Information Literacy Gone Bad Report
Lisa Levesque and Emily Shearer
April 7th, 2016
The purpose of this study was to create a website profile through a close examination of external inlink data. The website that we have chosen to profile is martinlutherking.org, a notorious hate propaganda site that disguises itself as educational. A close analysis of the content of external, inlinking sites was used to determine how martinlutherking.org is perceived on the web and the extent to which its posturing has been successful.
The inlink profile of martinlutherking.org will be composed of webpages that are either aware that this is a racist website or unaware of this fact. The content of these inlinking pages can be said to link to it either critically, uncritically, or with racist content.
Hyperlink analysis is a common webometric measure used to study and characterize the connections that exist between websites. Studies that examine hyperlinks have been able to use them as a measure of website visibility (Gao & Vaughan, 2005), latent political affiliations (Romero‐Frías & Vaughan, 2012), the connectivity of communities (Kenekayoro, Buckley & Thelwall, 2014), collective online identity (Ackland & O’Neil, 2011) and online recommendations (Bowler, Hong, & He, 2011). Hyperlink analysis was chosen for this study as an effective method for revealing online relationships between websites.
Hyperlinks are central to the function of the web, and their use in search engine ranking has been transformative. In their seminal paper describing the function the newly set up Google, Sergey Brin and Lawrence Page describe their hyperlinks as a variable in ranking results (1998). Comparing hyperlinks to academic citations, the number of inlinks to a site “gives some approximation of a page’s importance or quality.” This would be the innovation that set Google’s PageRank algorithm above its competitors (see appendix 4.1 for definition).
However, this analogy of hyperlinks to citations has been widely criticized, in part because of the potential for misunderstanding and manipulation. Seth Finkelstein describes highly ranked websites can confuse users who mistake popularity for authority. The founders of Google themselves admitted that ranking by hyperlink can lead to “complicated social effects” (2008). Finkelstein goes on to describe one such scenario in the context of hate propaganda. In 2004, the top search result on Google.com for “Jew” was the anti semitic website, Jew Watch (it is currently result 11). Google explained to upset users that their search results were the result of an algorithm and that they do not reflect a company bias. Jew Watch performed well in Google searches due to a number of PageRank boosting factors including “the longevity of ownership, the way articles are posted to it, the links to and from the site, and the structure of the site itself” (italics added, Anti-Defamation League, 2004).
While this particular controversy has subsided, controversies like this have taught hate groups “the power that comes from prominent placement in searches” (Finkelstein, 2008). The Internet is a powerful recruitment tool for hate groups (Lennings, Amon, Brummert, Lennings, 2010). The leader of Stormfront, a hate site that has been online since the mid 1980’s, has famously said that “their website has allowed them to reach more people and to recruit more people,” although these claims are unsubstantiated (Lennings et al., 2010). Stormfront is the owner of the domain of martinlutherking.org, which they use as a platform to disseminate hateful information about the famous civil rights leader Martin Luther King Jr.. The seeming legitimacy of the site, derived from its high rank in Google search results, confuses some viewers (Thomson, 2011). In response to concerns about this, Google has released a statement similar to the one they released about Jew Watch, stating that search results are “generated objectively and are independent of the beliefs and preferences of those who work at Google” (Thomson, 2011).
The Moz Open Explorer Tool was used to collect inlink data for martinlutherking.org. As a comparison, information was also collected about the website thekingcentre.org which is a legitimate website operated by The King Foundation, a charity established to protect the legacy of Martin Luther King Jr. Data was downloaded and visually inspected and coded. Human judgement was preferred for data analysis over that of a crawler. While a crawler could have been used to inspect these sites for keywords based on inspection criteria, it would have been inferior to human judgement in adding new criteria, recognizing every variation of wording, and discerning the genuine content of a webpage.
The Open Site Explorer Tool was used to collect the following data for each website: the total number of inlinks to each website and the total number of external websites linking in. Specific data collected was collected for links that: link to the root domain, are external, and are follow-type links. The top 100 pages that link in, according to Domain Authority (see appendix 4.1 for definition), were compiled and analyzed. The following information for each page was recorded: the title, URL, and Domain Authority.
The linking-in websites for both martinlutherking.org and thekingcenter.org were quantitatively analyzed using a coding schema, which was developed during the analysis based on re-occurring and stand-alone themes (Palys & Atchison, 2008). The standardized coding scheme for websites linking to martinlutherking.org consisted the content categories of Education, Racist, History, and Other (MLK Day, Travel, Quotes, and Domain Information). The coding scheme for thekingcenter.org included Resource Recommendations, Historical Descriptions, Writing and Speeches, and Other (Travel Guides, Martin Luther King Day, and Event Description). See appendix 1.1 and 1.2 for detailed coding descriptions for each website, and appendix 2.1 and 2.2 for compiled, coded data.
These categories were operationalized by creating subdivisions and explanations of the terms; for example, “History” was defined as related to the topics Biography, Civil rights and Trivia. It was found that many websites could be coded in multiple ways; therefore, the most dominant content of the site was identified and coded accordingly.
Coding for martinlutherking.org
Web pages linking-in to martinlutherking.org were individually visited and coded as either Critical, Uncritical, or Racist. These terms were operationalized before coding. A site was defined as Critical if it describes martinlutherking.org as an example of fraudulent site, or references it for the purpose of web literacy education (e.g. how to search, evaluate, and use digital resources on the internet) (University of Illinois, 2013). A site was considered Uncritical if it has a neutral tone (i.e. content that is neither positive or negative), such as referring to website as a reference without stating an opinion. A site is Racist if its content includes language that is discriminatory (eg. stereotypical, includes racial epithets), covers a racist theme (eg. the oppression of white Christians or the inferiority of a race), or provides links to a collection of racist sites.
Coding for thekingcenter.org
During evaluation of thekingcentre.org we did not encounter any inlinks from websites that contained racist content or that were critical of this site. All inlinks came from web pages that were uncritical of thekingcenter.org. Content categories for these web pages include: Recommended Resource, History, Writing/Speeches, and Other (which includes Travel, Martin Luther King Day, Event).
Statistics were generated for the categories and subdivisions for martinlutherking.org and thekingcenter for comparison. Pie charts were created for visual representation of the category proportions in relation to one another. Non-operational links were excluded from the 100 pages that link in, but noted at the end of the data set.
Comparison of martinlutherking.org and thekingcenter.org
As a comparison of the total number of links to these two websites shows, thekingcenter.org has far more total links (see Appendix 5.1). However, these are links of all types (internal, external, follow, no follow, etc.) and Moz Open Site Explorer does not allow users to tabulate the total number of external, follow-type inlinks.
Analysis of Martinlutherking.org
For the high level categories, 65% of the webpages were coded as Education, 21% History, 10% Other, and 4% Racist. Out of the total, 65% were critical, 31% were uncritical, and 4% were racist (See Chart in Appendix 3.1) All of the webpages that were critical of martinlutherking.org were classified as Education (100%). The webpages linking-in uncritically were largely those classified as History, constituting 68% of all uncritical sites. The category Other was 29%, and Education was 3%. The sites that link to martinlutherking.org uncritically were mainly comprised of blogs, forums postings, news articles, and other sites that included the inlink within biographical or historical content about King, and did so with a neutral tone. The results for the subdivisions consisted of 59% web literacy, 21% biographical, 6% awareness, 5% MLK day, 4% racist content, 2% domain information, and 1% for each travel, civil rights, and quotes. The majority of the inlinks were for web literacy or biographical purposes.
Importantly, the results show that the majority of webpages (65%) linking-in the martinlutherking.org were for were educational purposes, doing so critically and as part of web literacy education, or exposing martinlutherking.org as a racist site. A large proportion of these were university, college, and library websites (41%).
Although we had hypothesized this result, it is unexpected for the inlink profile of a racist website to be so highly composed of educational websites. The fact that 31% of inlinks came from websites with uncritical content show the need for better web literacy skills amongst Internet users.
Analysis of thekingcenter.org
The categories coded for thekingcenter.org had the following content: 33% History, 28% Other, 22% Writing, and 17% Recommended Reading. See Appendix 3.2. We did not encounter any inlinks from websites that contained racist content or that were critical of this site. All inlinks came from web pages that were uncritical of thekingcenter.org. As a comparison to martinlutherking.org, the inlinks for this website create a profile of a site that is used for legitimate research on Martin Luther King Jr.
Inlinks are a factor for PageRank, and so the fact that so many sites link in to martinlutherking.org may be affecting its ranking in search results. Sites are critical of martinlutherking.org should avoid linking to this site unless they wish to raise its ranking. This is an important consideration for universities, libraries, and educational sites who are linking to martinlutherking.org for web literacy puposes. Keith Thomson (2011) suggests a solution to avoid raising martinlutherking.org’s PageRank by linking-in. If websites wish to mention the site or link to it, they should do so by not hyperlinking the url or writing the link as “MartinLutherKing(dot)org”.
An-indepth anaysis of the sites linking-in to martinlutherking.org and thekingcenter.org allowed us to construct inlink profiles for both. For martinlutherking.org, the external, inlinking sites were found to include webpages that were aware of martinlutherking.org being a hate site, and sites that were unaware, possibly mistaking it for the authoritative site, such as those linking to it for MLK day celebration. Most sites that were linking-in were critical of the website.
Anti-Defamation League (2004). Google search ranking of hate sites not intentional. Retrieved from: http://archive.adl.org/rumors/google_search_rumors.html
Bowler, L. Wan‐Yin, H., Daqing, H., (2011). The visibility of health web portals for teens: a hyperlink analysis, Online Information Review, 35 (3), 443 - 470.
Brin, S. & Page, L. (1998) The Anatomy of a Large-Scale Hypertextual Web Search Engine. In: Seventh International World-Wide Web Conference (WWW 1998), April 14-18, 1998, Brisbane, Australia. Retrieved from: http://ilpubs.stanford.edu:8090/361/
Finkelstein, S. (2008). Google, links, and popularity versus authority. In Turow, J. & Tsui, L. (Eds). (2008). The hyperlinked society: Questioning connections in the digital age (104- 120). Michigan: University of Michigan Press. Retrieved from: http://quod.lib.umich.edu/cgi/t/text/text-idx?c=nmw;;idno=5680986.0001.001.
Hallam, A. Duality in consumer theory [PDF document]. Retrieved from Lecture Notes Online Web site: http://www.econ.iastate.edu/classes/econ501/Hallam/index.html
How Stuff Works. (2016). Why is the Google algorithm so important? Retrieved from http://computer.howstuffworks.com/google-algorithm1.htm
Kenekayoro, P., Buckley, K., & Thelwall, M. (2014). Hyperlinks as inter-university collaboration indicators. Journal of Information Science, 40(4), 514-522. doi:10.1177/0165551514534141
Lennings, C. J., Amon, K. L., Brummert, H., & Lennings, N. J. (2010). Grooming for terror: The internet and young people. Psychiatry, Psychology and Law, 17(3), 424-437. doi:10.1080/13218710903566979
Patel, N. (2014). What is a “good link profile” and how do you get one? Quicksprout. Retrieved from: https://www.quicksprout.com/2014/04/18/what-is-a-good-link-profile-and-how-do-you-get-one/
Palys, T., & Atchison, C. (2008). Research decisions: Quantitative and qualitative procedures, 4th ed. Print.
Roberts, K. F. (1998). Federal regulations of chemicals in the environment [PowerPoint slides]. Retrieved from http://siri.uvm.edu/ppt/40hrenv/index.html
Romero‐Frías, E., & Vaughan, L. (2012). Exploring the relationships between media and political parties through web hyperlink analysis: The case of spain. Journal of the American Society for Information Science and Technology, 63(5), 967-976. doi:10.1002/asi.22625
SEOmoz, Inc. (2015). External links. Retrieved from https://moz.com/learn/seo/external-link
Thomson, K. (2011). White supemacist site martinlutherking.org Marks 12th Anniversary. Retrieved from http://www.huffingtonpost.com/keith-thomson/white-supremacist-site-ma_b_809755.html
University of Illinois. (2013). Insider’s guide to the library. Retrieved from http://www.library.illinois.edu/learn/about.html
Yijun G, & Liwen V. (2005). Web hyperlink profiles of news sites: A comparison of newspapers of USA, Canada, and China. Aslib Proceedings, 57(5), 398–411. http://doi.org/10.1108/00012530510621851