spend 10 minutes each Sunday looking at your data: which experiments thrived, which ones struggled, and what you want to change.

Over time, you build a portfolio of tiny experiments — each one refined, each one teaching you something about how you actually work. The habits that survive are the ones that genuinely fit your life, not the ones an influencer told you to try.

Small bets, big returns

Behavioral research consistently shows that small, variable actions beat large, rigid ones. A 2-minute walk is better than a planned hour at the gym you skip. A single page read before bed beats the "read 30 books this year" resolution gathering dust by February.

Experiments keep you in the game. They lower the stakes enough that you actually show up. And showing up, repeatedly, in small ways — that's the entire secret.

How to start

Pick one tiny behaviour you're curious about. Give it a title, set a schedule (start with 3 days a week, not 7), and run it for two weeks. Check in daily — did you do it or not? At the end of the two weeks, review. Keep it, tweak it, or kill it. Then start the next experiment.

That's it. No grand declarations. No streaks to protect. Just curiosity, data, and iteration.