When you look for controversies, search where collective life gets most complex: where the largest and most diverse assortment of actors is involved; where alliances and opposition transform recklessly; where nothing is as simple as it seems; where everyone is shouting and quarreling; where conflicts grow harshest. There you will find the object of the cartography of controversies. (Venturini, 2010, p. 262)

Identifying debates in a corpus of over 4 million words limits the practicality of manually reading and identifying debates. Thus, we developed a 3-stage research protocol to streamline this process: 1) Searching for “disagreement terms” in the corpus, 2) Using the disagreement terms to extract quotes articulating debates in the corpus, 3) Organizing and visualizing issues, positions, and actors into DebateGraph.

Figure 1. The Concordance of Disagreement Terms used to search for debates in the corpus

A total of 1,824 “disagreement terms” were identified in the corpus. Additional “disagreement terms” could have been added to the list (e.g. differ, suggest, concern), however, 1,824 terms provided sufficient text to analyze and adequately identify major debates in the corpus. Importantly, this process allowed the actors to lead us to the debates in the corpus by following where they identify disagreements or controversies.

Each “disagreement term” entered into the Microsoft Word “find” tool generated between 6-420 unique instances and each was manually analyzed, reading several sentences before and after the identified term to understand the context. If the text surrounding the “disagreement term” was directly related to e-waste and specified a position or argument within the debate (not just a general comment stating that a debate exists) a selection of the text preceding and following the “disagreement term” was copied and pasted into a “corpus of quotes”.  As little information preceding and following the “disagreement term” was included to create succinct quotes that accurately articulate the context of the argument. Quotes were then “cleaned” to remove jibberish text and restore broken words resulting from website text-rips. The meta-actor (categorized via URL) and source document (originating URL) was included below each quote to allow users to “step off the controversy map” and trace quotes back to their original text.

A total of 251 quotes were extracted from the corpus using the “disagreement terms”. We then filtered the “corpus of quotes” into a more manageable “index of issues” to map in DebateGraph (described below), based on the following criteria:

• Originally selected quotes were refined, and, at times, disaggregated, to articulate individual arguments to ease the categorization of quotes into DebateGraph
• Quotes that did not explicitly state points of disagreement or did not clearly articulate an issue or position in the controversy were not included in the index of issues.
• Arguments that were replicated in several quotes we eliminated. The quote that was retained was closest to the original source, the best articulation of the argument and, the actors are explicitly stated.

This filtering process produced an “index of issues” consisting of 104 streamlined quotes that explicitly identified issues and/or positions in the controversy. The organization and visualization of the “index of issues” was portrayed in DebateGraph.org, which is a free web-based mindmapping tool, primarily used to visualize complex public policy issues.

The DebateGraph mindmap was divided into two core categories: Issues and Actors. The mapping of issues used the “original floating statement” (“E-waste is exported largely for the same reason manufacturing jobs have been sent overseas: lower labor costs and fewer regulatory burdens”) as a guiding point-of-origin to direct the organization of quotes added into the DebateGraph. Quotes were added in as either “issues”, “positions”, “supportive argument”, or “opposing argument”. The key issues identified emerged iteratively as we analyzed each quote, often quotes that would begin as an issue, and would later be better categorized as a position speaking to a larger issue. People or institutions identified in the index of issues were created as “protagonist” nodes and linked to positions with descriptors such as “informs”, “supports”, “opposes” indicating their relation to the debate.