- Tram Ho
First meeting Larry Page in the summer of 1995, Sergey Brin was a sophomore in the Computer Science department at Stanford University. Brin volunteered to be a guide for first-year students who had entered but didn’t know what to study. His duties include leading a tour of San Francisco. Page, a University of Michigan engineering alumnus, was on Brin’s team.
On that trip, the two clashed endlessly, arguing over the merits of different approaches to urban planning. “Sergey was quite a social person; he loved meeting people,” recalls Page, contrasting that quality with his caution. He has a really strong opinion on everything, and I guess I do too.”
“We both found each other quite obnoxious,” said Brin. “Obviously, though, we spend a lot of time talking.”
When Page showed up at Stanford a few months later, he selected Terry Winograd – a pioneer in human-computer interaction – as his mentor. Not long after, he began searching for a topic for his doctoral thesis. A thesis can shape a person’s entire career. He came up with about 10 ideas but was most interested in the World Wide Web.
Despite the fact that Stanford alumni became rich by starting Internet companies, Page found the web to be attractive only because of its mathematical features. Each computer is a “node”, each link on a Web page is a connection between those “nodes” – a classic graph structure. Page hypothesized that the World Wide Web could be the largest graph ever and is growing at breakneck speed. Many useful insights hidden deep are waiting for students to discover. Mr. Winograd concurred. Page began to consider the topology of the Web.
Back Rubs and PageRank
During his research, Page noticed that when you look at a web page, you can’t tell which page they’re linking to. He was bothered by that because it was obvious that it would be more helpful to know this.
Therefore, he researched on backlinks (links returned from other websites, blogs, forums, social networks to websites) and named the project BackRub. At that time, the Web consisted of about 10 million documents with countless links between them. The computing resources required to collect such large amounts of data go far beyond the usual confines of a student project. Page started building its crawler.
The complexity and scale of the idea intrigued Brin. He found the premise behind BackRub really compelling. “I talked to many research groups in the school and this was the most interesting project because it both solved the problem of the Web – representing human knowledge, and because I loved Larry,” recalls Brin.
In March 1996, Page pointed his self-developed crawler to a single website—his homepage at Stanford—and let it crawl from there.
Page theorized that the structure of the Web graph would not only reveal who was linking to whom, but more importantly, the importance of who linked to whom, based on various properties of the web. The Web site is performing the linking. Inspired by citation analysis, Page realized the number of raw links to a page would be a useful guide to that page’s rankings. He also found that each link needed to rank separately, based on the link count of its parent page. However, such an approach poses a recursive and convoluted mathematical challenge: you must not only count the links of a particular page, but also the links attached to the links. The math becomes extremely complicated.
Fortunately, Page had Brin as his partner, who had a natural gift for math. Brin is the son of a NASA scientist and a University of Maryland math professor. His family moved from Russia to America when he was 6 years old. During high school, Brin was considered a math genius. He enrolled at Stanford to develop talent.
Together, Page and Brin created a ranking system that rewards incoming links from important sources and penalizes backlinks. For example, Web sites associated with IBM.com may be owned by an industry partner, or by a teenage programmer. From human observation, the business partner is more important, but how does the algorithm understand that?
Page and Brin’s breakthrough was an algorithm – PageRank – that evaluates both factors: the number of links leading to a particular page and the number of links to each linking Web site. Going back to the example above, assume that only a few websites link to the teen programmer’s site. In contrast, thousands of websites link to Intel, and these sites also have thousands of sites linking back on average. PageRank will rank teen programmers not as important as Intel’s, at least in relation to IBM.
That is a simple view. More popular sites rank higher and vice versa.
While tracking the results, Brin and Page realized their data could impact Internet searches. Page and Brin also found that BackRub’s results outperformed existing search engines like AltaVista and Excite, which often returned irrelevant listings. “They just looked at the text without considering other cues,” says Page.
Not only that, the tool scales to the size of the web. Since PageRank works by analyzing links, the bigger the web, the better the tool. In fact, that inspired the two founders to name the tool Google – akin to googol, the phrase for 100 zeros following the number 1. They launched the first version of Google on the Stanford website in August 1996. To improve the service, they needed a large amount of computing resources, so they borrowed hard drives from the school lab. Page’s dorm room turned into a headquarters, a programming center, filled with machines.
The project became a legend in Stanford. At one point, the BackRub crawler consumed almost half of the school’s bandwidth. Even in the fall of 1996, the project almost crashed the Internet connection here, but fortunately, no one complained too much.
Open a company
While Brin and Page continued to experiment, BackRub and Google resonated inside and outside of Stanford. One of the buzz was Cornell University professor Jon Kleinberg, who is researching biometrics and search technology at IBM’s Almaden center. His approach to Web ranking is perhaps second only to PageRank. In the summer of 1997, he visited Page at Stanford, and the two exchanged research. He encouraged Page to publish an academic report on PageRank.
But, Page said he was worried that if he went public, someone would steal his idea. With PageRank, he felt like he possessed a secret technique. On the other hand, Page and Brin weren’t sure if they wanted to start and run the company. During Page’s first year at Stanford, his father died, and he completed his PhD in honor of his father.
Brin recalls talking to the consultant, who advised, “Look, if Google succeeds, great. If not, I can go back to graduate school and finish my thesis.” “I said, ‘Yes, why not? I’ll try it.”
And so, the genius duo Brin and Page founded Google on September 27, 1998 in Menlo Park, California.
Source : Genk