# Power laws and networks *A few things carry everything* --- Most distributions people assume are even are not. A handful of customers carry the revenue, a couple of contracts carry the margin, one or two people carry the function. Wherever outcomes are driven by things that feed on themselves - relationships, reputation, connections, reinvested winnings - the top few items end up mattering more than everything below them put together. The operating consequences run in both directions. Concentration is leverage when you know where it sits: the 80/20 cut, taken seriously, is an operating doctrine rather than a slogan - ITW ran its whole system on it, simplifying relentlessly around the few things that carried the value ([[The thin centre]] tells that story). And concentration is fragility when you don't: earnings that look diversified and rest on two contracts, a pipeline that is really three deals, a team that is really one person. The first question to ask of any healthy-looking aggregate is what's carrying it. Networks are where the maths comes alive, because each connection can make the others worth more. A participant joining a network raises its value for everyone already on it, which is why the moat of a network business is the connections rather than the software - [[Descartes]] is the worked case, two hundred thousand trading parties that no competitor can conjure. The same compounding works at street level: density is a local network, each added customer on a route cutting the cost of serving the rest, which is [[Cintas]]'s entire economics. Hubs, clusters and rich-get-richer dynamics fall out of the same machinery. The habit worth keeping: whenever something is described by its average - average customer, average deal, average return - ask for the distribution. If a power law is hiding in it, the average is the one number that's guaranteed to mislead. ---