Abstract:
Outside of a lab environment, it has been difficult for researchers to collect both behavioral and self-reported Web-use data from the same participants. To address this challenge we created Roxy, open source software that collects real-world Web-use data with participants’ informed consent. Roxy gathers Web log data as well as the text and HTML code of each page visited by participants. We describe Roxy’s data gathering capabilities and search functions and then illustrate how we used the software in a multi-method study. The use case examines selective exposure to political communication during the November 2010 U.S. general election campaign.

I’m very please to announce this online publication of the conference paper. A revised version will be submitted for publication in the JITP itself as a workbench note shortly. The website for the software is roxyproxy.org.

I’ve been remiss in posting to this blog recently for a variety of reasons, chiefly because I’m analyzing the (massive amounts of) data I collected for my dissertation about the role of technology in citizens’ selection of political news and information during the November mid-term election campaign. Also, I’m working on documenting the software I created with the help of Audacious Software to collect real-world web data for my dissertation, which is called Roxy.

What is the more traditional media’s relationship to these kinds of sources? The stories played out quite differently. In the case of Sherrod, virtually no fact checking (such as asking Sherrod about it) was done before the story jumped from blogs to cable TV and beyond. With the Wikileaks documents three newspapers combed through them for about 3 weeks before they ran the story, with the New York Times asking the White House for comment before it ran.

Obviously the source material was different, a video has an appearance of authenticity that thousands of documents do not. The video also came with a pre-packaged headline that seemed relatively easy to accept (even for Sherrod’s bosses who pressed for her resignation). The document leak was advertised as potential evidence of war crimes, but that wasn’t something traditional media were ready to embrace, which was appropriate and should have applied to the Sherrod video as well. Right wing bloggers like the source of the Sherrod video are certainly more familiar to journalists as sources than a transparency activist organization like WikiLeaks, although that is not a good reason to accept and print what they say.

“Journalists … If you want to know how our profession ends, look at Demand Media” (see Fry’s post)

On the one hand this is a business model for web content that really makes sense, on the other hand it turns formerly autonomous professionals into cogs in a giant content machine. However, I think a broader perspective is in order. Demand is trying to fill up the holes in the furthest reaches of the long tail. This may be a huge business, creating jobs in content creation where none existed before, or creating options for those who are out of work. The vast majority of journalism isn’t at the far reaches of the tail, it’s much closer to the top, which requires a different business model.

I do try to put down something about the methods and the main findings at minimum. It also helps to write it as if you were explaining it to a classmate so that you don’t use shorthand that won’t make sense to you later.

I couldn’t resist using a new software tool. I’m using Scrivener, a Mac-only application for writers. I like that it is designed for re-organizing and tagging which is what I’ll do when I have a better handle on how I want to organize the project. For now I can just write whatever comes to mind. Well-organized text files would probably work just as well.

Today, however, I followed a link from Jay Rosen’s Twitter that said “Blogs vs. big media in news diffusion, huh? Submit to frame check http://tr.im/sFL9 by @sjcobrien. Its one system, baby.” This took me to Chris O’Brian’s post where he argues against distinguishing between blogs and mainstream news. “I see blogs not as competition, but vital parts that help expand the conversation around news and information. I worry less about who is winning the battle of breaking news first, whether it’s mainstream sources, blogs, or Twitter.” It’s a valid point, but that is not a criticism of the research that was actually carried out. There are lots of other things one could or should focus on when researching the news media ecosystem but a study is just that. One article might add one piece of the puzzle and there is always more to do. It seems perfectly reasonable to explore where news memes originate particularly as we have lost several mainstream news organizations recently and the two major newspapers in Chicago are currently in bankruptcy. I could think of more useful “memes” to track, but it seems like an interesting piece.

It is necessary to develop some form of categorization of news sources (which need not be binary) to research this question. This is always imperfect but it can be done in better or worse ways. That’s where Scott Rosenberg adds to the conversation. Rather than criticizing the idea of doing the study at all, he digs into the methods and finds that “the study relies on Google News to draw a boundary between the news media and blogs. A site that appears in Google News is considered media; everything else is a blog. While this approach is convenient, it ends up slicing off some of the top layer of the blogosphere in arbitrary ways.” Good point. It’s really key to know that, by this definition, if a phrase shows up on the Daily Kos, which is a Google News source, it would count as mainstream news in this study. It’s important to develop better ways of categorizing news, which may (gasp) involve human decision makers as well as computers. Developing categories of news sources is not the same as labeling journalists who, like Chris O’Brian, may write for many different types of outlets.

Let’s discuss what kind of meaningful distinctions remain among news organizations so we can do better research rather than waving our hands and saying it’s impossible.

Along similar lines Google just announced a new tool, Google Squared that “automatically fetches and organizes facts from across the Internet” (see full announcement) For example, “A search for “small dogs” pulls a list of small dogs organized by size, weight and breed” (read further).

These first examples are fairly uncontroversial, particularly because this is a neat new technology and not something people rely on .. yet. Just to illustrate the potential minefield that could await Google, why doesn’t the unemployment rate graph that Google provides let me break down the figure by race, for example? I’ve heard on the news that the rates are very different for African Americans and I want to look into that in my county. I know how to download data from the BLS, but most people don’t and certainly won’t. Race classification is a political issue in France, however: “Classifying people by their ethnicity is illegal in France – the nation of “liberty, equality, fraternity” considers all people should be equally French with no differentiation” (The Guardian). I hope this highlights the potential difficulties.

There are much more subtle issues that can have major consequences for how we understand information – see Tufte’s quite illuminating book The Visual Display of Quantitative Information for examples.

You might notice I identified Tom as a “style columnist” since that is the section of the Washington Post in which his article appears, a fact that is incredibly easy to miss online where the context is more likely to be a link that says “look at this biased article in the Washington Post” rather than your own perusal of the style section. Alexandar wrote:

As a TV critic, it’s his job to offer a viewpoint. “I never talk about policies,” Shales told me. “I talk about how [Obama] comes across on TV. I like him based on what I see on television.”… But judging from the steady flow of complaints after each Shales review, a surprising number don’t see a distinction from the news pages.

Alexander feebly suggests labeling the piece as a review, and also acknowledges that this probably would not help.

Would these same conservatives be upset to find this article in the style section of their physical copy of the Washington Post? Probably. Would a lot of these people who are mainly interested in politics actually read the style section? No. News articles were circulated among groups long before Twitter or WordPress, but it was harder to accomplish. The one ideologue who does read the article can easily announce it to interested parties that share his/her views, which was not the intended audience of the style section.

The audience may have misunderstood the intent of the article, a television review, but the paper may also have misunderstood its readers and how those readers gets news – not just through the paper but through interpretive communities.