Google Insights for Search is a great new tool (if not the most elegant name) that allows you to investigate the prevalence of search terms over time in Google. Although it was designed for advertisers, it has a ton of potential research purposes, which Hal Roberts and Ethan Zuckerman have already begun to think about. For example, since it allows you to look at world-wide searches, you can investigate the popularity of different social networking platforms in various countries, as Ethan tells us the folks at Pingdom have already done.
One of the things I thought this might help us understand better is meme creation and tracking–in other words, when does a story emerge, go viral, and die. It also lets you compare the relative popularity of different news stories next to each other. For example, I wanted to see the importance of Iran, Iraq, Afghanistan and, for a pop culture reference, Paris Hilton. The results imply, sadly, that Googlers care a lot more about Paris Hilton than any leading foreign policy issue. Although in May 2004, we see a huge spike in searches for Iraq at the expense, it appears, of Ms. Hilton. The spike in Iraq searches coincides with the break of the Abu Ghraib abuse story.
RESULTS FOREIGN POLICY
I also was interested in tracking searches for politicians, so put in Bush, Obama, McCain and, again, Paris Hilton (who you can tell gets a little bit of a bump recently due to the McCain campaign ad criticizing Obama.) There is a huge spike for Bush around the 2004 election, which is also the first time we see an uptick for Obama, likely due to his speech at the Democratic convention that year. Obama’s recent rise in searches is also quite significant.
Finally, I wanted to compare newspapers to blogs, which was inspired by Hal’s post. However, I did it a bit differently than Hal, since I thought it might be more useful to compare specific media sources in the US. I chose two newspapers and three blogs/bloggers–the New York Times, Washington Post, Daily Kos, Huffington Post and Andrew Sullivan. In this analysis, it is clear that searches for online versions of newspapers blow bloggers out of the water. It is also apparent that bloggers and newspapers tend to attract extra eyeballs around similar news cycles–for example you’ll notice an uptick in coverage around US national elections in October/November 2004 and 2006.
RESULTS NEWS SOURCES
These are the type of ideas we also want to explore with Berkman’s MediaCloud project. By grabbing RSS feeds from hundreds of newspapers, blogs and email lists, time stamping each story and then tagging it with keywords, we’ll have a huge, searchable data set to start answering questions around meme tracking, agenda setting in mainstream media, and the impact of blogs and other online media in repressive regimes–among many other topics. Research this fall will likely be on the computer science side of the tool, but eventually we expect to have all the data freely available on the web so that any researcher can use it–just like Google has done with Insight (although even more data would be nice)!