Google is a search engine whose rise illustrates how collective judgment can be organized into a practical tool. Its PageRank algorithm treats links among Web pages as votes, weighing those votes by the importance of the pages that cast them. In the book’s account, this lets billions of locally maintained pages contribute to a fast, usually useful ranking: a form of collective intelligence built not by appointing one all-knowing judge, but by aggregating dispersed signals.
Rise as a Search Engine
Google began in 1998, when Yahoo! appeared to dominate Internet search and AltaVista or Lycos seemed likely alternatives if Yahoo! faltered. Within a couple of years, Google had become the default search engine for regular Internet users because it found relevant pages quickly and effectively. The change was not merely a matter of retrieving more material. Google’s distinctive achievement was to make the vast, distributed structure of the Web help identify which pages were most useful for a particular search.
The book makes that speed tangible through a series of searches. To find the full text of Bill Murray’s Caddyshack monologue, the author entered the terms “going for me” and “gunga”; among 695 results, a Penn State Soccer Club page posted by goalkeeper David Feist contained the whole riff. A search for “‘stock market’ challenger reaction” returned 2,370 pages and surfaced an article about the Mulherin paper as well as Mulherin’s own site, despite omitting his name from the query. Other searches found the needed Ramones lyrics. Together, the searches took about a minute and a half, while each individual result arrived in a fraction of a second.
The point of these examples is that Google did not simply search a small, curated directory. It surveyed billions of Web pages and repeatedly placed a page the user needed near the top. The familiar convenience of a search result conceals a difficult collective task: selecting a small number of relevant pages from an immense body of material without first asking a human expert to inspect it all.
PageRank and the Web’s Votes
The book identifies PageRank as the core of Google’s system, while noting that the company kept the details of its technology to itself. The algorithm was first defined by Google founders Sergey Brin and Lawrence Page in their 1998 paper, “The Anatomy of a Large-Scale Hypertextual Web Search Engine.” Rather than treat every page as an isolated document, PageRank uses the Web’s link structure as an organizational system.
A link from page A to page B counts as a vote by A for B. But the algorithm does not merely count links: it gives more weight to votes from pages that are themselves considered important. A page’s rank therefore depends both on how many other pages point to it and on the standing of those pages. In the book’s formulation, the result is a weighted average, not the simple average represented by the crowd estimating an ox’s weight.
This method turns an ordinary search into a distributed decision. The pages that receive the strongest votes tend to appear first, and the book observes that the top result—or the one immediately beneath it—is often the page with the most useful information. In one cited description, a search could survey three billion pages; the ranking compresses the judgments encoded in their links into a result delivered almost immediately. The Web’s structure becomes a means of making dispersed assessments legible at scale.
A Republic, Not a Perfect Democracy
The book stresses that PageRank does not give every page equal influence. Pages linked by many others carry greater weight, and already important pages can exert more influence when they vote for another page. Google is therefore described as a republic rather than a perfect democracy: the final ranking reflects unequal influence, not a straightforward one-page-one-vote count.
That inequality does not, in the book’s account, remove the crowd from the decision. The influential pages have their standing because smaller pages have linked to them. If the smaller sites repeatedly gave too much authority to irrelevant pages, the resulting searches would not remain accurate. The system’s central claim is thus conditional: a useful result at the top depends on the quality of the contributions throughout the network. The crowd rules through a weighted structure, and the structure can work only when the underlying pattern of links carries useful information.
Collective Intelligence and Its Limits
Google serves as one example of the book’s broader argument that people need not be individually expert for their combined judgments to be valuable. Human decision makers are limited in information, foresight, and willingness or ability to calculate; even so, aggregation can produce excellent collective results when the conditions are right. Google applies that idea to information retrieval: the separate links made by Web-page operators, each reflecting a local judgment about another page, contribute to a system-wide ranking.
The example also clarifies a general problem of decentralization. Dispersed participants can acquire specific knowledge independently, but valuable information may fail to travel beyond the part of the system where it originates. The book contrasts mere decentralization with aggregation: without a way to combine judgments, local knowledge remains fragmented. PageRank is presented as one way to turn such local signals into a collective whole, making them useful to people who never encounter the individual decisions behind the ranking.
The system is not offered as proof that every crowd or every market will be wise. Its performance depends on how contributions are combined and on whether the links provide a sound basis for ranking. The example’s force lies in that distinction: collective intelligence is not automatic, and neither unanimity nor equal influence is required. Google’s search results embody a designed method for aggregating unequal, distributed judgments—and demonstrate how that method can make a huge network useful in the time it takes to run a search.
