DEV Community

Nora Vale
Nora Vale

Posted on

Running a One-Month Self-Experiment With a Two-Minute Daily Game

A two-minute daily puzzle is a nearly ideal self-experiment: fixed rule set, one trial per day, reproducible input, cheap to record. I ran a month of it with a three-line notebook, and the design turned out to be more interesting than the results.

The notebook was: plan before the first move, the rung where the plan died, and the word I did not know. Three lines, about forty-five seconds, done before opening any answer page.

The design choices that made the data usable:

Record before reveal. The whole experiment depends on writing the plan before checking anything. If you check first, you record a hindsight rationalisation instead of a plan.

One sentence per field. Long entries kill daily habits. If the plan needs two sentences, it was not a plan yet.

No scores. I deliberately did not record move counts or times, because those are the numbers you can already feel. The interesting variables are the ones you cannot feel: which rung the plan broke, and which word you were missing.

After thirty entries two patterns showed up. The plan broke at rung two far more often than at the endgame, which contradicted my intuition. And five words accounted for more than half of the missing-word entries, which is small enough to be a curriculum.

The control is the tricky part. I changed only one thing, the notebook, and kept the puzzle constant, which means any improvement is confounded with the measurement effect. That is fine. A measurement effect that changes behaviour is a real intervention, and it is much cheaper than a genuine training regimen.

If you want a subject cheap enough to study properly, a daily game works because the trial length is bounded and the rules never change. I ran this on the four-letter ladder at Poople Unlimited. The general recipe transfers to any daily habit that has a clear start, a clear end, and an outcome you cannot fully control.

Top comments (0)