Skip to Content
Uncategorized

Search Us, Says Google

Can’t find it on the Web? The Google guys came up with a better way to sift through a billion Web pages-a stack of documents 100 kilometers high-in less than a second. As a result, they’re the hottest search engine around.

“A great comedy duo” is how Time Magazine describes Sergey Brin and Larry Page. But these two dropouts from Stanford’s doctoral program in computer science look to be the ones set to do the proverbial laughing all the way to the bank. That’s because they left Stanford in 1998 to launch Google.com, and in just a couple years their search engine has become the Web’s coolest.

Talk with Page and the Russian-born Brin, both 27, and one-liners are more likely to roll off their tongues than algorithms. But these are unquestionably brainy guys who started with the shared assumption that Web searching was broken: As the Web grew, finding good information got harder. So they devised a wholly new approach to sorting search results that they call PageRank: The more links there are to a page, the higher it vaults in Google’s hierarchy. Another innovation: To climb high in the search results, a page needs external validation. Proprietary algorithms that parcel out rankings to pages aren’t influenced by spamming and the other techniques that marketers sometimes use to boost a page’s rating in other search engines. The math gets complicated-some 6,000 computers get involved-but for the user, Google is simple and easy to use.

When Brin and Page began talking up their sorting strategy, Andy Bechtolsheim, a founder of Sun Microsystems, heard their pitch and promptly wrote out a check for $100,000 to help seed-fund their venture. Later, the founders assembled an all-star venture capital team, with money from both Kleiner Perkins and Sequoia Capital (between them, these firms funded Apple, Amazon.com, Cisco Systems, Netscape and Yahoo!). That kiss of the gods of Silicon Valley means big things are expected of Brin, Google’s president, and Page, the CEO.

And they have been delivering-a fact underscored in June when Yahoo! dumped its relationship with longtime search leader Inktomi and installed Google as its default search tool. Google is the first engine to index more than one billion pages, and it won this year’s “Webby” award for technical achievement, presented by the International Academy of Digital Arts and Sciences.

Brin and Page both like talking about how they got to where they are and where they hope to take Google. They provided commentary in conversations with Northern California freelance writer Robert McGarvey.

TR: Why is the Google.com home page so simple?
Brin: When we started, we didn’t have a webmaster. The result was a nice, simple interface. And we have stayed true to that because we realized it helps people get their searches done faster. They don’t want to hang out on a home page when they want to get information quickly. Page: Other search companies have turned into media companies. We have remained focused on search. Over half of our company is devoted to engineering.

TR: What makes one page more valuable than another?
Brin: The premise of our PageRank system is that a page that is pointed to by many other sites is important. In other words, external approval raises a page’s ranking. That’s where PageRank comes in-it is based on an analysis of the linkage structure of the Web. At the end, there are millions of variables, but we manage to do a lot of math, very fast, and in the end we rank pages in a way that’s very close to how you would do it intuitively.

TR: Is Google actually any different from other search engines in how it goes about hunting for information on the Web?
Brin: Google is different from other search engines in that the others use traditional techniques-which pages do the search terms occur on? Google factors in much more information including the link structure, fonts, headings, text of nearby pages and so forth.

TR: Why is it when I search for “Sergey” on Google, your Stanford home page comes up in the second position?
Brin: I was kind of upset about that myself. I was upset it was not number one. I should clarify. “Sergei” is usually spelled with an “i, ” not a “y. ” And then my page has good presence on the Web. There are links to quite a few articles that I authored, and those links in turn are linked to by other pages, so my placement by the PageRank system would be comparatively high.

TR: Is there any place where PageRank doesn’t work?
Brin: There are many places where it falls down. For instance, while the technique works very nicely to measure the relative value of a particular Web page, it doesn’t necessarily measure relevance: Is this page really delivering what you’re looking for?

TR: Couldn’t I create a series of Web pages at every site offering free space, link them to each other, and thereby create higher rankings for myself?
Brin: No, that wouldn’t work. The way the math is designed, you need external approval.

TR: Isn’t there a celebrated Google failing-namely that if the search is for “What is more evil than Satan,” Google points to Microsoft’s home page?
Brin: We fixed that. Now if you type in that phrase, you get articles about Google. This was a result based on technology that looks at what other Web pages are saying about a particular page. Lots of sites label Microsoft as evil-even though that is not how Microsoft is principally described. We view that as a not-great performance of that algorithm, which we’ve since worked on.

TR: How did you two start working together?
Brin: I had been in the PhD program for a couple years when Larry joined in 1995. I was working on data mining, the idea of taking large amounts of data, analyzing it for patterns and trying to extract relationships that are useful. When Larry joined, he started dabbling with the Web and started gathering large amounts of data. That data intrigued me, and I wanted to run various experiments on it.

TR: Why did you start a company?
Brin: For three years, we tried as hard as we could to avoid doing that. Of course, there was a temptation to launch a dot com. But we were pretty comfortable at Stanford. We got to a point, though, where we needed a lot of resources-for example, computers. We had snarfed a whole bunch of machines of all different types; we had sort of borrowed them from various people in the department.

Page: At Stanford we’d stand on the loading dock and try to snag computers as they came in. We would see who got 20 computers and ask them if they could spare one.

Brin: But this was getting out of hand. We saw that if we started the company, we would be able to bring Google to the world quickly-we wouldn’t have to spend all our time scouting for computing resources.

TR: Did you have a business plan?
Brin: When we decided to start a company-and we actually committed to it by purchasing disks ourselves, with our own money-we spent about $15,000 on a terabyte [a million megabytes] of disks. We spread that across three credit cards. Once we did that, we wrote up a business plan and, remarkably, we have stayed close to it over the last couple years.

TR: Was it difficult to raise funding?
Brin: The first check we got was from Andy Bechtolsheim-one of the founders of Sun, and a Stanford alum. We met him very early one morning on the porch of a Stanford faculty member’s home in Palo Alto. We gave him a quick demo. He had to run off somewhere, so he said, instead of us discussing all the details, why don’t I just write you a check? It was made out to Google, Inc., which did not exist at the time, and was for $100,000. So we set up a corporation, deposited the check, and by the time we closed the round we had raised almost $1 million. We continued to build our system and, after about a year, we needed more money. So we raised $25 million in venture capital. Everybody on Sand Hill Road knew who we were.

TR: I’m sure the VCs asked you time and again, What’s your revenue model? How are you going to make money?
Page: Since we’re a private company, it’s not necessarily a bad thing that people are skeptical about our business model. Doubts about the viability of this kind of business probably discourage others from jumping in as competitors. But we don’t have any skepticism. Our traffic is growing nicely-about 20 percent monthly. We answer 40 million queries a day, including 15 million at Google.com. That’s a very significant number. Very few Web businesses have anywhere near that scale. I won’t say exactly what we make per search, but most search companies, if you dig through their financial reports, are making about a penny per search from advertising. You can do the math: 15 million queries a day is about $14 million in revenue per quarter. I am not saying we make that much currently. We’ve been reluctant to do a normal ad deal because we feel we do a better job by providing a highly targeted ad. By the way, if we signed a deal with a Web advertising network like DoubleClick, we would be profitable right now. But we are in this for the long term, and we want to do the right things for our business. Being profitable immediately isn’t the right thing. We have other revenue sources besides ads. We provide search to about 80 sites, including some very big ones: Netscape, Yahoo! and Cisco; that’s a lucrative market.

TR: Why don’t you use traditional banner ads such as would be involved in a DoubleClick deal?
Brin: Understand that we do run ads. A lot of people say they never see ads on our site, but we show them on about 15 percent of our searches. They are text-based ads that load quickly and appear in blue boxes. They are targeted to your search. But we don’t want to use simple banners. They slow down the search experience-users want to get on and off a search page quickly, and we let them.

TR: Has anyone suggested that you bring in some “adult supervision” to manage the company?
Page: We always try to bring in as many adults as possible. But it’s difficult. There’s a trade-off between technological savvy and organizational savvy. And a lot of people we would want to hire-like Jeff Bezos at Amazon.com-well, he has his hands full.

TR: How do you two work out the division of responsibility?
Brin: Larry focuses a little more on the operations side-computers and things like that. I focus on research and marketing. But for the most part we’re interchangeable. It’s a great convenience to us; it gives us more flexibility.

Page: We’re like parents-the interactions are similar. Occasionally, somebody will get upset with an answer they get and they’ll run to the other parent, but we almost always give the same answer.

TR: A factoid making the rounds is that Google is the world’s largest installation of the open-source operating system Linux. How did you come to settle on Linux as a standard?
Brin: Linux is well supported, free and runs well on PCs. What’s not to like?

Page: At Stanford, because we took computers whenever we could find a spare, we ended up with one of everything. And we found that the PCs running Linux were faster, more reliable and a lot cheaper than anything else.

TR: What’s to stop another startup from coming along and beating Google at its own game?
Brin: I believe the search industry is like the semiconductor industry at a younger stage-search engines are becoming more complex, and the barriers to entry are rising. Google now has over 6,000 computers. Search has become expensive.

TR: What’s next for Google.com?
Page: We have a six-person research group that’s chartered to do things that are one year away and that may or may not work. We are on a quest to build the ultimate search engine. We think it will be a smart tool that understands exactly what you want, understands all the information on the Web, and then it gives you the exact right thing. We think Google is significantly better than anything else out there right now, but we have a huge list of things to do that will take us closer to the ultimate search engine.

TR: Things like what?
Page: Recently we have focused on making sure we have access to all the public information in the world. We’ve increased the size of our index to over one billion pages, and we will be very aggressive about continuing to expand that. If you assume each Web page would involve two pages to print if you printed out our indexed pages, it would be about 100 kilometers high. And we search that in less than a second. We want to access everything available, all over the world, in all languages. We recently added Chinese and Korean. Everybody here knows that the mission is to increase people’s experience when they are looking for things.

TR: Do you get acquisition offers for the company?
Page: Oh sure, all the time. We try to dissuade people from giving us offers, though. We’re growing at a good rate, we have been successful at attracting good people and we are increasing our traffic tremendously. We believe we are going to dominate the market-and if you believe that, it’s hard for anyone to pay you enough to justify selling. Search is the number-one application on the Web. And it’s easy for people to try out different search engines so they can compare. They notice differences and they tell their friends. Friends tell friends. And that’s how we grow.

TR: How is running Google different from being at Stanford?
Page: They’re kind of the same. I’m working on a lot of the same problems, just on a bigger scale. Maybe I’m too close to it to have perspective. Probably in three years, I’ll sit back and say, “Holy cow, what happened?”

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.