Linked: How Everything Is Connected to Everything Else and What It Means for Business, Science, and Everyday Life

Overview

Linked is a foundational work in network science by Albert-László Barabási, originally published in 2002. While the field has advanced significantly since its publication, making some of its discussions quite out of date, the book remains an interesting look into the history and early development of the internet and modern network theory.

Key Concepts

From Random to Scale-Free — A Taxonomy of Networks

• Random networks (Erdős–Rényi): The starting point of network theory; each pair of nodes is connected with equal probability $p$. The resulting degree distribution is Poisson-like — most nodes have roughly the same number of links, and very highly connected nodes are exponentially rare. Good as a null model but poor at describing real-world networks
• Small-world networks (Watts–Strogatz, 1998): Real social networks exhibit high clustering (your friends tend to know each other) yet short path lengths (any two people are connected by ~6 steps). Watts and Strogatz showed that rewiring just a few edges in a regular lattice produces this small-world property — a bridge between order and randomness
• Scale-free networks (Barabási–Albert, 1999): Barabási’s central contribution. Many real-world networks (the internet, citation networks, protein interaction networks, airline routes) have power-law degree distributions: $P(k)\sim k^{-\gamma }$ with $\gamma$ typically between 2 and 3. This means that while most nodes have few connections, a small number of hubs have enormously many — there is no characteristic scale, hence “scale-free”
  • Preferential attachment: The mechanism generating scale-free networks: new nodes preferentially connect to already well-connected nodes (“the rich get richer”). This is a formalisation of the Matthew effect / cumulative advantage observed in many systems

Hubs, Robustness, and Vulnerability

• Robustness to random failure: Scale-free networks are remarkably tolerant of random node removal — because most nodes are low-degree, removing them at random rarely disconnects the network; the hubs hold the structure together
  • Vulnerability to targeted attack: However, if hubs are specifically targeted (e.g., by a deliberate attack on the most connected routers, or by a disease targeting superspreaders), the network fragments rapidly — this is the Achilles’ heel of scale-free topology
• Cascading failures: In tightly coupled networks (power grids, financial systems), the failure of one hub can overload its neighbours, triggering a cascade that propagates far beyond the initial failure — Barabási discusses real-world examples including the 2003 North American blackout

Network Dynamics and Applications

• Epidemic spreading: On scale-free networks, the epidemic threshold approaches zero — even weakly transmissible diseases can persist because hubs act as superspreaders. This has direct implications for public-health strategy (vaccinate or quarantine the hubs)
• Biological networks: Metabolic networks, gene regulatory networks, and protein-protein interaction networks exhibit scale-free or heavy-tailed degree distributions; this architecture provides modularity (specialised functional clusters) and evolvability while maintaining global integration
  • Network medicine: Diseases can be understood as perturbations in network topology — comorbidities arise when disease modules overlap in the interactome; drug targets can be identified by their network position rather than just their molecular function
• The internet and information networks: The World Wide Web’s link structure is scale-free; Google’s PageRank algorithm essentially exploits hub structure by ranking pages according to the number and quality of incoming links

Limitations and Legacy

• Criticisms of universality: Since the book’s publication, researchers have shown that many networks previously claimed to be scale-free do not pass rigorous statistical tests for power-law distributions (Clauset, Shalizi & Newman, 2009); the exponent $\gamma$ and the range of the power law vary widely. The “scale-free” label was applied too broadly
• Beyond topology: Modern network science has moved beyond static degree distributions to study dynamics on networks (diffusion, synchronisation, game theory), temporal networks (links that change over time), and multilayer networks (interacting layers of different connection types) — extensions that Barabási’s framework laid the groundwork for

Personal Reflection

[To be added]

Related Books

• Entangled Life - Mycorrhizal networks are a biological instantiation of scale-free network topology
• The Information - Gleick covers the signals that flow through the networks Barabási maps
• Chaos - Nonlinear dynamics complements network dynamics — together they cover complexity science

---

Parent: Books