PageRank

PageRank is the algorithm at the core of Google’s search system, turning the web’s vast network of links into a way to rank pages for relevance. Rather than rely on a single editor or a simple count of links, it treats links as votes and gives greater weight to votes from pages that are themselves important. In the book’s account, the method exemplifies how a distributed crowd can produce a useful collective judgment—provided the system can aggregate its members’ contributions effectively.

Definition and Operation

First defined by Google founders Sergey Brin and Lawrence Page in their 1998 paper “The Anatomy of a Large-Scale Hypertextual Web Search Engine,” PageRank is a calculating method for assessing which web pages are most relevant to a particular search. Google keeps the details of its technology to itself, but the book identifies PageRank as the system’s core.

Its basic operation interprets a link from page A to page B as a vote by A for B. A page’s importance depends not only on how many votes it receives, but also on the importance of the pages casting them. A link from an important page therefore carries more weight than one from a page with less influence. Ranking is consequently not a flat tally: it is a weighted aggregation of judgments embedded in the web’s link structure.

The Web as a Voting System

The algorithm makes use of what the book calls the web’s uniquely democratic characteristic: pages can link to other pages, creating a distributed structure of endorsements. In a search, PageRank uses that structure to ask the web as a whole which pages are most useful. The page receiving the most weight comes first in the results, and the page immediately below it may also prove to be among the most useful.

This vote is not an equal-vote democracy. Large or already important sites have more influence because their links count more heavily. Yet their influence is not independent of smaller sites: those sites’ links help establish the larger sites’ importance in the first place. The book’s central qualification is that the crowd still rules, but a system can be smart at the top only if its judgments are sound throughout the network.

Aggregation and Collective Judgment

PageRank illustrates the broader conditions the book associates with wise crowds: diversity of opinion, independence, decentralization, and aggregation. A mechanism must gather dispersed judgments into a collective result; without aggregation, private information remains separate. PageRank supplies such a mechanism by converting links into weighted votes and combining them into an ordered list.

The book compares this weighted result with stock prices and football point spreads, which also combine judgments without simply averaging every contribution equally. It contrasts PageRank’s structure with a simple average, such as the average estimate in an ox-weighting contest. In each case, the method of aggregation affects how individual contributions shape the group’s verdict.

Google’s Search Results

The book presents PageRank as the basis for Google’s ability to search billions of web pages and return a useful result quickly. In one reported search sequence, the searches together took about a minute and a half; a typical individual Google search took a fraction of a second. Google began in 1998, when Yahoo! appeared dominant and AltaVista or Lycos seemed likely alternatives if it faltered. Within a couple of years, Google had become the default search engine for regular internet users, in the account given, because it found relevant pages quickly.

The algorithm’s significance in this example lies not simply in speed or scale, but in how those qualities are joined: millions of pages contribute signals through their links, and PageRank turns those signals into a ranking. The resulting order is not guaranteed to be perfect, but the book argues that the first result, or the one just beneath it, more often than not is in fact the most useful page. PageRank thus serves as a concrete example of collective judgment made operational through an algorithm.