Break One Arrow

Perturb one link while holding everything else steady so you learn which edge actually matters.

Your afternoon headaches could be the new monitor, the skipped lunches, the extra espresso, or the tense weekly call that precedes them. Last week you changed all four at once, felt better, and learned nothing, because now you cannot tell which change mattered, and you are stuck maintaining all four forever.

Break One Arrow is a small experiment: perturb one link in your suspected causal picture while holding everything else steady, so you learn which connection actually matters. The cue is a situation with multiple plausible causes where you are about to change several things at once, or already did. The operation: draw your suspected arrows (espresso leads to headache, monitor leads to headache); pick the one arrow that is cheapest and safest to disturb; change only that, for long enough to see an effect given honest delays; then observe whether the outcome shifts. If headaches persist without espresso, that arrow loses credit and you test the next. What changes is that you earn causal knowledge you can act on permanently, instead of a superstition bundle. Do not run one-arrow tests when the system is in crisis or the perturbation is risky; in emergencies you change everything first and assign credit later.