Question: How does the search engine do it? Answer: by sweating the details so users don't have to.
Ever think about how Google matches their search results to those short cryptic phrases you type when using Google.com?
Andrew Moore, director of Google Pittsburgh and professor of Robotics and Computer Science at the School of Computer Science, Carnegie Mellon University, thinks about it every day, along with members of his engineering team. The results of their steady contemplation equate to more relevant matches for you. And faster too.
Yet, deep thinking, in a Google sense, doesn’t mean just diving into a problem. Moore receives plenty of help. He has hired a combination of people whose expertise are in computer systems, as well as people who are heavy hitters in robotics, artificial intelligence and machine learning. While these concepts may be difficult for you to grasp, they provide him with two communities of deep thinkers, which he says have together come up with some of Google’s biggest successes.
A good example is Google’s ad targeting system. When they started in the advertising business, Google executives were worried that users would reject advertising because it represents not what users want to find, but what others want them to find. So, Google has developed a sophisticated approach to advertising that much of the population uses to find what they’re looking for. Moore’s team in Pittsburgh continues to work on refinements to the methods used in Google’s advertising prioritization programs to make those searches even better and faster. He brings up the case of somebody typing “tahoe” into the Google search field, and making sure that if the user wants to learn something about Chevy Tahoe, he sees results related to the vehicle; but for Lake Tahoe, he sees those related to the Nevada city.
Moore points out that if Google doesn’t feel there’s a high likelihood that an ad will match the right thought in your mind, there will be no ads on the page at all.
Similarly, when working on Google’s shopping capability, his team strives to provide complete information related to the product you desire—including specifications, images, prices and reviews—and to display it in a way that's easy to select the right resource. Importantly, Google doesn’t actually provide the information. It attempts to select information from other sources that are relevant and reliable. Moore says shopping doesn’t require the user to understand everything in the world, just what is most useful to make shopping easier.
One might think that doing this heavy thinking is Moore’s biggest challenge; but when asked, he didn’t hesitate, noting his biggest challenge is making sure the types of people he wants to hire are aware that he wants to hire them.
Fact is, with as big as Google has become, with its widespread fame and popularity, Moore believes that many people, even within a few miles of his growing Pittsburgh office, don’t know the office is there and that Google is looking for people.
He continues to hire people with backgrounds in computer science or related fields. Moore describes the hiring process as friendly but rigorous, typically involving interviews with a half-dozen people for one opening. Google seeks people who are passionate about the product and, oh, who like to program software.
Once somebody is hired by Google, it may feel like he or she is drinking from the proverbial fire hose—exposed to lots of new ideas, both simple and complex and from many angles, much the way that Google’s search engine deals with the same type of big data.
So be comforted Google is doing it, so we don’t have to.