The Google Guys by Richard L. Brandt Read Free Book Online

1998, it was distinguished by one big unique attribute: it worked.
    At its core is the PageRank system, invented by Larry (and named after him) while he was working on his Ph.D. It takes advantage of the unique properties of the Web—the network of links that makes its name so apt.
    Garcia-Molina recalls how it all started. He was Page’s adviser, and one day in 1995 his student came into his office to show him a neat trick he had discovered. The AltaVista search engine not only collected key words from sites, but could also show what other sites linked to them. AltaVista did not exploit this link information in the way Google would, but that day in Garcia-Molina’s office, Page suggested it would be a good way to rank the importance of sites.
    At first, it was just a game. “We had lots of fun that day seeing which computer science pages were most popular among the different universities,” recalls Garcia-Molina. They were pleased to find that Stanford’s database group, for example, drew more links than a similar department at rival University of Wisconsin.
    Larry had his own idea about links. He told Garcia-Molina, “If this is so important to us, why not make it part of the search process?”
    Larry’s idea was inspired by his scientific background. It was well known in the scientific community that when a researcher cites your paper in his own, it lends yours more credibility. The more citations you get, the more important your paper is perceived by the research community. This idea was codified in the “Science Citation Index” created in 1960 by Eugene Garfield, founder of the Institute for Scientific Information. Larry reasoned that Web links were analogous to scientific citations, and those with the most links probably were the most popular and would prove most useful to searchers. Those were the sites that should be listed first in the search results. He then began creating his own software for analyzing links between sites.
    This required some tricky programming. Not only did the system count links to a particular site, it went a step further by determining the importance of the sites doing the linking. This was done by counting the links to the sites one link back. This increased the complexity of the analysis enormously; in order to calculate relevance, PageRank also had to track the links two steps back and correlate that data with the key words. Larry first called the system BackRub, because of its property of tracing links backward. But he later settled on the more sophisticated PageRank, a double entendre with his surname.
    Sergey also fell into search engine research by chance. As a math and computer science major in the doctoral program at Stanford, he was working on a research project within the database group. In 1995, he and Brian Lent decided to try their minds on another computer science discipline called “associative data mining.” This is the process of finding pieces of information that commonly occur together. Retailers use it to search through their sales records and determine whether different items are frequently bought together by customers. Data mining was, however, a new field for computer science. It required archiving masses of Web data, so Sergey had to write a “crawler” program—software that visits Web sites, summarizes their content, and stores the data in a central location accessible to graduate students and search companies. Other search engines already had their own crawlers.
    Sergey is a terrific programmer and engineer. His data mining work, using the Internet, involved parsing through huge amounts of data. “He did it on a scale that others would not have even contemplated,” says Jeffrey Ullman, Sergey’s adviser. (Sergey’s paper outlining the Google search engine was itself cited in another scientific paper, “Quality of Service and the Electronic Newspaper: The Etel Solution.”) 2
    Sergey