This paper discusses early findings and methodological pitfalls of a study of a hyper–mediated election campaign in Canada. Web logs — or “blogs” — serve as the primary object of study. The paper focuses on possible sources of large scale data (aggregated blogger posts) and methods of determining the political influence of bloggers. A series of methodologies are proposed to resolve the over–reliance upon information aggregators and blog search engines provided by Google and Technorati.Contents
Introduction
Mapping code politics: An overview
Ranking bloggers in a leadership race
Research questions
Method
Analysis of rankings
A consideration of ranking methods
Proposed methods for Stage II: Federal election campaign
Conclusions
Introduction
Political conventions, once the highlight on the tele–visual calendar, are slowly giving way to less visible, digital forms of democracy, and now more recently, online forms of communication and organizing. Conventions are expensive and time–consuming. In many jurisdictions, the grandiose party convention has begun to mirror general elections, with electronic voting for individual party members slowly replacing the spectacle of hung–over delegates dancing and chanting on the floor of convention centres. As we see in this paper, even the very notion of closed door, centralized politics and “back room” party operatives working in political capitals, is being challenged by digital networks. This paper discusses the impact of a digital back room politics, and how one might evaluate the influence of its political actors — bloggers.
This short paper develops research methods for a political campaign research project housed at the Infoscape Research Lab at Ryerson University [1]. The paper bridges two stages of the project: first, an exploratory set of research projects focusing on online news, bloggers, and candidate Web sites in the leadership race for the Liberal Party of Canada, and second, initial preparations for a larger study of the forthcoming Canadian federal election, expected sometime in 2007. The paper is divided into two sections. First, we begin with a discussion of the initial methods and findings from the first completed stage. A critique of the methods employed in this first stage subsequently feeds into the paper’s second section — a discussion about possible revisions to methods in advance of this research project’s forthcoming study of the Canadian general election. The introduction of the paper also offers a broad survey of the research group’s conceptual framework: an analysis of “code politics” that takes Web languages, commands, and code as its primary object of study and analysis.
The first stage of the Code Politics project [2], conducted from August to December 2006, focused on three spheres of interest, first the influence of bloggers in the Liberal party’s leadership campaign; second, coverage of the leadership candidates and campaign issues in online news; and third, campaign tactics and strategies encoded within the leadership candidate’s respective Web sites. The project also set about comparing these three spheres, for example, by attempting to determine which issues promoted on candidate Web sites were being taken up by bloggers and journalists alike. This short paper, however, will focus specifically on research conducted on the influence of bloggers.
Mapping code politics: An overview
While drawing some influence, insights, and methodological considerations from a host of data samples and methods, the project was primarily concerned with the political use of Web–based platforms, content, and most importantly, the underlying language of the Web — commonly referred to as “code”. The most commonly discussed Web code or language, html (hypertext mark–up language) [3], served as a key object of study, although as the project progresses we are also finding that automated forms of content delivery and formatting (php and asp, for example) are becoming an increasingly popular technique to manage — and automate — additions and revisions to Web site content. It is our contention that html, php and other underlying Web languages remain largely understudied artifacts [4]. HTML serves as a digital shell that contains not only textual, and graphic images — both static and moving — but also a phalanx of commands that both structure the display and functionality of Web objects [5]. The broader Code Politics project then set about determining how electoral campaigns, and their operatives, literally encoded the Web for their political purposes. In short, the project sought to expand the syntax of Web–based research so as to deduce and deconstruct the efforts of political campaigns online. This “hyper–mediated” study of electoral politics (Howard, 2006), thus emphasized the need to vertically drill down from computer interfaces and screens into the underlying software, operating systems, languages and code formats, while also tracking Web code and content horizontally across online formats (primarily Web sites, blogs, online news pages, rss feeds, and search engines).
Conceptually speaking, Code Politics attempts to recognize both horizontal and vertical dimensions of the Internet, Web, and computer–based languages (Elmer, 2006). Horizontal approaches, commonly employed by rhetoricians, cultural theorists, and other narrativists typically focus on non–linear dimensions of cultural expression, language, and media. Hypertext and cybertext scholars have historically dominated horizontal studies of the Internet, questioning and critiquing forks in narratives, texts, and increasingly digital games (Bolter, 1991; Landow, 1994; Aarseth, 1997). Moreover, central questions which typically pervade horizontal approaches focus on authorship and meaning, common problems within the field of semiotics and cultural studies. A vertical perspective of the Internet, by comparison, questions the techniques, sites, and processes that both separate and bring together functions on the Web. In other words, the vertical approach assumes the Web to be a medium of layers which control — and most importantly interconnect — relationships between users, their computers, screens, and the seemingly limitless number of services and content stored and facilitated by the network of remote servers and routers. The languages embedded within the Web page therein serve to highlight, govern, and control such inter–connections and functions.
Ranking bloggers in a leadership race
A major part of the first stage of the Code Politics project focused on the role of bloggers in the leadership race. The campaign for the leadership of the Liberal party of Canada — historically Canada’s center–left governing party [6] — was the first leadership contest to face the scrutiny of organized, partisan (party affiliated) bloggers, both from within the party, and across the political spectrum, including, of course, bloggers supporting the minority Conservative government. In comparison to the general election of 2005/2006, that saw considerable political banter and posturing across partisan blogs [7], we felt that the leadership campaign would evoke distinct electoral and partisan dynamics not found in general elections. We hypothesized that Liberal party members, activists, and bloggers would use the leadership campaign to regroup and rethink their Internet and communications strategies since their defeat at the polls some nine months earlier in 2006. Conversely, we anticipated that other opposition parties and supporters of the minority Conservative government would use the opportunity to embarrass the Liberal’s leadership hopefuls, in addition to testing their own strategies for the upcoming general election. Plans for the party’s leadership convention in December 2006 supported our hypothesis. For the first time in Canadian politics, bloggers were given official media credentials at the Liberal party convention [8]. A “bloggers room” was also established metres away from the convention floor. The choice of former American presidential hopeful Howard Dean, as keynote speaker for the Liberal’s leadership convention, was another indication that the party might be looking to conduct a vigorous campaign on the Internet. Dean’s effort to secure his party’s nomination in 2003 and 2004 is widely cited — along with other partisan Web sites like moveon.org (http://moveon.org/) — as significantly uploading campaign organizing and fundraising onto the Internet (Trippi, 2004).
Research questions
Our research focused on a series of questions that sought to determine, in broad terms, the structure of the blogosphere — specifically those blogs that frequently discussed the leadership campaigns for the Liberal Party of Canada. We were primarily concerned with determining the influence and partisan affiliations — if any — of bloggers in the political context previously outlined above. Bloggers and other non–professional journalists had made a significant impact on American media culture and politics for some years [9], Matt Drudge’s gossipy Web site being among the more popular and influential [10]. The same, though, could not be said of Canada. The slow emergence, however, of organized rolls or networks of political bloggers in 2005 (Liblogs, Blogging Tories, NDP dippers, etc.) suggested that a select few blog organizers and opinion leaders were beginning to emerge. Thus at the outset of the study, about three months before the party would select a leader through elected delegates at a national convention in Montreal, we set ourselves the task of determining which blogs had the most influence over the course of the campaign and on the terms and issues contested within the campaign. Were they all Liberal party operatives? How would other opposition parities and the government’s supporters intervene in the blogosphere? In short, we sought to determine the partisan nature — and affiliations — of the most influential bloggers.
Method
At the beginning of our study in late August 2006, as the leadership race was nearing the selection of delegates for the December convention, we sought out a means of collecting weekly blogger posts. The broader Code Politics project was designed to track — and disseminate — on a weekly basis trends in hyper–mediated code politics during the Liberal leadership race. This subset of research on bloggers, particularly questioning the influence of bloggers, was in part driven by the need to identify a small but consistent data sample from which we could perform larger and more focused sets of analysis. As a simple indicator of the quantity of blog posts on the leadership race we first tracked, as a raw count, the number of blogs posts per week that mentioned a specific candidate by name as ranked on the Google blog search engine (see Figure 1).
Figure 1: Liberal leadership weekly blog coverage.
While the reliability of such results were unverifiable (coming from a proprietary Google algorithm), the chart nonetheless piqued our interest in determining which bloggers were making the most influential contributions to these trends in the blogosphere. Given that we saw markedly different coverage for the candidates, we also sought to determine which bloggers were formally or informally associated with leadership campaigns and candidates. An offshoot of this more qualitative research involved coding and tracking (on a weekly basis) partisan opinions expressed in the blogosphere toward the top four candidates (Michael Ignatieff the perceived front runner over the course of the campaign, Bob Rae, Stephane Dion, and Gerard Kennedy). Findings from this research are, however, still being analyzed and prepared for publication. For our purposes here, though, we should note that the data set we collected for this other sub–project was also used in constructing a ranking of bloggers, discussed herein.
Data was collected using the Google blog search engine (http://blogsearch.google.com). Each week the exact name for each of the top four candidates was used as a search query. The top 100 most relevant results per candidate were saved as an RSS [11]. We then imported these top blogger posts for each candidate into a .csv (comma separated value) spreadsheet format. Once transferred, the blog results were parsed into a number of headings, most notably, name of blog, teaser text, including our search terms highlight (candidate name), and date, etc. The Google blog search engine was chosen over others as we found it contained considerably more results (about five times the posts) than other leading search engines such as Technorati [12]. Google blog search also let us clearly demarcate our weekly search parameters. For the week ending 17 November 2006, we compiled a first top five bloggers list, ranked by number of posts within our data sample, from the sum of our four top 100 samples. Thus, the number of posts in not a reflection of the total output of the author, but the number of times the author’s post were present in our sampling method. After compiling the list (see Table 1), we then visited each site manually to code for party affiliation (e.g. Liberal, Conservative, Green, NDP, etc.), leadership candidate endorsements, and blogger networks (Liblogs, Blogging Tories, NDP dippers, etc.).
Table 1: Top five liberal leadership bloggers, with affiliations Rank Blogger (Affiliation) Number of posts 1. Paper Dynamite Online Print Archive by Peter Wrightwater
(No affiliation)17 2. Fuddle–duddle by Rousehouse
(Ignatieff, Liblogs, progressive bloggers)14 3. While the Earth Burns by Jeremy Kirouac
(Dion, Liblogs)7 4. The Adventures of Diva Rachel by Diva Rachel
(Kennedy, Liblogs, progressive bloggers)6 Bigcitylib Strikes Back by Bigcitylib
(no candidate affiliation, Liblogs, progressive bloggers)6 Dissonance and Disrepect by Loyalist
(Blogging Tories)6 Keith Torrie Today by Keith Torrie
(Iggy, Liblogs)6 Winnipeg Grit by Winnipeg Liberal
(no candidate affiliation, Liblogs)6 5. The Alberta Spectator by Werner Patels
(Kennedy)5 Cherniak on Politics by Jason Cherniak
(Liblogs administrator, Dion)5 Liberal Lite by Philly Cougar
(Kennedy, Liblogs)5 Political Staples by unknown
(Blogging Tories)5
In addition we produced a ranking of blogs for the entire campaign (from end of August to end of November, 2006, ending the Friday, four days before the leadership Convention began). For the overall campaign the ranking method remained the same, but the data set we used consisted of Google blog search RSS output from the entire campaign, amalgamated into one large file, and subsequently ranked by blog name as in Table 1. Once the blogs had been ranked with this method the top sites were again visited by the research team in order to determine their affiliations. (Table 2 and Figure 2).
Table 2: Top ten bloggers for entire campaign [13] Rank Blogger (Affiliation) Number of posts 1. Conservative Party of Canada — Reality Check 102 2. Cherniak on Politics:
(Liblogs administrator, Progressive Bloggers co–founder, Dion)85 3. Cerberus:
(Ignatieff)66 4. Liberal Outsider:
(Kennedy)59 A bcer in Toronto:
(Dion)59 5. Libnews.ca:
(No affiliations)53 calgarygrit:
(Progressive Bloggers co–founder, Kennedy)53 6. Paper Dynamite Online Print Archive:
(No affiliations)49 7. TDH Strategies:
(Progressive Bloggers co–founder, Ignatieff)48 8. Liberal Lite:
(Kennedy)46 altavistagoogle:
(Ignatieff)46 9. Werner Patels: Musings:
(Kennedy)43 10. The John Lennard Experience:
(Bob Rae)42 11. The Adventures of Diva Rachel:
(Kennedy)40 blog rae:
(Rae)40 democraticSPACE/blog:
(Kennedy)40 Jack’s Newswatch:
(Conservative)40 12. Ramblings of a Northern Ontario Liberal:
(Ignatieff)39 13. Liberal Life and Times:
(Kennedy)38 14. political staples:
(Conservative)35 15. Red Tory:
(Ignatieff)33
Figure 2: Total blogger posts as a percentage of total sample posts.
Analysis of rankings
The most glaring findings from the first (and only) weekly ranking of Liberal leadership bloggers, was the title and format of the top “blog”, the “Paper Dynamite Online Print Archive”. Upon visiting the site it was obviously not the typical diary–based blog format, but rather a news archive and aggregator of sorts. Furthermore, there were no personal comments and replies from visitors of any kind. Similarly, once the rankings were circulated across the Web, including those sent to individuals providing e–mail addresses, the Paper Dynamite Webmaster e–mailed the project director to point out that his site was not a blog. This issue, of course, points to problems we encountered with the opacity of ranking mechanisms on the Web [14], a topic we will explore in greater detail below.
Overall we found that Kennedy (7) and Ignatieff (5) supporters dominated the top bloggers list, 13 out of 21 of which were Liberal party members [15]. Of those blogs that identified themselves as Liberals, only one, “Libnews.ca”, failed to publicly endorse a leadership candidate. The top ranked blogs were, in other words, dominated by partisan blogs each pushing their respective candidate. Only one candidate’s blog was ranked, the former NDP premier of the province of Ontario, Bob Rae [16].
Partisan opposition to the Liberal party in our ranking bloggers was dominated by Conservative party members and supporters, a significant findings given that nearly half of the Canadian parliament is represented by members of Parliament from the Bloc Québécois and the New Democratic Party. The rising influence of the Green party in Canadian politics was also not evidenced in the top rankings of blogs. The inclusion of the Conservative’s “reality checks” site also clearly suggests their messages were the source of considerable discussion and debate within the blogosphere.
In terms of the main Liberal party bloggers on our list (Cherniak, Cerberus, and Calgary Grit), there are other indicators beyond our ranking that clearly suggests these bloggers served as opinion leaders for the party in the blogosphere. These three bloggers have provided not only resources for blogging (Web code and graphics images for party bloggers, but also small–scale research and party intelligence). The Cerberus blog, for instance, posted a running count of endorsements for the candidates [17]. Cherniak, a visible supporter of Dion (the eventual dark horse winner of the leadership race), received substantial media attention before, during, and now after the party’s convention, founded the Liblogs network of bloggers. Not surprisingly, his high profile and organizational history in the Liberal party blogosphere led him to question and try to determine his blog’s own influence on the outcome of the convention:
I can’t sleep, so I have some questions. I am trying to figure out what really happened on the convention floor and whether my blog had anything to do with it. If you have an answer to any of the following questions, I would greatly appreciate it. Private email will remain 150% private. Anonymous email or comments are also fine, but obviously anonymity makes it difficult to know whether you are telling the truth.
...[questions deleted]
(Yes, I recognize how silly this might be. Just laugh along with me and provide answers if you have them. I have considered not mentioning the blog part. The problem is that the blog part is the real research. This isn’t about whether I helped Dion win — it’s about whether blogs in general can have an effect beyond being an interesting read.) [18]
Lastly, in terms of blog networks (or rolls), those lists of blogs that seem to grow exponentially every week, our research was unable to determine the relationship between the partisan political rolls, and the more generic politically inclined rolls, such as “progressive bloggers”, organized by “Calgary Grit”. Though, as an aside, co–founders of “progressive bloggers” repeatedly told the press in the “bloggers room” that they were non–partisan and neutral observers at the leadership convention. It remains to be seen whether or not as we enter the more partisan general election cycle that such non–partisan networks begin to fade away in favour of more explicit politically affiliated rolls such as “Liblogs, Blogging Tories or NDP dippers”.
A consideration of ranking methods
While our blog ranking research revealed the evolution of the campaign online, our methods also raised questions about Web aggregators, in particular the Google blog search engine. Despite the fact that Google blog search provides the greatest number of results, Google’s proprietorial algorithm remains somewhat of a mystery. While some early discussions of the “page rank” search algorithm exist (Brin and Page, 1998), researchers continue to deconstruct the algorithm, testing search engine results for signs of potential bias (Henzinger, 2001; Vaughan and Thelwall, 2004). While other researchers have avoided Google, their reliance upon other search engines and aggregators should also be questioned. The ranking system of Technorati, for example, a blog search engined used in Drezner and Farrel’s (2004) discussion of blogger inluence and agenda setting research, also relies heavily upon a sole indicator of relevance, reputation, and influence: inlinks. And again, like Google, it is difficult to get exact information on Technorati’s algorithm. Rather, one tends to find general statements, like the one below from a Technorati employee posted on the search engines blog, qualifying their measurement of inlinks:
Technorati determines a blog’s ranking based on the number of links from unique blogs over the last six months. If John links to Susie five times in five months Technorati will count five new links to Susie from one unique source. Susie’s rank will change based on this one new source. [19]Search engines like Technorati or Google though remain one of the few avenues for general, large–scale blog research. And while one may critique search engine algorithms, manual or random sampling of blogs could also easily produce samples that reflect and reproduce the biases of the researcher and project. Drezner and Farrel’s (2004) relatively early research on American blogs, however, offers a helpful hyper–mediated method that encompasses their use of the Technorati blog search engine, that is to say they study the influence of bloggers across online formats — both blogs and news sources. Drezner and Farrel’s research questions both the agenda setting functions of bloggers (e.g. are there a select few influential bloggers?) and their impact on the broader mass media, referred to in the vernacular as bloggers “breaking the story”. Thus while the following Technorati–derived chart identifies influential blogs (represented by circles) through measuring total inlinks, Drezner and Farrel enhance these findings with methods that determine which blogs are both read or cited by the mass media.
Figure 3: Drezner and Farrel’s (2004) visualization of blog influence and distribution,
where degree is the number of incoming links and rank refers to the rank of the blog overall.
Thelwall and Stuart’s (2007) use of Nielsen’s Blogpulse search engine by comparison takes advantage of proprietorial info–tools that allow for the diachronic analysis of specific keywords in blog posts (tracking the discussion of London subway attacks, Katrina, etc. in the blogosphere). And while Thelwall and Stuart do not voice any concerns about the inability to know the exact ranking method Blogpulse employs, they do highlight another significant limitation in the use of search engines such as Google, Technorati and Blogpulse, the frequency with which general research keywords return substantial off–topic results. van Doorn, et al. (forthcoming), by comparison, offer a topic focused approach, shared in part by Foot and Scheider (2006): “One alternative to content relevance [through the use of search engines and keywords] as the primary criteria for inclusion in the Web sphere ... is the inclusion of Web sites produced by actors identified as relevant to the sphere.” [20] While still collecting their blog sample from an aggregated source, blog portals (from the Netherlands and Belgium), the focus of van Doorn, et al. on national and language specific sites narrows their data sample, and consequently offers an ability to make stronger conclusions about the demographics under analysis.
While each of these three studies will influence the future direction of our project as a whole, as we move from relying upon any one single aggregator or measurement of relevance, authority, reputation of influence, there is still an argument to be made in support of employing blog search engine aggregators, especially the widely used Google and Technorati search engines. As an industry leader in information aggregation there is some validity in using the Google results as a sample. Roughly half of Internet users use Google for their search engine queries [21]. As a consequence one could argue the major search engines, including Google, offer a “representative” view of how a large proportion of Internet have Internet content — including blogs — formatted for them by the main online information aggregators. Furthermore, in our own research, we found that “glitches” or anomalies in Google’s algorithms turned out to be very instructive. For example, the inclusion of “Conservative Party Reality Checks” by Google — the online war room for the Conservative party that posted responses and retorts to issues arising from the Liberal’s leadership campaign ” in its blog search engine, while raising some valid questions about their sample, nonetheless provides valuable insight into how the blogopshere was strongly hyperlinked into the Conservative party’s Web site. One might conclude that the heavy inlinking from the blogosphere was in part instrumental in including the Conservative party’s Web site in Google’s blog search engine, though this is only speculation at this stage. On the whole, though, while we believe there is some merit to these arguments in favour of using large information aggregators to collect blog samples, we are not convinced that it adequately addresses the bias of the sample, and moreover, fails to integrate a hyper–mediated perspective that our study calls for (compared against online news, Web pages, etc.).
Proposed methods for Stage II: Federal election campaign
Our methodological concerns could perhaps be summarized by asking the simple question — what did our research actually measure? We might also have questioned the limitations of obtaining data by focusing only on leadership names as search terms. Looking forward, though we would prefer to determine how we could perhaps triangulate multiple methods and measurements, some already identified above, to develop a broader index of blog influence. Given the amount of commentary herein focused on Technorati and Google, it should come as no surprise that for our next study of the Canadian general election we are looking to expand our possible data samples. Unlike Foot and Schneider (2006), whose studies of Web campaigning in the United States faced a large complex set of Web actors and bloggers, the Canadian political blogosphere is surprisingly structured along party lines. The success of the Liblogs roll has been emulated by the other parties offering ready–made lists of partisan bloggers. Other data samples we are considering include so–called professional or journalist bloggers. And again, as opposed to the U.S. or western Europe, Canada has relatively small number of national newspapers and mainstream political magazines. The ability to cull together a list of national journalist–bloggers then is neither difficult nor time consuming. Blogger specific networks could then be juxtaposed against more general Google or technocrati produced lists of blogger posts. Furthermore, as an addendum or complement to inlink measurements, we are proposing to aggregate the “blogs that I read” blog roll, to further refine and qualify the use of inlinks, which we consider to be far too general a measurement of influence. We are therefore looking to distinguish links. So far, we have identified: links to posts (suggesting recommended or valued commentary/posts), inlinks from blog networks (suggesting memberships and associations), inlinks from commercial aggregators and advertisers/Web metric companies, and the aforementioned “blogs that I read” inlinks (which themselves can be broken down by type of bloggers: partisan, professional, journalist, political opponent, etc.). As we begin to develop a typology of links (see Table 3), our methods to determine influence will become much more precise and grounded in specific communities that have distinct dynamics and purposes.
Table 3: Typology of electoral inlinks Type of inlink Measurement Inlinks from partisan networks Political association and engagementk Inlinks from “blogs I read” Influence/opinion leaders Inlinks from advertisers Professional/revenue aspirations Inlinks from party/candidate sites Possible paid political association Inlinks from blog networks Possible neutral or pundit blogger
In addition to a qualification of links as indicators of influence within a federal election, the quantification of blog “comments” might also provide a more precise indicator of readership and influence within said blogosphere, or at the very least indicate a deeper sense of engagement and interactivity with a respective blog. Merely studying the amount of daily or weekly posts would clearly not provide this level of detail into the workings and communicative characteristics of blogs.
Moving to the hyper–mediated aspect of our research project, we also anticipate developing a blog influence index that takes into consideration posts, stories, and names of blogs that are cited in the mainstream press, much like the early work of Drezner and Farrell. Comparisons of top political issues in the newsphere will also be juxtaposed with the most frequently mentioned issues in our election blogosphere results. And lastly, as part of our cartography of blogs, we anticipate charting blog locations (province, or rural/urban split) in an effort to determine if rural, politically marginalized regions of the country, or so-called “battleground” ridings and provinces, see heightened blogging activity. This issue was of particular interest to the leadership portion of our project as the costs of attending the leadership convention for party members living in remote regions were prohibitive.
Lastly, the project is developing an “issue geography” that measures the proximity between the partisan blogger networks and official party websites. This aspect of the project will determine — through content analysis that will produce frequency of terms (social and political issues, phrases, platforms, candidate and party names) — which partisan blog networks stray the most from the official party platforms and communications from their respective parties.
Conclusions
Our early research has concluded that the blogosphere is indeed a dynamic and complex set of relationships and networks, exhibiting distinct patterns. Our initial research clearly suggests that Canada’s two largest parties, the Conservatives and Liberals informally and formally dominated the political blogosphere throughout the Liberal leadership race. It remains to be seen, however, whether the other mainstream political parties in Canada replicate these organizational strategies across the blogosphere and whether the number of blog opinion leaders grows in both numbers and political affiliation.
Political forms of analysis, particular in a small (population wise) country such as Canada provides a particularly insightful case study as partisan networks are seemingly more focused and defined. Given the immense geography of the country one might also assume, though it has yet to be proven, that the vast expanse of the nation might produce a ready–made topography for Internet communications and campaigning, particulary in light of the highly centralized political system and requisite need to centralize or otherwise organize remote and disparate party members, campaigns, and messages.
Moving forward, we identified three areas to develop in our upcoming study of the impending Canadian general election. First, in an effort to make the collection and characteristics of our data samples more transparent and our conclusions more verifiable, we are proposing to both expand our samples and rank blogs through a more complex index. Of course these two goals will confront the research team with an all too familiar dilemma, that of inefficient and time consuming human coding and analysis on the one hand, and the often problematic automation of these processes on the other [22].
About the authors
Greg Elmer is the Bell Globemedia Research Chair in the Infoscape Research Lab (http://www.infoscapelab.ca) at Ryerson University in Toronto. Peter Malachy Ryan, Zach Devereaux, Ganaele Langlois, Joanna Redden and Fenwick McKelvey are graduate students in the Infoscape Research Lab at Ryerson University.
Correspondence to Greg Elmer: gelmer [at] ryerson [dot] ca
Acknowledgments
The research was funded in part by grants from the Social Science and Humanities Research Council of Canada (SSHRC), the Canadian Media Research Consortium, and the Bell Globemedia Research Chair, Ryerson University. The paper was also written with the support of a visiting faculty fellowship at the Royal Dutch Academy of Arts and Sciences (Virtual Knowledge Studio).
Notes
1. http://www.infoscapelab.ca.
2. Weekly analyses form this first stage, in addition to a final report, and these are available at http://www.infoscapelab.ca/gritrace.
3. For a brief history of html see MIT’s World Wide Web Consortium (W3C) at http://www.w3.org/MarkUp/historical.
4. Matt Fuller’s (2005; 2003) attempts to define “software studies” as a field of study, Adrian MacKenzie’s (2006) discussion of “cutting code” or Richard Rogers’ (2004) discussion on research tools and “information politics” are among the most insightful exceptions.
5. Web objects can generally be defined as snippets of modular code that can be easily copied, pasted, and adapted for use within Web sites. Object technologies come in four common forms at the moment: 1) scripts (e.g. Java and PHP); 2) embedded multimedia components (e.g. applets and plug–ins); 3) Web server components (e.g. APIs); and, 4) intelligent agents and mobile codes (e.g. client side executed code instructions). Objects are meant to increase intelligent use of the Web because programmers can use them over and over again once an object has been created, thus the modular nature. For more on object technologies, please see the W3C’s Web site on Web objects and intelligent agents: http://www.w3.org/OOP/ (2007).
6. The Liberal party’s ideological leanings and policy agendas are often compared to the Democratic party in the U.S. or the British Labour party under Tony Blair.
7. The Canadian Broadcasting Corporation provided an insightful tracking of blogger activity during the campaign at http://www.cbc.ca/canadavotes/analysiscommentary/blogreport.html.
8. Bloggers were given credentials at both the Republican (http://www.conventionbloggers.com/) and Democratic conventions in 2004; see http://edition.cnn.com/2004/TECH/internet/07/23/conventionbloggers/.
9. A 2004 Pew Internet report on the role of the Internet in American politics noted that top blogs such as Daily Kos and Instapundit had already overtaken political magazines such as the New Republic as sources of political news; see http://www.pewinternet.org/PPF/r/151/report_display.asp. See also Keren’s (2006) overview of political blogs and their impact on American political culture.
10. Drudge was reportedly the first outlet to break the Monica Lewinsky–Bill Clinton scandal; see “Scandalous scoop breaks online,” BBC News at http://news.bbc.co.uk/1/hi/special_report/1998/clinton_scandal/50031.stm.
11. A particular example of a common Web object used today is application programming interface (API) codes that can be used to analyze Really Simple Syndication (RSS) feeds, a popular form of syndicating blogs and other newswire services. RSS is a common euphemism for what was originally called Resource Description Framework (RDF) Site Summary, a standardized format “to give the gist of the site and enable syndication” (Hjelm, p. 97). RDFs are very simple eXtensible Markup Language (XML) grammars that provide a defined structure to enable better sharing of Web site documents over the Internet. XML is, put simply, a refinement of html code that helps increase the useful description of Web documents using metadata (or data about data). Metadata provides a computer extra information about any particular XML–encoded document. For example, RSS feeds have a common XML tagging structure that uses the code “<author>” to provide search engines with easy searching of the “authors” of RSS feed delivered articles. Not unlike html, an author such as “Johan Hjelm” would be consistently coded as “<author> Johan Hjelm<author>” for a document that he wrote on a blog. An API code can then be written to automatically search for this tag on RSS feeds, if one was interested in tracking a specific author like “Johan Hjelm”.
12. Technorati has since apparently improved its database. A search for the “Liberal Party of Canada” in both Technorati and Google blog search returned roughly the same for both search engines — 38,000 posts (22 January 2007).
13. Top fifteen bloggers. 25 September — 26 November 2006: Over 4,600 blog posts were collected during this period and were analyzed as a total sample to produce this ranking.
14. http://technorati.com/weblog/2005/09/50.html.
15. This was a very small issue for our study as our keyword searches employed relatively uncommon names. Only “Stephane Dion”, the eventual winner, the names of the leadership hopefuls were not.
16. http://searchenginewatch.com/showPage.html?page=2156451.
17. http://canadiancerberus.blogspot.com/2006/09/liberal-leadership-blogger_28.html.
18. From Cheniak on Politics, 6 December 2006.
19. http://technorati.com/weblog/2005/09/50.html.
20. van Doorn, et al., forthcoming, p. 34.
21. http://searchenginewatch.com/showPage.html?page=2156451.
22. Given time and space constraints this paper has not adequately discussed the role that experimental info–tools played in the analysis discussed herein. For a short discussion of our tools please see Elmer (2005).
References
Espen J. Aarseth, 1997. Cybertext: Perspectives on ergodic literature. Baltimore, Md.: Johns Hopkins University Press.
J. David Bolter, 1991. Writing space: The computer, hypertext, and the history of writing. Hillsdale, N.J.: L. Erlbaum Associates.
Sergey Brin and Larry Page, 1998. “The anatomy of a large–scale hypertextual Web search engine,” Computer Networks and ISDN Systems, volume 30, number 1, pp. 107–117. http://dx.doi.org/10.1016/S0169-7552(98)00110-X
Daniel W. Drezner and Henry Farrell, 2004. “The Power and Politics of Blogs,” paper presented at the 2004 meeting of the American Political Science Association, Chicago (1–4 September), at http://www.utsc.utoronto.ca/~farrell/blogpaperfinal.pdf.
Greg Elmer, 2006. “The vertical Net,“ In: David Silver and Adrienne Massanari (editors). Critical cyberculture studies. New York: New York University Press.
Greg Elmer, 2005. “Re–tooling the network: Parsing the links, codes, and commands of the Web world,” Convergence: The International Journal of Research in New Media Technologies, volume 12, number 1, pp. 9–19. http://dx.doi.org/10.1177/1354856506061549
Kirsten A. Foot and Steven M. Schneider, 2006. Web campaigning. Cambridge, Mass.: MIT Press.
Matthew Fuller, 2005. Media ecologies: Materialist energies in art and technoculture. Cambridge, Mass.: MIT Press.
Matthew Fuller, 2003. Behind the blip: Essays on the culture of software. Brooklyn, N.Y.: Autonomedia.
Monika R. Henzinger, 2001. “Hyperlink analysis for the Web,” IEEE Internet Computing, volume 5, number 1, pp. 45–50. http://dx.doi.org/10.1109/4236.895141
Johan Hjelm, 2001. Creating the semantic Web with RDF: Professional developer’s guide. New York: Wiley.
Philip N. Howard, 2006. New media campaigns and the managed citizen. Cambridge: Cambridge University Press.
Michael Keren, 2006. Blogosphere: The new political arena. Lanham, Md.: Lexington.
George P. Landow (editor), 1994. Hyper/text/theory. Baltimore, Md.: Johns Hopkins University Press.
Adrian MacKenzie, 2006. Cutting code: Software and sociality. New York: Peter Lang.
Richard Rogers, 2004. Information politics on the Web. Cambridge, Mass.: MIT Press.
Joe Trippi, 2004. The revolution will not be televised: Democracy, the Internet, and the overthrow of everything. New York: ReganBooks.
Mike Thelwall and David Stuart, 2007. “RUOK? Blogging communication technologies during crises,” Journal of Computer-Mediated Communication, volume 12, number 2, at http://jcmc.indiana.edu/vol12/issue2/thelwall.html. http://dx.doi.org/10.1111/j.1083-6101.2007.00336.x
Niel van Doorn, Liesbet van Zoonen, and Sally Wyatt, forthcoming. “Writing from experience: Presentations of gender identity on Weblogs,” European Journal of Women’s Studies.
Liwen Vaughan and Mike Thelwall, 2004. “Search engine coverage bias: Evidence and possible causes,” Information Processing and Management, volume 40, number 4, pp. 693–707. http://dx.doi.org/10.1016/S0306-4573(03)00063-3
Editorial history
Paper received 27 February 2007; accepted 15 March 2007.
Copyright ©2007, First Monday.
Copyright ©2007, Greg Elmer, Peter Malachy Ryan, Zach Devereaux, Ganaele Langlois, Joanna Redden, and Fenwick McKelvey.
Election bloggers: Methods for determining political influence by Greg Elmer, Peter Malachy Ryan, Zach Devereaux, Ganaele Langlois, Joanna Redden, and Fenwick McKelvey
First Monday, volume 12, number 4 (April 2007),
URL: http://firstmonday.org/issues/issue12_4/elmer/index.html
