The two men who would build Google spent their first day together arguing. It was the summer of 1995; Larry Page, 22, was at Stanford for a program for accepted graduate students, and Sergey Brin, already two years into the computer science PhD program, had been assigned to show him and the other newcomers around San Francisco. The subject, by several accounts, was urban planning, though it could have been anything. Each came away unimpressed.
“I thought he was pretty obnoxious,” Page said later. Brin’s version was more generous only in its symmetry: “We both found each other obnoxious.” They were two strong, fast, fiercely certain intellects, and the friction never quite went away. It just turned out to be productive.
A thesis about the shape of the web
Page enrolled that fall and went looking for a dissertation topic. He landed on a question about the World Wide Web — not its content but its structure, the way pages pointed at one another through links. Treated as a graph, with each link an edge, the whole web could be studied as one enormous mathematical object. His advisor, Terry Winograd, told him it was a good idea and to chase it.
The insight underneath was simple and, in retrospect, obvious. A link from one page to another is a kind of vote. A page that many pages link to is probably important; a link from an already-important page should count for more than a link from an obscure one. Search engines at the time ranked results mostly by how often the search words appeared on a page, a method that was easy to fool and frequently useless. Page and Brin — Brin, the mathematician, having been drawn into the problem — built a system that ranked pages by the structure of links pointing at them. They called the ranking method PageRank — a name that reads equally well as a description of what it did, ranking web pages, and as a nod to Page’s surname.
To gather the raw material, Page wrote a web crawler. In March 1996 he set it loose with a single starting point: his own Stanford home page. From there it followed links outward, and outward again, pulling in the back-link data the algorithm needed.
BackRub
They named the project BackRub, for the back links it analyzed. It ran on Stanford’s own network — later, once the name changed, at addresses like google.stanford.edu — and it worked alarmingly well. The first version went up on the Stanford site in August 1996 and reportedly consumed close to half of the university’s entire network bandwidth. Word spread on campus. People used it because it found what they were looking for, which the competition often did not.
In 1998 the two laid out the system in a paper titled “The Anatomy of a Large-Scale Hypertextual Web Search Engine.” In it they explained the name they had by then adopted: they chose “Google,” they wrote, because it was a common spelling of googol — the number written as a one followed by a hundred zeros — and it fit their goal of building a search engine on a very large scale.
That is the tidy official version. The real story is that someone couldn’t spell.
The typo
In September 1997, Page sat in a Stanford office with a fellow graduate student named Sean Anderson, working at a whiteboard, trying to name the thing. They wanted a word that suggested an immense quantity of data. Anderson said googolplex — a one followed by a googol of zeros, a number with no practical use except as a thing too large to imagine. Page shortened it aloud to googol.
Anderson, seated at a terminal, went to check whether the domain was available. He typed it from memory and got it wrong: g-o-o-g-l-e. The misspelled name was free. Page liked it better than the correct one. On September 15, 1997, he registered google.com.
A check made out to a company that did not exist
By 1998 the operation had outgrown the dorm. Through a Stanford professor, David Cheriton, the founders got an audience with Andy Bechtolsheim, a co-founder of Sun Microsystems and a man who recognized a good machine when he saw one. They demonstrated the search engine on Cheriton’s porch. Bechtolsheim grasped the economics in about the time it took to run the numbers in his head — sponsored links, a few cents a click, a million hits a day. He didn’t need the rest of the pitch.
He went to his car and came back with a check for one hundred thousand dollars, made out to Google Inc. Cheriton, watching his own porch turn into a funding round, matched it with a second check for the same amount. There was a complication: Google Inc. did not exist. The founders had not yet incorporated, and they could not deposit the check until they did. It sat for about two weeks while they built a company around it. On September 4, 1998, Google was formally incorporated, and that month the operation moved into a garage in Menlo Park, rented for $1,700 a month from Susan Wojcicki, who was carrying a new mortgage and would later run YouTube.
The two men who had spent their first day together arguing now had a company, a name born from a spelling error, and a check they’d had to build a corporation to cash. Bechtolsheim’s hundred thousand dollars, among the first money anyone put into Google, would in time be worth well over a billion.