SEW met the people behind Google Crisis Response at I/O 2012. The work of this small, relatively unknown, team is indicative of the future of information retrieval on connected devices and their journey illustrates why search must continually evolve.
Whilst Google’s 12th birthday and 13th birthday was characterized by a sense of volatility reflected in radical changes to leadership and hard line approaches to spam, Google’s 14th birthday couldn’t be more different. Direction seems to be set and there is a strong sense of stability in the changes that the Company has have made and confidence in the direction they are taking. Symptomatic of renewed confidence and direction is the change in name to one the team the search industry looks most towards – namely Matt Cutt’s team is no longer called Search Quality and is now called Knowledge.
From Search Engine to Information Provider
Whilst there is some consternation towards the role that Google’s Knowledge graph will play in the future due to fears that it will lead to increasing disenfranchisement of webmasters to control the appearance of their own content, the work of the Knowledge team is emblematic of how Google sees search evolving and the competitive edge they must maintain to stay relevant. As Matt Cutts said at SES San Francisco this year, Google is constantly re-evaluating the core principles upon which they built the search user experience to ensure the engine doesn’t “age” unintentionally.
Google Knowledge is emblematic of Google’s values after 14 years for two reason. Firstly, it’s still all about “Faster is Better” which has always been Google’s mantra and secondly Google Knowledge fundamentally about scaling data built on open standards. However, I would say that the true precursor to Google Knowledge is Google.org’s social good project, Google Crisis Response.
When Hurricane Katrina hit New Orleans in August 2005, Google really started to show it’s “age”. Users were flocking to the search engine for immediate answers and information but Google was only able to update as fast as the newswires and much of the most useful data that could have been deployed, was silted in unreadable and incomputable formats that were just unreachable for the average web user.
At the time, to solve the problem, Google had to collaborate with information providers to manually surface some of the information on a piecemeal basis. To the Google.org team, looking at the search query data and speed of response from the current process, that they would need to create a Crisis Response team to cope with possible catastrophic events in the future.
Crisis Response to Prevention Opportunity
In conversation over the phone, Pete Giencke, GIS data engineer with the Google Crisis Response Team, told SEW that the Haiti Earthquake in 2011 was, “when Google Crisis Response came into it’s own.” The first “product” (for lack of a better word), was PersonFinder which generated over fifty-five thousand records of people looking for information on loved ones. The technology was made to be open source from the outset and subsequently was used for the Japan and Chilean earthquakes. “Crisis response is a global effort, so it is not always in the English language. For example, during the Japan earthquake local contacts could send any data they had via the same tools and to get information out very quickly.”
This set the tone for the day-to-day activities of Google’s Crisis Response team. “Now we work with our satellite imagery providers, local organizations and other external data providers to help them embrace open standards such as a standard called the Common Alerting Protocol (CAP). Our aim is to help their data flow freely to users.”
Supporting open standards is crucial for the Crisis Response to function effectively and rally a co-ordinated response between all the different organizations. In the recent Colorado wild fires, the National Oceanic and Atmospheric Administration (NOAA) provided warning data of “fire weather”, whilst the United States Forest Service provided data on the location of fires. Meanwhile the National the State of Colorado provided data on evacuation route information whilst the Red Cross provider data on local shelters.
Crisis response alerts are now geo targeted to the browser so that desktop searchers in a nearby area will automatically get an alert within search results when they visit Google.
When a crisis does hit, Google’s Crisis Response team use open data features with Google Earths KML markup to show evacuation zones and route maps, but the key is to work with authoritative local media outlets to disseminate the information. During Hurricane Irene, Google published a threat evacuation layer onto Google maps on the WNYC radio website to reach both browsers and listeners. The map was deliberately made to be embeddable aswell, so that other news and social networks could also alert their people to the hazards.
During hurricane Isaac, Google worked with NOLA.com radio station to do the same thing, but also found that 20% of search queries came via mobile devices from people looking for Geo information, shelters and evacuation routes. Subsequently, Crisis Response is now built into Android Jellybean so that users will automatically receive a push notification for nearby hazards, such as a tornado warning, without the user needing to search.
The concerted effort of these different organizations illustrates the power of open data and the important role digitally “connected organizations” can play in the future.