About Wonkosphere

Wonkosphere is designed for all political writers and bloggers, media people in newspaper, TV, and radio, political workers, activists, and political junkies who need to stay on top of the 2008 Presidential race but can't spend all day searching for the hottest and most relevant material. Wonkosphere is also for those who seek an analysis of the political buzz that is timely and unbiased.

Wonkosphere is different from other sites because we track hundreds of blogs and web sites per day, more than any human can read. Our patented technology text analysis technology identifies posts that matter, not just posts that are popular. If you're tracking individual candidates, our unique analysis will show you their share of buzz in conservative and liberal blogs, highlight the most representative posts, and tell you whether the tone of discussion about your candidate is trending up or down.

Wonkosphere is a service of Crawdad Technologies LLC, located in Chandler, Arizona.

Measures & Sections

Buzz Share

This measures the proportion of buzz for each candidate in all blogs for the 24 hours preceding the latest update. Percentages for each candidate are calculated every four hours. To get the value we first analyze each blog post for each candidate, and score the prominence of the candidate's name within the post. We sum these prominence scores across all candidates then express their scores as a percent of the total.

We also calculate buzz share value for each candidate for each calendar day, broken out by political persuasion of the blogs mentioning the candidate. These values are plotted for the last two weeks in the graphs on the candidate pages. The blue line shows buzz share for liberal + independent blogs, and the red line shows buzz share for conservative + independent blogs.

Tone

Tone measures the amount of positive and negative language used in statements about the candidate in blog posts. On the home page and top of the candidate pages, we indicate whether the tone value is high, low, or average compared to the average range of tone values for all the candidates in the last 24 hours. These values are broken out for liberal and conservative blogs.

We also calculate an average tone value across all posts each candidate for each calendar day, broken out by political persuasion of the blogs mentioning the candidate. These values are plotted for the last two weeks in the graphs on the candidate pages. The blue line shows buzz share for liberal + independent blogs, and the red line shows buzz share for conservative + independent blogs.

The gray band shows the average range for all candidates for each day. If the candidate's tone is above the grey band, the buzz about them is more positive than the norm, and if it's below the grey band, the buzz about them is more negative than the norm. If it is within the band, it should be considered average.

Representative Posts

At the top of the home page just under the buzz share chart, and on the buzz page, we show the most representative posts for the 24 hours preceding the most recent update. These are the posts that best represent the blog posts in general and are not necessarily the "hottest" single issues being discussed at a given time.

One Step Ahead

This is a ranked list of blogs whose posts the day before yesterday best represented all blog posts yesterday. These blogs got to stories first or were "leading the pack" for other reasons. This list is updated once a day.

Hot Links

These are links exracted from the blog posts themselves that appear in multiple blogs. They might be links to blog posts, news stories, web sites, or other web content that many bloggers are talking about. Where possible we get a descriptive title for these links, but where that is not possible we provide the source name, or sometimes just a link.

FAQ

Click items to expand/collapse (requires JavaScript). Internet Explorer users: To collapse list, allow active content.

How is Wonkosphere different from other campaign websites?

The difference is in objective analytics, where Wonkosphere is unique in the blogosphere. RSS feeds and other political sites provide a list of recent blog posts without regard to how important they are. Wonkosphere analyzes every post to filter out the ones that are unimportant or redundant. Some political sites offer selected posts, but the posts are selected by humans who bring their natural biases to the selection process. Wonkosphere selects posts using patented natural language processing algorithms that are run by a computer, eliminating humans and their biases from the selection process. The buzz share percentages and tone categories on the site all result from this objective analysis process.

Is Wonkosphere a substitute for reading blogs?

No. Wonkosphere is a "reading aid" for the political blogosphere. We help make reading political blogs more manageable by selecting posts that are especially important and relevant using unbiased automatic analysis of blog posts. Wonkosphere enables you to go beyond your Favorites List and track blogs that perhaps are not on your radar screen, without getting sucked into the black hole of reading everything.

How does Wonkosphere technology work?

The core of our system is a network text analysis technology developed at Arizona State University called Centering Resonance Analysis (CRA), which is licensed to Crawdad Technologies. It applies natural language processing techniques to convert blog texts into networks, then analyzes these networks to determine the relative importance of all the words in the text. This allows us to judge how much any given post is about a particular candidate, and how positive or negative the writing is. Centering Resonance Analysis has been shown in research to represent the "collective consciousness" of how an audience interprets a text, and has been shown to be more accurate than simple "word counting" methods. Crawdad's development of CRA has been funded, in part, Air Force Office of Scientific Research.

Where do you get the blog posts?

Wonkosphere monitors the live feeds of over 850 blogs and web sites. Beginning with a core set of the most popular plitical blogs, we selected the rest using a "bootstrap" procedure based on network analysis of blogrolls. Each of the blogs is in turn carefully classified as conservative, liberal, or independent. From time to time, we search for new blogs using this procedure as well as analysis of entries in our Hot Links section on the sidebar of the content pages. The blog feeds are monitored for new posts, and when found these are extracted, analyzed, and entered into our database for inclusion in the selection process.

Can I suggest a blog to add?

Yes. Use the link on the Contact Page.

How often do you update?

The Buzz Share and Tone Graphs on the candidate pages, as well as the One Step Ahead section of the sidebar, are updated once a day shortly after midnight MST. The other sections of the site are updated every four hours. The timestamp at the top of each content page indicates the time of the most recent update.

What is Wonkoblog?

Wonkoblog provides daily posts with expert, non-partisan interpretation of the data presented on Wonkosphere, plus occasional commentary on happenings in the political blogosphere. Visitors are encouraged to register and add their comments to the Wonkosphere posts.

Is Wonkosphere non-partisan?

Yes. We provide coverage of conservative, liberal, and independent blogs. All posts are analyzed and selected automatically, keeping human bias out of the process. On Wonkoblog we strive to provide balanced coverage of all candidates and political perspectives. Naturally, because of changing events the posts will vary from day in terms of coverage and how positive/negative they are about particular candidates and perspectives.

Is the Mud Meter™ Coming Back?

Yes. Crawdad debuted the Mud Meter™ in the 2004 Presidential contest between Bush and Kerry, and measured mud-slinging by each candidate against the other. Once the primary contests are decided and we move into the general election, we will add this measure to Wonkoshere.