Google Israel's Chief Geek Yossi Matias Offers a Peek Inside the R&D Center

The AutoComplete function on Google Search? Google Flu Trends? The Dead Sea Scrolls project? All developed with the help of Google's Israel R&D unit. It's head Yossi Matias sits down with Fast Company for a chat.

<a href=Yossi Matias, Head of Google R&D Center Israel" width="350" height="525" />Working as the head of a Google research and development center ranks up there as one of the world's ultimate geek jobs. Yossi Matias, managing director of Google's R&D center in Israel, has had some interesting projects go through his lab. Google's facilities in Tel Aviv and Haifa have played key parts in developing Google Instant, GMail, new functionality for YouTube and Google Trends. Along the way, they've even helped develop a flu epidemic prediction system and begun to digitize the Dead Sea Scrolls. We talked with Matias at Google's New York offices recently about the expanding role of Google's Israel R&D unit, cool tricks it helped pioneer (like the Choose Your Own Adventure-like YouTube Annotations), and what they have in store.

FAST COMPANY: What was Google Israel's role in developing Google Instant?

Yossi Matias: Our R&D center, which is quite sizable at 150 people plus on staff, is located in both Tel Aviv and Haifa. We operate as a single center working on global technology, products and innovations for Google. One of our most significant and growing areas is search. In particular, one of the technologies we developed was Google Autocomplete, also known as Suggest. Google Instant obviously takes this notion of Autocomplete in a leap forward in the sense of accelerating the search experience by giving search results as you type. One of the core technologies behind that is our Autocomplete technology, which comes out of Israel. Obviously we're very fortunate to be part of the bigger team which developed Google Instant. The team was primarily in Mountain View but also consisted of help from other labs around the world including our team in Israel, which was responsible for putting in Autocomplete technology.

How did the Dead Sea Scrolls project come about?

Let me give the bigger picture and plug in how the Dead Sea Scrolls fits. In general, one of our core teams centers around search. Here we have some quite visible projects such as special search results which appear at the top of the screen. For example, when the Olympics, World Cup and Oscars took place, searches related to these events triggered all sorts of results at the top. This auto-technology also came out of Israel. In addition, we also are working on some next generation search technologies.

In addition to search, we have some global initiatives. One is based around data analytics and visualization. For example, Google Trends and Google Insights for Search also come out of Israel. Google Visualization API and Chart Tools, which enables one to display visual information of textual data in a compelling, open cloud way, also comes out of Israel.

If you're familiar with Google Analytics, we have teams closely working on some projects with Mountain View for it. We also have some teams working on applications related to Gmail and others, along with another team working on networking infrastructure—how to make sure Google's network services all these applications well.

Google is one of the biggest networks in the world, so obviously there are many interesting and challenging technological problems there.

In each of these areas, we have projects and efforts that we are leading. Many of the projects started up as initiatives bought up by our engineers—that's part of the culture of Google. We foster innovation and allow engineers to come up with their own ideas which sometimes develop into highly visible projects. A good example is the Visualization API, which actually started up with engineers developing some visualization for a particular product and discovering that developing visualization was much more difficult than it should have been. We also asked ourselves "what can make it a simple task for anybody to do?" and that's how it came into effect.

Similarly, a very visible technology that came out of Israel was YouTube annotations. Many random videos on YouTube have clickable annotations that enable someone to click on a blurb that takes you to more information or even another video. This created some very interesting applications from the media perspective.

For example, the London Police used it for a very creative clip which made a branching story designed to educate teenagers about the possible implications of taking a knife to school. The video gives them a choice of "take" or "don't take" that creates a branching story with different ends. This even won some prize I believe at a European festival. This is a technology which was actually started by an intern at our office which later became a 20% project. A 20% project is a project executed by engineers outside of their normal project work for up to 20% of the time in order to foster innovation and all sorts of wild ideas that can create stuff like that.

These are the kind of innovative projects and technologies we do at Google Israel. In addition, and to give context, one thing we do all the time is to encourage our engineers to find new ways to benefit all sorts of world-wide causes. Various people are taking various initiatives—for example, two female engineers who took a wonderful initiative to encourage young students of high school age to emphasize mathematical and computer studies with great results. In this context, some time ago we looked into the question of how to help facilitate making great global projects more available to users. A few years we helped Yad Vashem [Jerusalem's Holocaust memorial] take their video archives and make them available through a YouTube channel, which gave them magnitudes more views.

Now when the Israel Antiquities Authority started this wonderful project, we began a conversation with them and we are very interested in facilitating making this data available. It is certainly part of Google's mission to organize the world's information and make it universally available. We started conversations with them some time ago on various levels. Personally, I'm very interested in that. We got to a stage where I'm very happy about how we aligned our interests on making this data available. Google will be part of the initiative to make this (the Dead Sea Scrolls) available.

Can you give an example of what sort of projects for Google Analytics and Search were developed in your lab?

One exciting project was for Google Insights for Search. One purpose for the Google Insights for Search is to allow any user to get a glimpse into what users are searching for and what their interests are. This has already been used by scholars, economists, health scientists, and others as a tool for getting instant surveys and identifying trends in a social sense. Obviously, we know that what people are searching for has a high correlation to their interests and actions. We learn a lot on the aggregate from the searches. What we learned previously was that we had to conduct an expensive survey of users interest in a topic which would take time and money to only answer a handful of questions for a limited need. One can instead use something like Insights for Search to get an instant survey on many topics, almost without limitation.

Marketers use it quite often to understand trends in user interest. Economists started to use it. Hal R. Varian, chief economist at Google, wrote a paper based on Google Trends and Google Insights for Search where he showed how you can use the various trends of user searches in different categories for economics. What we show in Google Insights is not only particular keywords but also categories of keywords. For instance, trends of all searches related to automotive sales. There are over 600 subcategories there in Google Insights for Search. The innovation is that you can take all searches in a certain category and see that there are often correlations between searches in a given category to various economic indicators.

There have been works measuring this correlation and showing how this can be used for good forecasting of these indicators. Forecasting in order to get a quicker idea of where these economic indicators are heading—something called "nowcasting." This is now being used by quite a few scholars, including in the Federal Reserve Bank and other places, as a tool to learn real-time information about economic trends.

A various nice demonstration of what can be achieved by aggregating search trends was done through a product called Google Flu Trends. This was launched by Google to show indications of possible flu outbreaks by applying mathematical models to a mixture of keywords. These keywords were identified as possible indicators of a flu phenomena happening somewhere. Correlating trends of searches were combined with data taken in collaboration with the CDC, thereby building models which could identify possible flu outbreaks earlier than would be possible using formal institutions. We're very excited—this was one of our early projects when we just opened the Israeli office.

What's the relationship between the Israeli R&D center and Mountain View like?

So the Israeli R&D center is part of Google's global engineering organization. On the one hand, there is a very strong autonomy in driving efforts and projects, some of which I mentioned. On the other hand, each and every one of them are part of bigger projects and efforts. We work very closely with all these offices. For instance, I am here today (at Google's offices in New York) because we are working very closely with teams in New York on various projects and efforts. I will spend next week in Mountain View because we are working with many teams there as well as other teams worldwide.

We work in strong collaboration with other teams according to the areas on which we work, but most of what we do in Israel are projects we feel a strong sense of ownership on. Some are projects which were conceived in Israel, so to speak. Others, at some point, we took on after they started somewhere else.

Are the employees at Google Israel primarily Israeli? Are there foreigners working there?

That's an interesting question. When we look at the demographics of Israeli engineers and scientists, there is a very good mixture of attributes in experience, origins, education ... all this stuff. So for example in terms of age and experience, we have some with more than 20 years experience and some right out of school. Some are even still in school and with us because they had some good experience from the Israeli Defense Forces. In terms of origin, many obviously came from top universities in Israel.

We have a lot of top talent in Israel coming out of Israeli universities, but we have some who, so to speak, "made aliyah" [immigrated to Israel] and it was a different opportunity for them to come and live in Israel. We have some Israelis who spent a few years in the U.S. and came back to Israel. So in terms of education, we have people educated from top Israeli universities like Tel Aviv University, Hebrew University, Technion, Ben-Gurion, Weizmann, and others. We also have some others who went to schools like MIT, Carnegie Mellon University, and other top schools in the U.S.

We have a good mix—the nice thing is that we are a global company and working at the R&D center is very easy for anyone who would go to any other Google office to consider going to Israel.

What's the relationship between the Israeli R&D Center and the startups Google purchased in Israel like?

As a center in Israel, we have a lot of interaction with the ecosystem of the Israeli high-tech industry. Personally, I have strong roots in academia in my background. I am still on the faculty of Tel Aviv University and on leave as a professor of computer science there. I also, as an entrepreneur, had various start-ups which I was involved with or started in the past. So we work very closely with the high-tech companies, with the startups, with venture capital and universities.

For example, quite often when we have visitors we bring them over to Garage Geeks and other events with start-ups, or bring start-ups to our office. There are multiple connections to the start-up world for us. Since the Google approach is to have open platforms and open technology, we encourage start-ups to leverage from these open technologies and work with them. We encourage them to leverage from the cloud and collaborate in whatever they do in an open way. That is one area of connection.

In terms of acquisitions, those acquisitions we made became part of our team. The teams that came over because of acquisitions are now just inherent parts of our team.

[Note: Parts of this interview were condensed for length and readability]

Add New Comment

2 Comments

  • Tyler Gray

    Clarified some language in this post. Google Flu Trends was not developed in Israel, but was cited by Matias as an example of how trends and search aggregation technology worked on at Google Israel were used.