Please do not cite or circulate the printed version of this HTML document. Please refer instead to the original online version of this document, which is available online at , or contact the author(s).

Research Report

Doing Search as Research

Mapping Localisation and Issue Maturity with Search Engines

Tessa de Keijser, Fernando N. van der Vlist [alphabetical]

(Graduate School of Humanities,) University of Amsterdam, the Netherlands

Working paper

Published online: 1 February 2014

Abstract

We report on whether and in what way Google's device-specific definition of the “local” may be used to perform cross-country and comparative media analysis with local domains of Google. We suggest to think of search engines in a framework of actor-network-theory which allows us to focus on the negotiability of the spaces they demarcate, and how both human and non-human actors operate side by side to achieve this. Search as research is a valuable research practice to look at contested issue spaces, such as the issue of euthanasia in Web search spheres. We wondered to what extent different socio-cultural and legal attitudes towards euthanasia are apparent in the types of sources and actors that show up in the search results for each of the selected local Google domains, and how far these different source types or actors are removed from the “top of the Web”. Can we use these measures to understand the extent to which that issue has matured? To operationalise our research questions, we first undertook a qualitative content analysis to understand local discourse and to be able to identify patterns and anomolies across countries. From there, we then did cross-country comparisons of thee local issue spaces and found that the current state of the debates are mostly about the legislative dimensions, ethical dimensions, and informative, “neutral” resources such as studies and reference sources. Overall, we identified legislative, ethical and oppositional concerns as the most frequent categories, but even though the same categories show up for each of the local domains in our source set, they do not necessarily refer to the same sources. Finally, we conclude that the “local” is in fact a negotiated space between the users, through Google's algorithms, and that in order to determine issue “maturity” we would need to study multiple social snapshots or “present states” of the “movements” that characterise the social over time.

Keywords

Internet research, digital methods, search engine studies, query design, issue maturity, Google, euthanasia

Internet research
digital methods
search engine studies
query design
issue maturity
Google
euthanasia

1. Introduction

We explore and report on how we could analyse search engine results for certain social research purposes. More specifically, we enquire whether and to what extent Google's device-specific definition of the “local” may be used to perform cross-country and comparative media analysis with local domains of Google. Rogers (2013) suggests to use search as research as a form of search research for which we rely on Google's capacity to serve as a research machine (100). In other words, we do “social studies via or on top of engines” (111). The research conducted here should thus be seen as part of the broader frames of Internet research and Web search studies in particular. The specific deployment of methods, devices and techniques that characterise this research situate it in the line of work on “digital methods” (Rogers 2013). We use medium-specific approaches that “follow the medium” by repurposing the devices and techniques of the Web, in order to be able to make claims about culture and societal change more broadly. We thus rely on this philosophical framework to study web search engine results by doing search as research in order to “emphasise the engines' capacity as indexer as well as author of the order of things” (111), but also to advance our understanding and critique of search engines and how they capture and produce culture society.

1.1. Search Engine Studies

For years now, Google Search has been the most used search engine in nearly all countries worldwide, but its critiques are still being developed. Despite their embeddedness and transparency as infrastructures (Star and Ruhleder 1996), search engines are still perceived of as exactly that: “taken for granted as a means of operation” (Feuz, Fuller and Stalder 2011). Therefore, research directed at making explicit the workings of the infrastructure that is Google is important. In reaction to this, Michael Zimmer (2010) revealed the growing importance and interest in studying web search engines. He writes, “Along with the necessary technical design and evaluative research, significant contributions have been made to understand web search engines within the context of transaction log analysis and user studies, within political, ethical, and cultural perspectives, and to utilise legal and policy analysis to help understand where possible remedies to many search-related concerns might exist.” (Zimmer 516). Moreover, Zimmer (2010) suggests four directions for further multidisciplinary approaches of the design, use and implications of search engines: search engine bias, search engines as gatekeepers of knowledge, values and ethics of search engines, and framing the legal constraints and obligations (516–517). Even though many studies work in more than one of these directions, his suggestions do not address localisation and personalisation specifically. This is, then, where our exploration primarily contributes.

In order to understand localisation, we look to the device itself. In the nineties, search engines still used to compete over their algorithms and different ranking schemata. For example: PageRank, HITS, SALSA (Zimmer 2010). But gradually search engines started using the same models (Rogers 114). As a result, the large, highly popular, general purpose search engines such as Google, Yandex or Baidu can now arguably all be used to study this particular modeling of culture and society. If we are to understand Google, we need to at least have a basic understanding of its workings. Broadly speaking, Google organises the results based on two engines each with countless parameters: the first is PageRank, which calculates document authority based on link analysis, and the second is a document matching engine that matches the given query input to a structured index or database (information retrieval) (Brin and Page 1998; Rieder 2012). Understanding the algorithm means understanding both components, but also the changes it went through over time. The Moz Blog keeps track of such announced changes in the algorithm – which is supposedly changed about 500–600 times a year (“Google Algorithm Change History”) – for search marketers who wish to optimise their websites. Looking at the general direction of its development, it seems that “The common themes in all of these algorithm updates boil down to surfacing higher quality contents that are more likely to answer individual searchers' questions.” (Kocher 2012). On February 12, 2012 a significant update to the algorithm under the code-name “Venice” was implemented. As Ramsey (2012) notes, “Simply put, the Venice update localised organic results on broad search queries.” (“Understand and Rock the Google Venice Update”). This includes “Improvements to ranking for local search results” and “Improved local results” (indeed, the algorithm only ever seems to “improve”). From that point onward, the localised “organic” results would be reliably tailored to the searcher on the city level. The fundamental problem, however, is that we cannot have a complete a priori understanding of the tool we are using for research, which problematises the notion of a controlled experiment. Rather, we treat Google as a black box: we can control the input and a number of settings so that we can analyse the output, and infer what happens inside the black box itself.

1.2. Analysing Search Engine Results

Search engine results mark a site of practices of information politics, a Web epistemological question of what gets privileged (Rogers 2004). As Van Couvering (2007) demonstrates, the service-for-profile market logic that aims at delivering a free service in exchange for a marketable profile has become a dominant discourse in search engine production. On the basis of in depth interviews with search engine producers she concludes that the “quality [of search engine results] is linked to long-term satisfaction on the part of customers, which in turn is linked to revenue, the positive goal or norm of the market scheme”. What is privileged or considered to be a “relevant” result is the outcome of a complex, biased and non-transparent process. Rogers (2004) makes a distinction between front-end and back-end politics here, to mark the difference between the questions of “Which information is allowed to be displayed; What constraints are placed on the scope of issues and range of arguments discussed?” (10) and the “politics behind how a search engine or portal selects and indexes its information” (1, emphasis in original). Not all links are equal, and the privileging mechanisms that they follow become “naturalised” over time, which makes this all the more urgent.

Actor-network-theory (ANT; Latour [2005] 2007) is a specific kind of network analysis that is useful to our study for two reasons: firstly, because it takes into account not only the semiotic dimensions (e.g. information politics) of the search engine network but also its material dimensions. Secondly, it is non-essentialist and thus avoids reducing explanation to a single vantage point. Speaking in the vocabulary of ANT (in a science and technology studies tradition), the inside and the outside of the Web search sphere (or actor-network) are delineated by the search engine, which establishes itself as an “obligatory point of passage” (OPP) (Callon 1986). An OPP is a point or space where local and global networks converge and where the local negotiates with the global. It is a point of mediation between the two, and the stronger it becomes established the more power it holds (resulting in what Grimmelmann dubbed “The Google Dilemma”). Thinking of search engines in this framework allows us to consider the negotiability of the spaces they demarcate, and how both human and non-human actors operate side by side (“flat ontology”) in both local and global networks in order to do so.

1.3. Euthanasia, Issue Networks, and Trajectories

Search as research is a valuable practice to look at contested issue spaces, such as the issue of euthanasia in Web search spheres. Death and dying are issues that on the one hand can be seen as universal – everybody encounters death at some point, regardless of the location a physical body is in. Yet, the way in which we deal with death and how life and death are given meaning also has strong roots in the social and the cultural. One area in which this is clearly visible is the ethical, moral and legal debate around euthanasia. Current attitudes and legal frameworks that deal with euthanasia vary greatly between countries, starting from the question of what exactly constitutes “euthanasia”. In New Zealand, for instance, euthanasia remains an unspecified term and is illegal (“Euthanasia”). In Canada, a difference is made between passive euthanasia (withholding food or medication) and active euthanasia (administering lethal medication), yet both of them are still illegal (“Euthanasia in Canada”). In the Unites States, such a specification of the term euthanasia also exists, but there the distinction is made between voluntary euthanasia, non-voluntary euthanasia and involuntary euthanasia, all of which are illegal. What matters for U.S. citizens is the distinction between euthanasia by itself (which in this case means the administering of drugs by a physician) and physician-assisted suicide (where the patient takes the drugs). While the former is illegal, the latter is currently acknowledged by law in four states (“State-by-State Guide”). In these countries mentioned above, the debate seems to revolve around establishing what constitutes as euthanasia (formalisation), and whether or not it is legal.

However, euthanasia is not only thought about in the legal/illegal dichotomy. In Netherlands, euthanasia has been legally defined, and so the debate takes on a different form. Since 2001, the administration of lethal drugs to a terminally ill patient by a physician, or the intake of those drugs by the patient, monitored by a physician, is not punishable by law (“Euthanasie”). However, recent controversies regarding the strict guidelines around euthanasia (with one man brought to court for helping his terminally ill mother in law and a physician that committed suicide after threatening to be indicted for administering a lethal doses of morphine to a terminally ill patient) sparked new debates about easing some of these guidelines in Netherlands. This is in stark contrast with other countries (such as those mentioned above) where euthanasia is underspecified, seen as illegal, or where no legal framework around the issue even exists for such nuances to be made. Thus, it seems that the various debates around euthanasia in different locations can be characterised by their “maturity” as socio-cultural issues. It should be noted that the notion of “maturity” implies that issues have “trajectories”, which are followed. Understood in terms of Actor-Network Theory, this “maturity” could then be seen as the degree to which euthanasia as an issue network is stabilised – its obduracy (Pinch, Ashmore, and Mulkay 1992) – in different geographical regions, jurisdictions, or societies.

2. Research Questions

First, we aim to contribute to the discussion by enquiring into to some broader questions. Can we perform research by using search as research? How does Google demarcate the local and to what extent may Google's device-specific definition of the “local” be used to perform cross-country and comparative media analyses. Can we read the search results as indications or as a reflection culture and societal change?

Second, we raise questions that are more specific to our case study. As introduced above, the wide variety of interpretations of euthanasia and its legal status as an issue differs greatly between countries. Therefore, we wonder to what extent these socio-cultural and legal attitudes towards euthanasia are apparent in the types of sources and actors that show up in the search results for each of the selected local Google domains. We also seek to explore how far these different source types or actors are removed from the “top of the Web” and to see if there is a hierarchy among them. Can we use these measures to understand the extent to which that issue has matured?

3. Methods

3.1. Configuring a Research Browser

On the application-level, we have configured Mozilla Firefox to not store browsing history, and to send a do-not-track-request with our browsing traffic. On the device-level, we “disentangled” ourselves from Google (Borra 2013) by decreasing personalisation: turning off Google Instant, autofills and autocomplete, clearing histories and cookies, opting out of customisation based on search activity, raising the number of results to be shown per page from the default of 10 results per page to 20 results per page, and logging off. Even though opting out will remove personalisation on Web-history and (social) network, it will still retain localisation on geography.

3.2. Query Design

If we want to study how Google reflects societal change, we need to design our query so that it can be used to answer our research question. This means we need to do query design. “‘Query design’ is the practice of formulating a query so that the results can be interpreted as indications and findings (however cluelike), as opposed to mere information retrieved or optimization and manipulation exposed (however fascinating)” (Rogers 110–111). Working from an initial problem related to the “transliteration” of our query into other languages (how does one translate the different forms of “euthanasia” into other languages that do not yet have precise terms for this), we decided to narrow our analysis down to only English-speaking populations and to the query [“euthanasia”]. This way, the additional complexity that comes from this step of translation would be avoided, and allows us to let Google “fill in” the different interpretations of this underspecified (or rather socio-culturally contested) term. Additionally, reports on the frequency of the search term have been generated per region (see next section) using Google Trends to quantify the actual usage of the search query for these regions and subregions (Figure 1a–1g).

Fig. 1a. Regional interest (Australia) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1b. Regional interest (Canada) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1c. Regional interest (Ireland) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1d. Regional interest (New Zealand) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1e. Regional interest (Singapore) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1f. Regional interest (United Kingdom) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

Fig. 1g. Regional interest (United States) in “euthanasia” from 2004–present as measured by the number of searches in Google Search. Report generated in Google Trends on 21 Nov. 2013, 20:55.

3.3. Preparing a Source Set

To fit our query design to the right set of local domains to be studied, we gathered lists of English-speaking populations per region (“List of countries by English-speaking population”) (Table 1), number of internet users (“List of countries by number of Internet users”) and the penetration per region (“List of countries by number of Internet users”) (Table 2). These lists could then be triangulated using the DMI Triangulate tool (Table 3) to exclude from our analysis those countries that did not meet our criteria. We then selected sources that met three specified thresholds: first, population size had to be greater than 400,000 so that we could ensure the credibility of the sample for that country (larger populations make for a more stable average). Second, at least 60% of the population of a country would need to register as English-speaking. And third, Internet penetration in those countries would have to be at least 60%. In combination with the first parameter, we ensured that a significant number of Internet users would be covered this way (excluding small islands or archipelagos). Furthermore, the triangulated list with more than 60% of the population speaking English (Table 4a, list 1) was later reduced to include only those who have English as an official primary language (Table 4b). This rules out uncertainties (and our personal inferences) regarding the default querying language of these countries (e.g. do Dutch people generally query in English or in Dutch?). Finally, we specified the search engine and country code top-level domains (ccTLD) for each of these regions based on dominant search engine usage in that same region (Table 4a–b). For this, we used Alexa's ranking of top sites per country to determine this. Not surprisingly, all of these countries listed Google as their dominant search engine and all have their own ccTLD. Since we could use the same search engine for all of our searches, we could also compare results between countries.

3.4. Running the Query and Analysing the Results

We queried [“euthanasia”] in the local Google domains for each of our selected countries in the source set: google.com.au, google.ca, google.ie, google.co.nz, google.com.sg, google.co.uk, google.com (Table 4b). The output results for each of these queries were saved in HTML format and stored on a local hard drive. In doing all this, we are breaking the Google Terms of Service (e.g. by not accessing Google through its interface, saving and storing the results, and by making a derivative work; “Google Terms of Service”). From the results, we categorised the top 20 returned links (excluding image results, and Google News results from our actual analysis). We applied the categories a posteriori and based on the most prevalent content or intention, and compared the analysis between the two of us to ensure at least some representativeness. Furthermore, in imposing these categories we gradually redefined what they include or exclude as we went through the content analysis. Results were also triangulated, counted and visualised.

4. Results

Although debates around euthanasia occur worldwide, the scope of our research was limited in a number of ways. While we did consider the internet penetration for each region, we cannot conclusively rule out the possible digital divides in the countries that we chose (the results might represent some privileged part of society). Because of time limitations we also did not use proxies to route our requests through the country in question. This might have skewed the results to the extent that localisation by geography would have occurred, although as described in 3.1 we tried to configure a research browser as carefully as possible. Another point to take into consideration relates to the way Google as a search engine works. It is problematic to analyse exactly why certain results rank higher than others (e.g. Google properties are returned higher and Wikipedia is often at the top of “informational queries”) or why not every result is ranked (e.g. social media pages and discussions usually do not appear in the results). As we have no completely transparent view of how Google works, we can only treat it as the black box that it is, merely researching the in- and output and thereby inferring what happens in the box itself. This means that we have to be careful in making claims about the results that show up “on the top”.

4.1. Qualitative Content Analysis: Local Discourse and Anomalous Sources

Based on our analysis, we can look at the present discourse around the term “euthanasia” to infer how people think about it, but we can also separate more-or-less common results from more unique results. One interesting result is the link to the United States Holocaust Memorial Museum (categorised by us as “Other”). This link directs to the specific page about the “euthanasia” program that the Nazi's in the Second World War devised in order to kill anyone who was disabled or mentally ill. Such a result places the use of the term “euthanasia” in a historical perspective and shows how the notion of euthanasia as “the good death” – from its Greek etymological origin – is tied up with specific perspectives about the value of life and death the moral question of ending it. It shows that the debate around euthanasia is also a debate about agency and who determines the possibility to act. In this respect, the use of the words “being killed by” to define euthanasia by some sources links stands out, as it implies that a murder is committed (even if the patient takes his own life) and thus makes the term euthanasia devoid of agency on the side of the patient. The rhetorics of the opposition are typically either about comparing the present situation in Belgium, religious, or about the belief in the power and advancement of medical sciences. It is perhaps all the more puzzling that the results show that the term is now not only used in relation to humans. Instead, a specific and significant set of links point towards the safeguarding of animal welfare and care. Canada, Ireland, Singapore and the U.S. list websites that point to veterinary information on the practice of putting animals to sleep – beings one can say, with relatively little agency.

4.2. Cross-Spherical Analysis: Comparing Local Issue Spaces

Looking at the categories of our analysis, it seems that the current state of the debates are mostly about the ethical dimensions, legislative dimensions, and objective resources such as studies and reference sources. To see who occupies these issue spaces – or at least as we categorised them – their distribution was compared in Table 5a–h (further studies could look into the influence of each actor in this regard). Figure 2 shows the categorical distribution of the issue, which allows us to compare the number of links per category per country in two ways: between countries across categories (Figure 2a), and between categories in particular countries (Figure 2b). It seems that there is no pattern in the division of these various dimensions of euthanasia in terms of category ranking across countries. However, it should be noted here that news sources are prominently featured in all local top results, although the content of these stories varies greatly. In Australian results, for instance, there were mainly news results about the opening of a euthanasia clinic in Adelaide (Table 5g). The United Kingdom takes a leading position with no less than nine news articles in the first twenty results. Although the sources vary, many of these stories are about a Belgian case that raises new questions about euthanasia in Belgium itself. This case returns in all the result lists, which is probably due to its “newsworthiness.” The kind of sources discussing it vary here as well, as do the number of returned links (Table 5d). Overall, we identified legislative, ethical and oppositional concerns as the most frequent categories.

Fig. 2a. Comparative bar chart visualisation of Table 5a. Horizontal axis: top 20 Google results per local domain divided into a posteriori defined categories. Vertical axis: number of links per category per local domain. Colours represent the countries in question.

Fig. 2b. Accumulative bar chart visualisation of Table 5a. Horizontal axis: top 20 Google results per local domain divided into a posteriori defined categories and showing the number of links per category. Vertical axis: country in question. Colours represent the various categories.

In triangulating the results, we also found that all of the countries have the Wikipedia page about euthanasia as their top result – and the “general” English language wikipedia at that (as opposed to the Wikipedia page about euthanasia in that specific country, which often does exist) (Table 6). Most notably, as can be seen in Figure 3, most actors only appear once across countries. Most of those that do appear twice, all appear in the same two domains: Singapore and the United States (Figure 3b). Thus, there seems to be a strong overlap between the two spaces. Furthermore, most of the actors that do appear two or more times across all countries appear in the U.S. list as well. The various local domains are thus presenting us with the same themes, but do so not necessarily through the same sources (actors or links) (Table 5a–h). Besides the Wikipedia results, there is no apparent pattern as to which types of sources appear closer to the “top of the web’ in terms of their ranking.

Fig. 3a. Colour-coded table of Table 6. It shows the triangulated list output colour-coded by frequency and in its original order. Visualisation generated by the DMI Triangulate tool.

Fig. 3b. Colour-coded table of Table 6. It shows the triangulated list output colour-coded by frequency and ordered by frequency and alphabet. Visualisation generated by the DMI Triangulate tool.

5. Discussion

5.1. Issue Maturity and Studying the Social

Returning to our introduction of actor-network-theory earlier, we can now briefly discuss the “maturity” of the issue as we understand it in terms of its obduracy (Pinch, Ashmore, and Mulkay 1992) or as “stabilisation of controversies” (Latour 15). In our analysis, we diagnosed the “present state” of (a part of) the social, but this can only give us a “material” account of sociology: what Latour ([2005] 2007) calls a “sociology of the social”. Instead, Latour suggests that we should develop a “sociology of associations” (11) and “follow the actors themselves” (12). For our analysis, this means that we have captured only a single snapshot of the present material state of the social itself, which is in fact characterised by the very “movement” behind it. In order to determine issue “maturity” then, we could imagine that studying the social would entail producing (or capturing) multiple of these snapshots or “present states” of these “movements” over time (hours, days, weeks, years, depending on the speed of the issue, the actors involved, and the researcher himself). We should start following the trail of breadcrumbs (“traces”) that actors leave behind, so that we can understand the specifics of the relations they assemble in the present. We are thus arguing that we cannot make claims about the social (and the maturity of euthanasia as we understand it is very much a social problem) from this single snapshot. Rather, we can only say something about its present material state with our imposed categories. We need to “trace” the “associations” that make up the social, “following only those [actors] that have currency in the rear-world of the social.” (47). That way we can infer meaning from these snapshots about the movements that actually make up the social.

5.2. Negotiating Localisation and Personalisation

What we study here as the “local” is in fact a negotiated space between the users, through Google's algorithms. The local as a measure of personalisation is not a stable, clearly delineated territory, but is constantly being re-produced. In fact, Feuz, Fuller and Stalder (2011) concluded that Google's personalisation “promised an “augmented reality’ in which machine intelligence interprets the user's individual relationship to reality and then select what's good for each”. In other words, Google authors a different reality for each specific user that they think would fit their needs. Even though we de-personalised our results as much as possible by configuring a research browser, we already see that based on this single parameter of location, different realities are being negotiated. The problem, then, is that we do not have any way of understanding the amount of influence we can exert over this negotiation as users in this network largely populated by non-human actors.

So what kind of version of the local is Google trying to negotiate then? The fact that Singapore and the United States show great overlap gives us new insight into the question of how Google frames the local. Either the local is based on the U.S. and the double results that appear in other country lists originate from there, or the U.S. results are an aggregated set of results from other countries. Seeing as most of the results are located in the U.S., the doubles as well, the first option would seem more likely. Not only the overlap of direct links is significant here. As we have shown in our results, even though different sources are presented, they can still deal with the same topic (such as the recent debates around the euthanasia case in Belgium). This alludes to what Eli Pariser (2011) famously called the “filter bubble”, in which the far-reaching personalisation of Web search results in the construction and maintaining of bubbles in which only some content tailored to your profile is returned. Moreover, Richard Rogers' critique (2009) of the inculpable engine also becomes more and more relevant: “personalization takes the search engine off the hook, because the “blame” or responsibility for the results is partly one's own.” (183). All this could be read in response to Grimmelmann's “Google Dilemma” (2009), as a way for Google to relocate part of its responsibility to the user.

5.3. Suggestions for Further Research

As we explained above, the classification we made of the returned results was made based on the most prevalent content or intention of a website. Yet, further research that deploys Google as a media source and research machine could also be of more interest. Using the DMI Lippmannian Device (formerly named Google Scraper), a tool to scrape Google results, it becomes possible to do source partisanship and issue commitment research. This way, the types of actors that are part of the euthanasia issue network and the degree to which they are committed to, or can exert influence over, this issue could be mapped out. Using the device, it is also possible to generate keyword frequency source clouds, which could for instance be used to examine how big the pro and con lobby is, for instance, and from what perspective their arguments stem from (e.g. religious or humanistic). Finally, to complement our Anglocentric analysis, it would be valuable to include other cultures that may think (radically) different about these topics as well.

6. References

Borra, Erik K. “Things Internet Researchers Should know About Google.” Digital Methods Initiative Wiki. Digital Methods Initiative, 13 Nov. 2013. Web. 21 Nov. 2013. <https://wiki.digitalmethods.net/Dmi/ThingsInternetResearchersShouldKnowAboutGoogle>.

Brin, Sergey, and Lawrence Page. “The Anatomy of a Large-Scale Hypertextual Web Search Engine.” Seventh International World-Wide Web Conference, 1998. Web. 20 Nov. 2013. <http://infolab.stanford.edu/~backrub/google.html>.

Callon, Michel. “Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay.” Power, Action and Belief: A New Sociology of Knowledge? Ed. John Law. London: Routledge, 1986. 196–233. Print.

“Euthanasia.” New Zealand Medical Association. N.p., n.d. Web. 20 Nov. 2013. <http://www.nzma.org.nz/policies/advocacy/position-statements/euthanasia>.

“Euthanasia.” The Free Dictionary. N.p., n.d. Web. 20 Nov. 2013. <http://legal-dictionary.thefreedictionary.com/Euthanasia+and+Physician+-Assisted+Suicide>.

“Euthanasie | Levenseinde En Euthanasie” Rijksoverheid. N.p., n.d. Web. 20 Nov. 2013. <http://www.rijksoverheid.nl/onderwerpen/levenseinde-en-euthanasie/euthanasie>.

“Euthanasia and assisted suicide.” NHS. Gov.uk, 1 Aug. 2012. Web. 20 Nov. 2013. <http://www.nhs.uk/conditions/euthanasiaandassistedsuicide/Pages/Introduction.aspx>.

“Euthanasia in Canada.” Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc., n.d. Web. 20 Nov. 2013. <http://en.wikipedia.org/wiki/Euthanasia_in_Canada>.

Feuz, Martin, Matthew Fuller, and Felix Stalder. “Personal Web Searching in the Age of Semantic Capitalism: Diagnosing the Mechanisms of Personalisation.” First Monday 16.2 (2011): n. pag. Web. 20 Nov. 2013. <http://firstmonday.org/article/view/3344/2766>.

“Google Algorithm Change History.” The Moz Blog. SEOmoz, Inc., n.d. Web. 20 Nov. 2013. <http://moz.com/google-algorithm-change>.

“Google Terms of Service.” Google. Google Inc., 11 Nov. 2013. Web. 21 Nov. 2013. <http://www.google.com/intl/en/policies/terms>.

Grimmelmann, James. “The Google Dilemma.” Law Review 53.4 (2009): 939–950. Print.

Kocher, Jill. “Understanding Google's Algorithm Updates.” PracticalEcommerce. 26 Oct. 2012. Web. 20 Nov. 2013. <http://www.practicalecommerce.com/articles/3785-Understanding-Google-s-Algorithm-Updates>.

“International Perspectives Legal Status of Euthanasia and Assisted Suicide.” ProCon.org. ProCon.org, 28 Sept. 2010. Web. 20 Nov. 2013. <http://euthanasia.procon.org/view.resource.php?resourceID=000136>.

Latour, Bruno. Reassembling the Social: An Introduction to Actor-Network-Theory. 2005. Oxford: Oxford University Press, 2007. Print. Clarendon Lectures in Management Studies.

“Legality of euthanasia.” Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc., n.d. Web. 21 Nov. 2013. <http://en.wikipedia.org/wiki/Legality_of_euthanasia>.

“List of countries by English-speaking population.” Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc., n.d. Web. 21 Nov. 2013. <http://en.wikipedia.org/wiki/List_of_countries_by_English-speaking_population>.

“List of countries by number of Internet users.” Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc., n.d. Web. 21 Nov. 2013. <http://en.wikipedia.org/wiki/List_of_countries_by_number_of_Internet_users>.

“List of countries where English is an official language.” Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc., n.d. Web. 21 Nov. 2013. <http://en.wikipedia.org/wiki/List_of_countries_where_English_is_an_official_language>.

Pariser, Eli. The Filter Bubble: What the Internet Is Hiding From You. New York: The Penguin Press, 2011. Print.

Pereira, J. “Legalizing Euthanasia or Assisted Suicide: The Illusion of Safeguards and Controls.” Current Oncology 18.2 (2011): e38–e45. Print.

Pinch, Trevor, Malcolm Ashmore, and Michael Mulkay. “Technology, Testing, Text: Clinical Budgeting in the U.K. National Health Service.” Shaping Technology/Building Society: Studies in Sociotechnical Change. Eds. Wiebe E. Bijker and John Law. Cambridge, MA: The MIT Press, 1992. 265–289. Print. Inside Technology.

Ramsey, Mike. “Understand and Rock the Google Venice Update.” The Moz Blog. SEOmoz, Inc., 12 March 2012. Web. 20 Nov. 2013. <http://moz.com/blog/understand-and-rock-the-google-venice-update>.

Rieder, Bernhard. “What Is in PageRank? A Historical and Conceptual Investigation of a Recursive Status Index.” Computational Culture 2 (2012): n. pag. Web. 20 Nov. 2013. <http://computationalculture.net/article/what_is_in_pagerank>.

Rogers, Richard A. Digital Methods. Cambridge, MA: The MIT Press, 2013. Print.

———. Information Politics on the Web. Cambridge, MA: The MIT Press, 2004. Print.

———. “The Googlization Question, and the Inculpable Engine.” Deep Search: The Politics of Search Engines Beyond Google. Eds. Felix Stalder and Konrad Becker. Innsbruck: StudienVerlag, 2009. 173–183. Print.

Sottimano, Dave. “Google ccTLDs and Associated Languages & Codes Reference Sheet.” Distilled. Distilled, 2 Aug. 2012. Web. 20 Nov. 2012. <http://www.distilled.net/blog/uncategorized/google-cctlds-and-associated-languages-codes-reference-sheet>.

Starm Susan Leigh, and Karen Ruhleder. “Steps Toward an Ecology of Infrastructure: Design and Access for Large Information Spaces.” Information Systems Research 7.1 (1996): 111–134. Print.

“State-by-State Guide to Physician-Assisted Suicide” ProCon.org. ProCon.org, n.d. Web. 22 Nov. 2013 <http://euthanasia.procon.org/view.resource.php?resourceID=000132>.

Van Couvering, Elizabeth. “Is Relevance Relevant? Market, Science, and War: Discourses of Search Engine Quality.” Journal of Computer-Mediated Communication 12.3 (2007): 1–18. Print.

“What Is Involuntary Euthanasia?” wiseGEEK. Conjecture Corporation, n.d. Web. 20 Nov. 2013. <httphttp://www.wisegeek.com/what-is-involuntary-euthanasia.htm>.

Zimmer, Michael. “Web Search Studies: Multidisciplinary Perspectives on Web Search Engines.” International Handbook of Internet Research. Eds. Jeremy Hunsinger, Lisbeth Klastrup, and Matthew Allen. Dordrecht: Springer Netherlands, 2010. 507–521. Print.

Info
Title: Doing Search as Research
Subtitle: Mapping Localisation and Issue Maturity with Search Engines
Type: Research report; Assignment
Author.name: T. (Tessa) de Keijser; F. N. (Fernando) van der Vlist
Author.affiliation: Graduate School of Humanities, Faculty of Humanities, University of Amsterdam
Instructor.name: Dr. B. (Bernhard) Rieder; E. K. (Erik) Borra; Prof. Dr. R. A. (Richard) Rogers; Dr. A. (Anat) Ben-David
Instructor.affiliation: Dept. of Media Studies, Faculty of Humanities, University of Amsterdam
Abstract: We report on whether and in what way Google's device-specific definition of the “local” may be used to perform cross-country and comparative media analysis with local domains of Google. We suggest to think of search engines in a framework of actor-network-theory which allows us to focus on the negotiability of the spaces they demarcate, and how both human and non-human actors operate side by side to achieve this. Search as research is a valuable research practice to look at contested issue spaces, such as the issue of euthanasia in Web search spheres. We wondered to what extent different socio-cultural and legal attitudes towards euthanasia are apparent in the types of sources and actors that show up in the search results for each of the selected local Google domains, and how far these different source types or actors are removed from the “top of the Web”. Can we use these measures to understand the extent to which that issue has matured? To operationalise our research questions, we first undertook a qualitative content analysis to understand local discourse and to be able to identify patterns and anomolies across countries. From there, we then did cross-country comparisons of thee local issue spaces and found that the current state of the debates are mostly about the legislative dimensions, ethical dimensions, and informative, “neutral” resources such as studies and reference sources. Overall, we identified legislative, ethical and oppositional concerns as the most frequent categories, but even though the same categories show up for each of the local domains in our source set, they do not necessarily refer to the same sources. Finally, we conclude that the “local” is in fact a negotiated space between the users, through Google's algorithms, and that in order to determine issue “maturity” we would need to study multiple social snapshots or “present states” of the “movements” that characterise the social over time.
Keywords: Internet research, digital methods, search engine studies, query design, issue maturity, Google, euthanasia
Length.words: 4,620
Length.reading: 26 mins
Element.table: Table 1; Table 2; Table 3; Table 4a; Table 4b; Table 5a; Table 5b; Table 5c; Table 5d; Table 5e; Table 5f; Table 5g; Table 5h; Table 6
Element.figure: Fig. 1a; Fig. 1b; Fig. 1c; Fig. 1d; Fig. 1e; Fig. 1f; Fig. 1g; Fig. 2a; Fig. 2b; Fig. 3a; Fig. 3b
Date.submitted: 22 Nov. 2013
Date.evaluated: 25 Nov. 2013
Date.publishedonline: 1 Feb. 2014
Language: English (United Kingdom)
Documentation.style: Modern Language Association (7th ed.)
Export.citation: BibTEX
Export.print: javascript:window.print()
2012– fernandovandervlist.nl
v1.2.28