Varian's analysis

GOOGLE chief economist Hal Varian predicts big things from robots and statisticians.

HAL Varian was an academic economist when he bumped into Eric Schmidt. "Hey," the Google CEO told him, "I've just joined this great little start-up. We've got 400 people. You should come and take a look."

That was in 2002 and Varian had just stepped down as the dean of information at the University of California, Berkeley. He took the trip to Mountain View, near Stanford University and close to where the search engine company had been launched in a suburban garage. The rest, as they say, is history.

At 64, Varian is one of the older staffers at Google, less well known than Schmidt, now executive chairman, or Larry Page and Sergey Brin, the brilliant founders of a company that has grown to 24,000 employees. However, he has been influential in transforming Google from a classy bit of software into a verb, a mindset and a global productivity tool with a market capitalisation of about $US170 billion. And there’s more to come – last month Google launched a $US12.5 billion bid to acquire mobile phone company Motorola.

Building dollar value is where Varian, as chief economist, has made a big difference. When he started at Google, the company had a powerful search product, but no clear idea of its business model. When Varian asked what he should work on, Schmidt replied: “Why don’t you take a look at the ad auction? I think it might make us a little money.”

AdWords, a system that auctions the right to place an ad in Google’s content, was already in place, but it was only when Varian added economic insights from the research literature that it became a huge money-earner.

On a recent visit to Australia, Varian lamented the gap between economics and business, the bifurcation that means there is one vocabulary for business people and another for economists. “They split off because economics was not as relevant or maybe economists were not as interested in being relevant to business,’’ he says. “I would like to see them come back together.” They’re getting closer. Since Varian took the plunge, Microsoft and Intel have employed chief economists to work alongside other executives in the C-suite.

Varian is also closely involved in forecasting, otherwise known as “nowcasting’’ or “predicting the present”. It uses special search functions, such as Google Insights for Search (which lets you compare search patterns across place, time and categories) and Google Correlate (which allows you to upload data and correlate it against Google search inquiries over time). Both functions are free, but Correlate uses only US data. They both demonstrate a simple yet profound idea – that you can discover what is happening in the real world by looking at search patterns in cyberspace. Will Holden outsell Ford this month? Is your newspaper ad campaign generating interest in your product? What’s likely to happen to house sales in your region? When do most people look for a hangover cure? Will anyone be travelling to Florida this year?

Insights and Correlate began when Google staff found that queries about flu predicted official flu figures from the US Centres for Disease Control and Prevention issued weeks or months later. Epidemiologists are interested in this ability to spot a looming epidemic and its location. “Nowcasting’’ has also predicted jobless numbers in the US by looking at search queries on claiming benefits. And Google searches show that cars are a measure of consumer confidence in the US: When people are feeling flush, they look for BMWs and Mercedes; and when they’re not, they look for a Lexus. Europeans do the same with their vacations – when they are confident, they search for luxury over more modest holidays. “We think there are a lot of hidden nuggets in Correlate.

We have basically crowd-sourced it and put it out for people to find interesting things to do with it.’’

The finance sector has seized on it, trying to see whether Google searches can predict market sentiment. Google+, the company’s answer to Facebook, has also excited those who want to know what we really think and Varian says it is only a matter of time before a tool such as Google Insights will look at social networking sites and capture sentiment.

Varian’s team has created a Google price index based on price changes on the web. The tool has not been released, but offers a daily measure of inflation, tracking CPI figures closely. With similar work by Massachusetts Institute of Technology academics, it has sparked debate on how governments might modernise their methods of collecting inflation data.

It is this deconstruction of digital data that will make the statistician among the most important people in a company in future, according to Varian. “Virtually every large company and many small ones have a data warehouse where they collect real-time data about their transactions and their organisation. The challenge is making sense of that data and for that you need the statistician.’’

Knowledge equals productivity: Last month Deloitte Access Economics predicted Australian business would receive a $27 billion productivity boost from the internet this year. In a recent study, Google found students who used its search funtion answered questions three times quicker than those who used the library. Productivity also comes from reduced communications costs. “A tiny company with a dozen people has access to infrastructure that only the largest multinational could afford 15 years ago,” Varian says. “You can put in Voice over IP, you have Google Docs, Wiki, email, social networks, chat. These tools would have been ridiculously expensive 15 years ago and now they are free, ubiquitous. So you can co-ordinate productive activity globally at a very low cost.’’

Varian puts robots and collaboration at work at the top of his “futures’’ list. “One good bet in forecasting the future is that what rich people have now, the middle class will have in five years and poor people in 10 years. Think of mobile devices. Once, only large companies had big information systems and now anyone can have one. Robotics is where computers were 15 years ago. All the big manufacturing plants have robots, but they’re really expensive and have to be cared for by specialists. But they are getting cheaper and cheaper.’’

Take the car: “I took a ride in a Google self-driving car and it went down the freeway at 100 kilometres an hour, drove around the neighbourhood. It was amazing. Rich people have chauffeurs, but I think that in 10 years we will all have chauffeurs – robotic chauffeurs – because the technology has advanced to the point where it is absolutely real-world stuff.”

Or medicine: “Right now doctors are using laproscopic devices, but you can automate a lot of that. Why is someone stitching you up by hand? We know about sewing machines.’’

Tools for social networking will increase collaboration at work as a generation that routinely uses shared computing facilities, Google Docs and Google spreadsheets at school and university hits the office. He offers a (possibly apocryphal) story. A CEO asks his CIO: “Can I have a tool where I could have a spreadsheet and multiple people who work for me can enter numbers and we can add them all up?’’ The CIO says: “Well, we can do that if you give us a year and a half and $20 million dollars.’’ And the CEO says: “Really? Because my daughter’s soccer team has that already!’’

Another prediction: Most people will experience the internet through a mobile phone, not a desktop computer. That means more use of voice recognition software, partly because it is hard to type on a phone and partly because billions of people will have mobiles, but little or no literacy.

Last year Google announced it would give $2 million to Salman Khan to continue building the Khan Academy he set up in 2006. It has 2400 seven-minute educational videos on YouTube and is a “virtual’’ college for everything from physics, history and computer science to the entire US high school maths curriculum – from basic arithmetic to vector calculus.

Varian has kept one foot in the academy and loves the cross-fertilisation of ideas this allows, although he cautions the real world “can be overrated’’. Still, he draws on his Google experience when he fronts his Berkeley students. “They’ve grown up with technology and they are very quick to adopt things and try things. Our brand identity among college students is very high. They all want to work for Google.’’

SEARCH RESULTS FOR HAL VARIAN
Hal Varian explains how Google’s ad auction works on this Youtube clip. It is worth noting that the auction runs in tiny fractions of a second each time a user enters a search term.

Varian has written two bestselling textbooks, but it is the book he co-authored with Berkeley colleague Carl Shapiro more than a decade ago, Information Rules: A Strategic Guide to the Network Economy (Harvard Business School Press, 1999), that people still remember as an excellent survey of the digital economy.

Varian received his PhD in economics from the University of California, Berkeley in 1973; and he has taught at MIT, Stanford and Oxford, among other universities. He was the founding dean of the School of Information at Berkeley
and remains an emeritus professor there.

Varian on how the web changes management: “If you look at the beginning of the 20th century, we saw the rise of mass production. Henry Ford and the entire team were down on the factory floor raising this, lowering that, speeding up the assembly line, changing the way things were built and they were able to extract far more efficiencies than were available before. I think the same thing is happening now with digital technology. When we’re all networked, we all have access to the same documents, to the same capabilities, to this common infrastructure, and we can improve the way work – intellectual work, knowledge work – flows through the organisation.”

You can see what Varian’s robotic future looks likes at YouTube. Search for "self-driving cars" or "autonomous parking".

For more on how academics are looking at using prices collected on the internet as an inflation index, check out the Massachusetts Institute of Technology Billion Prices Project.