# Linked **Albert-László Barabási**  --- _Networks are not random. They are shaped by growth and preferential attachment, and that shape determines almost everything._ The architecture of connections shapes what's possible. Why do viruses spread unstoppably? Why do economies cascade into failure? Why do a few players dominate every networked market? The answer in each case is the same: the structure of the network isn't uniform. It's scale-free, and that property is not a curiosity. It's the mechanism. Most real networks follow power-law distributions: a few nodes carry enormous numbers of links, while most carry very few. These hubs aren't accidents. They emerge from two forces operating together. Networks grow by adding new nodes. New nodes attach preferentially to already well-connected nodes. The rich get richer, exponentially, and the resulting structure looks nothing like a bell curve. It has fat tails. Hubs. Extreme concentration. --- **Robustness and fragility coexist, and that coexistence is the defining property of scale-free networks.** Remove random nodes and the system barely notices; there are too many small nodes to care about. Remove the hubs and the system collapses quickly. This makes targeted intervention wildly asymmetric from random removal, a distinction that matters in epidemiology, in financial regulation, and in competitive strategy. The implication for strategy: in networked systems, winning means becoming a hub. Microsoft, Google, Amazon don't simply have more customers; they have exponentially more connections, creating [[Switching costs]] so deep that dislodging them becomes a question of topology rather than product quality. You can't compete your way out of a hub's network position from the periphery. The strategy has to be about establishing early connections at the right moment, because preferential attachment rewards the node that already has the links. --- **Weak ties matter more than strong ones.** Granovetter's insight, confirmed by network science, is that strong ties cluster together in dense cliques where everyone already knows everything. Weak ties bridge separate clusters, carrying novel information between groups that don't otherwise communicate. Jobs, ideas, and fads spread through weak ties, not strong ones. Clustering is a generic property of networks, not a social quirk. In scale-free networks, epidemic thresholds effectively vanish. Traditional diffusion models assume that a pathogen has to reach a certain transmissibility to spread at all. Scale-free structure eliminates that threshold: even a weakly contagious virus persists, because hubs become infected early through their many connections and then infect hundreds of others. The only effective intervention is targeting the hubs directly. Randomly vaccinating the population is far less efficient than vaccinating the most connected individuals. Markets function less as free-floating exchanges and more as directed networks of firms, suppliers, and customers. Relationships prioritise long-term reliance over one-off transactions. In a network economy, buyers and suppliers become [[Complements]], each making the other more valuable, and cascading failures become the natural consequence of that interdependence. No institution can function in isolation; that's not a design flaw but a structural property of interconnection. Nodes always compete for connections, because links represent survival. Some networks can undergo what Barabási calls Bose-Einstein condensation, where one node takes all. Microsoft, at the height of its Windows dominance, carried that signature: not just a large hub, but a condensate that made any competing node nearly unreachable. ---