a16z Partner Martin Casado Reviews the Academic and Commercial Evolution of Network Effects
Andreessen Horowitz partner Martin Casado reviews the academic and commercial evolution of network effects: In 1974, Rohlfs at Bell Labs proposed a simple model of communication networks based on users and their interactions. In 1985, Katz and Shapiro attempted to extend this, but he believed that the extension lost the implicit concept of the network itself, being almost tautological and useless. In 1994, Liebowitz and Margolis criticized Katz and Shapiro's work and began to better classify it; their paper is worth reading. In 2000, Rochet and Tirole separately proposed two-sided markets as an independent category. Meanwhile, commercial media intercepted discussions at different nodes, forming various working definitions, ultimately constructing a Babel tower of commercial terminology. Casado added that Brian Arthur's work on positive feedback loops predates its use in network effects (also predating Katz's paper), and was later adopted as a causal force of positive feedback, with its scope seemingly narrowing further. He pointed out that there is currently almost no consensus, and network effects are often misused; the true existence of super-linear interactions among users and advantages of leaders over followers is rare in the tech field. The concept has evolved from specific communication network models, through economic promotion and criticism, to the separation of two-sided markets, ultimately fragmenting in commercial discourse. Funding and entrepreneurial narratives rely on these different definitions to assess moats, and vague definitions increase the risk of misjudgment. Source: Public Information
ABAB AI Insight
The concept of network effects originated from Rohlfs' modeling of interdependent demand for communication services in 1974, focusing on user interactions. In the mid-1980s, Katz and Shapiro extended it to compatible systems and hardware/software networks, influencing antitrust discussions. In the 1990s, Liebowitz and Margolis proposed classifications and criticisms, and in 2000, Rochet and Tirole separated two-sided markets, leading to a divergence in academic paths. On the capital side, commercial media and entrepreneurial discourse intercepted these academic discussions at different stages, forming various working definitions used to assess the moats of platforms, social, and data businesses. The strategic motivation is to use the "network effects" label to explain the advantages of leaders, but inconsistent definitions lead to confusion in judgment, with investment and competitive analysis relying on vague terminology. Similar cases include the phenomenon of scale effects, flywheels, and data advantages being misused as network effects, as well as the independent application of two-sided market theory in platform economics. The current concept is in a phase of diffusion from specific network models to broad commercial terminology, lacking consensus. Essentially, this involves conceptual reconstruction and the transfer of pricing power. The original network model based on user interactions loses its core after being generalized, with the mechanism being that academic simplification and commercial interception jointly create multiple definitions, making "network effects" a flexibly usable moat label rather than a verifiable structural characteristic. ABAB News · Laws of Cognition
- When definitions scatter, moats become slogans.
- True network effects are rare; most are scale effects.
- The Babel tower of terminology grows taller, with less consensus.