The M-System

The M-System is a resource-allocation model for people whose interest list is long enough to embarrass them at parties. It has three parts. A stability pillar funds everything else and needs to run near-autonomously, roughly the point where you can solve most of its recurring problems without asking anyone. A growth pillar is the one deep pursuit you are actively building right now, timeboxed to a season rather than a lifetime, with an explicit graduation criterion: it either becomes the new stability pillar or it closes, on purpose, rather than by neglect. Between them sits the curiosity lab, a zero-commitment span where new interests get captured, archived, and occasionally cross-pollinate with whatever you are building, exactly the way a Zettelkasten or a personal knowledge base is supposed to work.

Why it is not just “focus better”

Specialist advice assumes one pillar and treats everything else as distraction. The M-System assumes the opposite starting condition, that wide curiosity is a fixed trait rather than a bug to suppress, and asks a resource-allocation question instead of a discipline question: given a fixed weekly budget of time and attention, how should it split across a known-good asset, one deliberate bet, and an open information-gathering budget. That is structurally the same question the explore-exploit tradeoff answers for a multi-armed bandit, and the same one portfolio theory answers for capital: do not put everything in one asset, and do not spread so thin that nothing compounds.

The failure modes it is built to avoid

Two patterns burn people who try to run wide without a system. Switching pillars before either one clears its own valley of abandonment produces a string of unfinished starts. Running an unbounded curiosity lab with no stability pillar underneath it produces the opposite failure, forty-seven browser tabs and nothing shipped. The system fixes both by making the pillar count and the switching rule explicit instead of implicit.