In the trillion club
On January 16, 2020 (one month ago) Google joined the club of technology companies that have surpassed the trillion dollar market cap. In mid-December 2019, the two founders of Google, Sergey Brin and Larry Page, decided to leave all operational positions at Alphabet, the company with a Montessori name that currently controls Google.
One might think that the exit of the founders in favor of more business-minded executives helped to make Google's share soar. The Stock Exchange has always rewarded Google, but always with a certain concern for the initiatives and interests of Brin and Page. The famous monshoots that sometimes overflowed into unknown territories and barren of resources. A bit like Fitzcarraldo.
In fact, the two founders continue to control the company with their privileged voting rights on the board of directors.
Perhaps it should be so, even if the matter may make investors itchy. Brin and Page invented Google from scratch. They created the technology that made their fortune. They have turned it into an operational form of AI that is shared and benefits us all. Maybe the same thing will happen in the future!
With incredible perseverance and humility, they turned a graduate project into a monstrous service and business that radically changed at least three industries, including the trillion-dollar advertising industry.
Steven Levy's story
In this story Steven Levy narrates, with accuracy and humor, the birth of this extraordinary project. Above all, he narrates how he took shape from two students at Stanford, perhaps the most important incubator in the world. He also shows us how important perseverance is as much as genius and inventiveness.
Precisely the latter together with luck, initiative, sacrifice and correctness in behavior were the mixture of the big bang from which, from scratch in a university dormitory, the creation of Google took place.
Steven Levy is the journalist and writer who has covered for over thirty years, for "Wired" and other magazines ("Newsweek", "The New York Times Magazine", "The New Yorker", "Premiere", "Rolling Stone" etc.) the companies, people and technologies of Silicon Valley, the engine of the technological revolution.
In 1984 he published one of the first books explaining the hacker ethic with great clarity, Hackers. The heroes of the information revolution (also translated into Italian). He has also published books on Apple, artificial intelligence and Google itself (In the plex). Many of these are also found in the Italian language, including the latter
Have fun and above all inspire yourself!
Stanford, summer 1995
In the summer of 1995, just before entering graduate school at Stanford, XNUMX-year-old Larry Page attended an admitted student program that included a tour of San Francisco. His mentor was a rollerblading, trapeze-loving, math-minded graduate student in computer science of the same age who had attended Stanford University for two years.
"I found him rather obnoxious," Page would later say of his guide, Sergey Brin.
As the son of computer scientists, Page grew up in Lansing, Michigan, with computer language as his first language, and later earned a computer science degree from the University of Michigan. He wasn't the sociable type—people who spoke to him often wondered if there was a hint of Asperger's in him—he could get others on edge simply by keeping quiet.
Yet to those who knew him, Page's intelligence and imagination were evident, as was his ambition. In 1995, Stanford was not only the best place to major in cutting-edge computer science, but, because of the internet boom, it was also the ambition capital of the world.
Page chose to work with the Human-Computer Interaction Group of Stanford's computer science department. His tutor, Terry Winograd, describes Page's outlandish ideas as more "science fiction than computer science."
Both were stubborn when it came to sticking to their beliefs.
Larry&Sergey
Brin, for starters, was born in Russia and was four years old when his family immigrated to the United States. His English still retained a Cyrillic flavor, and his speech was peppered with anachronistic Old World expressions, such as the use of "whatnot" when peers said "stuff like that." He had come to Stanford at nineteen, after passing through the University of Maryland, where his father taught. He was one of the youngest students to start a PhD at Stanford.
Sergey was an eccentric boy who zipped through the halls of Stanford on roller skates.
But the professors understood that behind that extravagance was a formidable mathematical mind. Shortly after arriving at Stanford, he passed all the tests necessary for his doctoral degree and was free to take classes until he found one that would allow him to do a thesis. He supplemented his studies by practicing sports such as swimming, gymnastics and sailing.
Despite their temperamental differences, in some ways Page and Brin were identical. Both felt at home in the meritocratic environment of academia, where brains were worth more than anything else. Both had a deep understanding of the value of an ultra-connected world, which they could avail themselves as Computer Science students. A world which, moreover, was beginning to spread to the rest of society. Both shared an unwavering faith in the primacy of data.
When Page settled at Stanford, he became best friends with Brin, to the point where people thought they formed one identity: Larry&Sergey.
Looking for a thesis
What helped cement their relationship was the fact that at the time they met neither of them had decided on their dissertation topic.
Brin had a grant from the National Science Foundation and they weren't solely focused on data mining. However, he helped start a research group called MIDAS, which stood for Mining Data at Stanford.
In a resume posted to the Stanford website in 1995, he spoke of "a new project" for generating personalized movie ratings. Another project she worked on with tutor Hector Garcia-Molina and another student was a system that detected copyright infringement by automating searches for duplicate documents.
Page was also looking for a topic for his dissertation. She had presented an interesting project to his tutor, Terry Winograd, to be carried out in collaboration with Brin. The idea was to create a system where people could annotate and comment on websites.
But the more Page thought about it, the more he saw its difficulties. On big sites, there would probably be a lot of people who wanted to leave a comment. How to figure out who should comment or the comment that he would be seen first? There was a need for a rating system.
It was out of the question to have a human determine the ratings: it would have been impractical and unreliable. Only algorithms — well designed, efficiently executed, and based on solid data — could have delivered reliable results.
So the problem became finding the right data to determine which comments were more reliable or more interesting than others. Page realized that this type of project already existed and that no one else was actually using it.
He said to Brin, "Why don't we use web links to do this?"
BackRub
Page, a child of academia, understood that web links were like citations in a scientific article. He was widely recognized that really important articles could be identified without needing to read them. It was enough simply to realize how many other articles cited them in their footnotes and bibliographies.
Page believed that this principle could also be applied to web pages. But it wasn't going to be easy to press the right data. Web pages made their outbound links transparent: in the code, there were easily identifiable markers for destinations that could be navigated to with a mouse click from the page.
But it wasn't all that obvious that it linked to that page at all. To find out, you had to collect a database of links that all connected to a certain page. Then it went backwards.
That's why Page called his system BackRub. He once told a reporter:
Early versions of hypertext had a tragic flaw: you couldn't run i backlink. BackRub it was meant to nullify this problem.
Winograd thought it was a great idea for a project, but not an easy one. To get it right, he told Page, he really would have to capture the entire link structure of the World Wide Web. Page said that, indeed, he would. He figured it would take him a week or something. “And of course,” he recalled later, “it took, like, years.”
But Page and Brin went on the attack. Every two weeks Page would go to Garcia-Molina's office to ask for records and other equipment. "Okay," Garcia-Molina said, "this is a big project, but you have to give me a budget." He asked Page to pick a number, to know how much of the web he should scan and to estimate how many disks it would take.
"I want to scan the whole web," Page said.
Python
Since Page didn't feel like a programmer capable of taking on this challenge, he asked a friend to help him. Scott Hassan was a full-time research assistant at Stanford. Hassan was also good friends with Brin, whom he had met at a game of ultimate Frisbee his first week at Stanford.
When Hassan took over the job, he commented, "Page's program had so many bugs that it wasn't much fun."
Hassan rewrote the program in Python, a more flexible language that was becoming popular with web developers. This program would have scoured the web for data. The program visited a web page, found all the links and logged them sequentially in a queue. Then she went to check if he had visited those pages before. If not, it placed the link in a list of pages he should visit in the future. Then he repeated the process. Since Page was unfamiliar with Python, Hassan became a member of the team.
Brin, theenfant prodige of mathematics, he took charge of the great enterprise of finding the mathematics necessary to unravel the skein of links collected by Hassan's procedure during the monstrous session of processing the pages of a web in exponential growth.
While the small team was on its way somewhere, its final destination wasn't entirely clear. Remember Hassan:
“Larry didn't have a plan. In research, you explore something and see what makes the most sense."
The idea of the search engine was born
In March 1996, they started a test starting with a single page, the homepage of the Stanford computer science department. The program located the links on the page and scattered to all the sites that pointed to Stanford, and then to the sites that linked to those sites, and so on.
“We realized that this part of the program worked really, really well,” Page recalls. "So I said, 'Wow! The big problem here is not leaving notes. Now we should use it not only for annotation rankings, but also for research".
It seemed like the most obvious application for an invention that gave a rating to every page on the web.
"It was pretty clear to me and the rest of the group that if there was a way to rank, not just on the content of the page, but based on the value the world placed on that page, it was going to be something very valuable to research." .
AltaVista & Co.
The leader of web search at that time was a program called AltaVista developed by the Western Research Laboratory of Digital Equipment Corporation (DEC).
Even the creators of AltaVista had the good intuition to collect the entire web and make it accessible with a search engine. They had missed the neglected link structure. AltaVista analyzed the content of each individual page — using the frequency of words used as a measure — to see if a page was a relevant match for a given keyword in the context of a search.
While there wasn't a clear-cut way to make money from search, AltaVista did have a number of competitors. In 1996, executives at several companies were bragging that they had the most useful service, but conceded that in the race between the omnivorous web and their developing technology, the web was winning.
AltaVista's chief engineer said the best approach was to use massive amounts of silicon to fix the problem and hope for the best.
None of the teams working at the search engine companies thought of using links.
The invention of PageRank
The links were the reason for the success of a research project running on a computer in a Stanford dorm room. Larry Page's PageRank was quite powerful because he intelligently analyzed those links and assigned them a number on a scale of one to ten.
This assignment allowed to define the relevance of the page compared to other web pages. BackRub had simply counted the inbound links, but Page and Brin quickly realized that it wasn't just the number of links that made a page relevant.
The higher the status of the page that generated the link, the more relevant was the link that generated the page itself, and also increased the ranking of the web page when calculating its number with PageRank.
PageRank had another huge advantage. For search engines that relied on traditional content analysis approaches, the web posed a daunting challenge.
There were millions and millions of pages, and as more and more were added, the performance of those systems gradually degraded. For those sites, the rapid expansion of the web was a problem, draining their resources.
But thanks to PageRank, the BackRub process got better as the web grew. New sites meant more links.
"PageRank had the benefit of learning from around the World Wide Web," comments Brin.
If Michigan is better than Stanford
Brin and Page fell into a whirlwind of rapid iteration and throwing. If the pages for a given search weren't quite in order, they went back to the algorithm to see what went wrong.
Page used the ranking of the word "university" as a litmus test. She paid particular attention to the ranking relative to hers alma mater, the University of Michigan, and the University of Stanford.
Brin and Page thought Stanford would rank higher, but the University of Michigan surpassed it. Was it a flaw in the algorithm? No.
“We saw that Michigan had more material on the web, and that explained its better ranking,” Page says.
This result showed the "intelligence" of PageRank. It made BackRub much more useful than the results you got from commercial search engines, whose list of "university" research institutions seemed totally random. The first result for that generic term in AltaVista was the Oregon Center for Optics.
Up until this point, the task of compiling a list of universities and ranking them by importance had been a complicated, intellectually demanding and laborious business.
Some trade magazines employed large crews who worked for months for just that purpose. There was little chance a machine would crank out a rating consistent with the impressions of a well-educated citizen, but BackRub knew nothing about these statistics.
He only knew how to exploit the fact that the links created by a web community had produced a better ranking than one compiled by a group of editors or curators of any expert journal.
Beyond the thesis project?
Page and Brin had thought of the project as a possible topic for their senior theses. But it was inevitable that they also began to imagine a different outlet that could bring them money. Stanford's Computer Science program was as much a business incubator as it was an academic institution.
However, Brin and Page were still reluctant to strike out on their own for the time being. Both had aimed at Stanford with the intention of following in their fathers' academic footsteps.
But selling the license for their search engine wasn't easy. Yahoo didn't see the need to buy the search engine technology. Any chance of a deal with AltaVista was nullified by DEC headquarters in Massachusetts: "It wasn't invented here," they said.
Page and Brin almost managed to strike a deal with Excite, a research firm started by a group of Stanford geeks much like Larry and Sergey. But the deal fell through, due to the "adult supervision" of Excite's new CEO, George Bell, a former Times Mirror magazine executive.
The comparative test with Excite
The BackRub team had visited Bell for a demonstration, opened BackRub in one window and Excite in another for a comparison test.
For the first research he used the word “internet”. Excite's first results were Chinese web pages where the English word "internet" stood out among a jumble of ideograms. The first two BackRub results, however, showed pages informing about how browsers are being used.
It was exactly the kind of result that would most likely satisfy the searcher.
Bell was, however, visibly impressed. Stanford's product was really good. If Excite had incorporated a search engine capable of immediately giving the information people were looking for, Bell explained, users would have left Excite instantly.
Since the advertising revenues came from the people who visited the site — "stickiness" was the most sought after parameter by the websites of the time — using BackRub technology would have been like shooting oneself in the foot.
Perplexities about leaving the academic environment
Scott Hassan recalls urging Larry and Sergey just then, in early 1997, to leave Stanford and start their own companies. He still remembers:
“Everyone else was already doing it. There was Hotmail and Netscape which were very good. Money flowed freely into the Valley. So I told him, 'The idea is the search engine. We absolutely have to do it.' They didn't think so. Larry E Sergey they were both very determined to stay within the academic realm of Stanford».
“We weren't… in an entrepreneurial mood at the time,” Brin said later.
In September 1997, Page and Brin rebranded BackRub to something they hoped would be better suited for the market.
They rated "The Whatbox," only to dismiss it because it sounded too much like "wetbox" (a slang term for "vaginal lube"). Of course, that wasn't a fair denomination.
Then, Page's roommate at the Stanford dorm suggested "googol," a mathematical term that identifies the number 1 followed by 100 zeros.
The choice of name and logo
“The name exactly matched the scale of what we were working on,” he explained Brin some years later. In reality it turned out to be an apt denomination only some time later. Today, in fact, we have billions of pages, groups, images and documents, as well as hundreds of millions of searches every day.
Unfortunately the internet address for “googol” was already taken, as they soon noticed. Luckily Page made a mistake typing the name into the search box. He wrote “google” instead of “googol”. And Google was an available domain name.
“Google was also easy to type and mnemonic,” recalls Page.
One night, using an open-source graphics program called GIMP, Sergey already threw out the homepage graphics. He designed a logo by separating each letter of the name and giving each a different color. The logo was reminiscent of a children's wooden cube puzzle. He conveyed a sense of affable whimsy.
“He wanted it to be playful and youthful,” recalls Page.
The homepage had a box for typing in your search phrase and, underneath, two buttons, one for search and another labeled "I'm feeling lucky." A surprising bet. Unlike its competitors, Google seemed capable of fulfilling a search on the first try.
Unlike many other web pages, Google's homepage was so sparse that it looked like a sketch.
The more stuff on the page, the slower it loaded. Both, especially Page, believed that speed was a key factor in satisfying users. Page later found it amusing that people praised the homepage's zen design, with all that white space.
"The minimalism was due to the fact that we didn't have a webmaster and we had to do everything ourselves," he acknowledged.
A classified project
Meanwhile, BackRub-gone-Google had grown to such an extent that it was becoming difficult to manage within Stanford's facilities. It was becoming less and less of a research project and more of an internet startup run by a private university.
Page and Brin's reluctance to write a paper about their work had become legendary in the Stanford computer science department. People said, "Why are they so secretive? It's an academic project, we should know how it works,” said Terry Winograd.
Page seemed to have fallen into a conflict of interest. On the one hand, he had firmly subscribed to the hacker ethos of shared knowledge. It was what his project promoted: making knowledge accessible to make the world a better place.
But he also had a strong drive to protect his hard-won discoveries. He was reminded of Nikola Tesla, who died in poverty while his inventions enriched others.
Furthermore, writing an article about their experience was ultimately not as inspiring to Brin and Page as being able to build something.
Winograd finally got them to explain PageRank at a public event. The two presented a paper entitled The Anatomy of a Large-Scope Hypertext Website Search Engine at a conference in Australia in May 1998.
Stanford is starting to feel cramped
Soon, Google was handling a whopping 10.000 queries a day. At times it consumed half of Stanford's internet bandwidth. Its consumption of equipment and bandwidth was voracious.
“We begged, we borrowed,” says Page. There were a lot of computers around, and we managed to get hold of a few.”
Page's dorm room was essentially Google's operations center, with a motley assortment of computers from various manufacturers plugged into a homemade version of a server rack — a cabinet built out of Lego bricks.
To store the millions of pages scanned, Page and Brin had to purchase the high-capacity mass storage drives themselves. Page found a place that sold refurbished records for a tenth of the original cost.
They worked fine as long as you didn't replace the disk management software.
“We procured 120 drives with a capacity of nine gigs each. In all one terabyte of space — says Page — but it was clear that we would soon need more. And then what else will there be need?, we wondered. Does all this really happen?”
You start looking for money
Stanford wasn't kicking them out. The difficulties generated by the nascent Google were balanced by the knowledge that the department was doing something important.
"I think they would have presented a great doctoral thesis," says Grace-Molina who worked to retain Brin and Page at Stanford—. I also believe that their families supported them in their studies. But starting a business had become too big an attraction to resist.”
There was no alternative, no one would ever pay Google enough. And satisfied users of the service were confirming that Google was making a real difference. The problem remained, not least, of the resources needed to support this exponential growth.
Brin and Page turned to Professor Dave Cheriton for advice. Cheriton had started his own company at Stanford. He then sold it to Cisco for $220 million. Cheriton suggested they meet Andy Bechtolsheim, his former business partner.
That same evening, around midnight, Brin rushed to write an e-mail to Bechtolsheim. He responded immediately by asking the two students to report at eight the next morning at Cheriton's home, which was on the road he drove to work each day.
Here comes 100 thousand dollars
At this unusual hour, Page and Brin demonstrated the search engine in Bechtolsheim on the Cheriton porch which had an internet connection. Bechtolsheim was impressed, but eager to get to the office, he cut it short by offering the two a check for $100.000.
Brin recalls: “We didn't even have a bank account. 'Cash it when you get one,' replied Bechtolsheim and zoomed off in his Porsche.'
He did it nonchalantly, while sipping a cappuccino any morning of any day on his way to work. In fact, he had just invested in a company that would change the way the world accessed information.
Brin and Page celebrated with a snack at Burger King. The check sat in Page's dorm room for over a month.
… and then the first million of many others
Shortly thereafter, other angel investors joined Bechtolsheim. Among them was Dave Cheriton, the professor who had introduced them to Bechtolsheim. Another early investor was a Silicon Valley entrepreneur named Ram Shriram. His company had recently been acquired by Amazon.
Shriram had met with Brin and Page in February 1998. While he was skeptical of the search engine business model, he was hugely impressed with Google. After the meeting, Shriram invited the two students to her home to meet her boss Jeff Bezos.
Even Bezos was positively impressed by the passion and "healthy stubbornness" of the two students. Brin and Page explained why they would never place advertisements on their homepage. With Bezos, along with Bechtolsheim, Cheriton and Shriram as private investors, Google's fundraising totaled $XNUMX million.
Google.com Inc is born.
On September 4, 1998, Page and Brin filed to form the Google corporation and moved off campus. Sergey's girlfriend at the time was friends with an Intel manager named Susan Wojcicki. Wojcicki and her husband had just bought a house on Santa Margarita Street in Menlo Park for $615.
To finance the mortgage, the couple asked Google for $1700 a month to rent the garage and some rooms in the house. At that point, they hired their first employee, fellow Stanford student Craig Silverstein.
He had come into contact with Brin a Page showing an algorithm for compressing all the links in order to make their memorization and functioning more efficient.
An office manager was also taken on board.
As if they were still students, Brin and Page maintained a presence at Stanford, co-teaching a class, CS 349, "Data Mining, Search, and the World Wide Web." The class was held twice a week that semester.
Brin and Page presented the course as a "project class." The students would work with the repository of 25 million web pages that Google had collected. Google was now a private company.
They also had a research assistant. The recommended first reading was their own essay, but later in the semester the course focused on comparing PageRank to the work developed by Jon Kleinberg.
In December, after final projects were completed, Page emailed students to a milestone party. In the invitation it was written .
“Stanford's search project is now Google.com: the next generation Internet search company. Polynesian attire is recommended, and bring something for the hot tub."
