Interview with Thomas Hofmann, Google

«Every employee can be an innovator»
 Thomas Hofmann, Engineering Director Google Switzerland, about building a start-up company, new challenges and the innovative future of Google.

You have studied computer science and philosophy in Bonn and Paris, not your everyday combination. What made you choose these fields? 

I am interested in pushing the limits on tasks that computers can perform. Think «Artificial Intelligence», but with a modern slant towards statistical methods and machine learning. Philosophy provides an alternate way of thinking about deep problems and I have always considered it very valuable to not only to have been trained in engineering and mathematics, but to have had a broader education that included philosophy and the humanities in general. 
At the time, when you finished your studies, did you have a concept of where the internet was heading? 
When I finished my undergraduate studies, I was using email, telnet, ftp, gopher, later the mosaic browser, but I thought of these as tools for the scientific world more than anything else. It was only as a Ph.D. student in the second half of the 1990ies that I realized where the Internet was heading and what immense role my own research area could play. 
You had interest from Yahoo! and Google but you co-founded the start-up company Recommind. The company recently won an award. Building a start-up is an incredible experience. What do you consider your greatest profit from being part of Recommind from the very beginning? 
Founding a technology company and designing products from scratch that are based on one's own ideas and inventions has been one of the most fulfilling experiences of my life. The excitement and thrill is wonderful, the complexity of the challenges on all fronts formidable. I have rarely learned so much in so little time. What has fascinated me most is the ability to shape the identity and culture of a company, something that gets set in the first year or so. 
Couple of years later you came back to Germany and started to work for TU Darmstadt and the Fraunhofer Institute. What challenges attracted you? 
Frankly, after seven exciting years in the US, I was looking for opportunities in Europe, because this is where my family and I wanted to live. My goal was to take on a position which would offer me the right mix of applied and basic research as well as some level of influence and responsibility to work on projects that could impact the society at large. At the time the position as a Fraunhofer Director seemed to offer me that. Yet, the reality of my role was different from what I had expected and the opportunity to work for Google out of the Zurich-office seemed hard to resist. 
Since July 2006 you're Director of Engineering at Google in Zurich. At what stage did you realize the potential of Google as a company? 
I realized that Google was quite special in early 2000, when I met Larry Page at the Internet Archive in San Francisco. Of course, I was also already a daily user of Google search like many Computer Scientist. Since there was no good business model at the time, I remained skeptical that Google could sustain its rate of innovation and deliver on its ambitious mission. Soon I realized that I was wrong! Since then, I have been learning year by year how much of a difference it makes to hire the best people and to establish an innovation culture like Google has. 
Google is and will be playing a central role in shaping the internet and therefore the future in general. A great responsibility. 
I agree. Many products and services are indispensable in our life, but Google's services and products deal with the basis of how we form an opinion and gain knowledge -information. And not only will we use the Internet more and more in the future, but we will also live an important part of our life out there. That's why «putting users first» is the key principle we always start from. 
How does Google innovate? 
Innovation happens in many ways. Our starting point is the insight that every employee can be an innovator and may have a big idea that can be the seed for a whole new product or an innovative improvement of an existing one. In some sense, everything we do and how we organize ourselves is centered around innovation: it
is our live-blood. I would highlight the empowerment of small, agile teams as one important aspect. While we provide top-down direction and strategy to our teams, much of the real innovation happens in small teams, usually a handful software engineers and a product manager. They suggest what they want to accomplish
quarter by quarter, they are often the ones to develop a vision for their project and use their technical expertise to deliver on it. 
Google has a very open approach to research and innovation. Google expects all engineers to be engaged in the innovation process. They are expected to spend 20% of their time working on a project of their own choosing. What has come out of this? 
Well there are many examples of important products that have started as 20% projects. Google News is an early example, Google Transit and Orkut are other ones. Google Suggest (just launched a few week ago) is a more recent example. Yet, there is much more that gets accomplished in 20% time. Innumerable valuable improvement to existing products have been made in 20% time, many initiatives that improve communication and knowledge sharing among engineers are powered by it. One benefit that is often overlooked is that it creates «random» connection in the social network of engineers (by one developer working with another team during his 20% time), thereby avoiding compartmentalization and stagnation. So in many ways, 20% time is a key component in maintaining our unique culture. 
Why is it so hard to predict changes in technology? From a layman's perspective some rules like Moore's law look rather reliable. 
One may be able to predict the scaling behavior of key technologies (e.g. network bandwidth or storage) by Moore-like extrapolations without being able to predict which innovations exactly contribute to the progress. What is hard is already the next step: How will the behavior of users change? How will users make use of better technology? Some of the change will just be incremental and quantitative, e.g., I may store more of my digital photos and in higher resolution, if storage is cheaper. Yet, at some point there may be a qualitative change, a phase transition that leads to something new. I may simply put all my photos out there to share them with the rest of the world and a completely different set of social dynamics is created. Communities emerge and the whole behavior of what it means to take photos and what for changes. Or think of the YouTube phenomenon, which is enabled by technological advance, but not «determined» by it. These changes are hard to predict. Similarly, some changes happen through a combination or convergence of technologies. We may be
able to understand and predict each one independently, but understanding their interactions is a much more challenging task. 
The internet is fast moving. What is here today, is not exactly gone tomorrow in the sense that the information is lost, but trends come and go at a quick pace. Google recently announced an alternative to Second Life called Lively. What roles will social networks play in the future? 
Many new products will be social and many existing ones will become more social in the years to come. Lively is one specific product that goes in that direction. Even more important in my eyes is Google's OpenSocial, a common API for social applications across multiple websites. This allows developers to create applications that access a social network's friends and update feeds. Instead of building everything in-house, we engage with the open source developer community to address the many specialized social applications that people may want to use. I think this is a great example of how Google can spur innovation outside of the company. 
Branching out into fields like online email services, social networks, Google is growing in «non-traditional» fields. What is happening at the core. What will the future search engine provide? 
Search still remains the key paradigm in organizing information. One of our most important and ongoing innovations in the last years has been to make more information accessible through the same single search box of our web search engine. We are not just searching over web pages, but we return videos, images, books, news, maps, you name it. This has been called «universal search» and is only possible, because we developed a much improved infrastructure that is capable to route every Web search query to various back-ends. Of course, there are also the long-standing challenges of better understanding what a user is looking for. Understanding the query and the user context better is critical to improving search engines in that regard. Finally, better leveraging what others users do also bears quite some potential. Google Suggest is an example of that nature: we help people formulating their queries by showing them searches that have been popular with other users. 
Google is often attacked, recently by rival search engine Cuil. How does one defend a position as a leader? 
It is great to see other search engines try out new approaches. This is the right thing for the user and it also stimulates us to constantly re-think what we are doing. In general, it is mandatory for us not to rest and not to take our lead position in many countries as granted. We should neither be self-content, nor reactive, but rather actively pursue new avenues and constantly challenge ourselves – which is what we do. 
What can we look forward to hearing from you at the European Futurists Conference Lucerne? 
Nobody can predict the future and nobody can plan innovation. But as an organization, despite all the uncertainty, you can be prepared. I will highlight some concrete organizational innovations in Google that help us being ready for the future. 

Thomas Hofmann has been a researcher in Artificial Intelligence and Machine Learning for more than 10 years. Joining Google in 2006, Thomas is currently an Engineering Director responsible for search quality and monetization at Google. He also serves as a co-Director of the Zurich engineering site.

Thomas Hofmann
Keynote «Innovation at Google»,
Culture and Convention Centre KKL, Lucerne, Switzerland
October 28, 2008; 14.45 h



Hotel Schweizerhof  Auviso