The Nationality of Issues. Rights Types

What may be expected when interpreting search engine results, especially the rankings of sites for particular queries? What may be expected when comparing results across the many local versions of Google? In the project The Nationalities of Issues: Rights Types, we entered the word ‘rights’ in various languages into the local Googles in order to obtain hierarchies of rights types per country. Are there distinctive rights that reach the top in Finland, the Netherlands, France, Italy, Switzerland, Germany, Austria, Sweden, Russia, Japan, Canada, the United Kingdom, Australia, Philippines, Ivory Coast and other countries? (Rogers et al, 2009).
When repurposing Google for national research some considerations need to be taken into account. Following, two issues with designing queries are discussed as well as how they affect the analysis of the outcomes.

Firstly, language is important in cross-national web analysis. A query in either Dutch or French in google.be produces quite different results. Generally speaking, a query would be made in a country’s dominant language(s). Dominant languages might be interpreted as official language(s), predominantly used on a national web, or by the population. The language is always determined on the basis of the research question. In some cases the language chosen might be the language of a minority (e.g. Arabic in the Netherlands). Note that languages related to a specific geographical area might be an indication for local concerns about the topic of inquiry (e.g. compare Finnish in google.fi to Spanish in google.com.ar, the former results might be interpreted as local while the latter might include transnational results).

Secondly, I want to point at the effect of underspecified versus refined queries on the results returned, what may be termed “algorithmic effects”. Or as Page and Brin phrase it, “the benefits of PageRank are the greatest for underspecified queries” (Page et al 1998: 9). A refined formulation of keywords will decrease the algorithmic effects in the results, whereas the more general the term, the more algorithmic effects appear. In other words, Google returns what can be found in the index. If there is not a lot in the index for a very specific query, there are also less hyperlinks, or votes, to the resulting sources.

In this case study, the aim was to stress test our effort to find the local, culturally specific with Google, hence the underspecified query “rights.” In other words, we tried to analyze Google results in various national domain Googles, in order to find out to what extent local and culturally specific rights were returned in the top results. So the first step was to query the term rights in the local languages in the local Google versions (e.g. oigused in Google.ee and direitos in Google.pt, see table 1 for the complete list). The second step was to read and interpret the results and make lists of the top ten distinctive rights types, leaving them in the order that Google provided.

Google.se with query “rattigheter” (13.07.09)
Google.fi with query “oikeudet” (13.07.09)
Google.ee with query “oigused” (15.07.09)
Google.lv with query “tiesibas” (16.07.09)
Google.co.uk with query “rights” (13.07.09)
Google.nl with query “rechten” (13.07.09)
Google.be with query “rechten van” (15.07.09)
Google.be with query “droits” (14.07.09)
Google.lu with query “rechte” (15.07.09)
Google.de with query “rechte” (15.07.09)
Google.at with query “rechte” (15.07.09)
Google.ch with query “rechte” (15.07.09)
Google.fr with query “droits” (14.07.09)
Google.pt with query “direitos” (14.07.09)
Google.es with query “derechos” (13.07.09)
Google.it with query “diritto al” OR “diritto all” OR “diritto alla” (13.07.09)
Google.ro with query “drepturile” (13.07.09)
Google.mo with query “drepturile” (13.07.09)
Google.ru with query “prava” (13.07.09)
Google.com.tr with query “haklari” (17.07.09)
Google.jp with query “\x{6a29}\x{c801}” (16.07.09)
Google.hk with query “\x{6b0a}\x{c801}” (17.07.09)
Google.com.ph with query “karapatang” (16.07.09)
Google.ci with query “droits” (17.07.09)
Google.com.au with query “rights” (14.07.09)
Google.ca with query “rights” (15.07.09)
Google.ca with query “droits” (15.07.09)
Google.com with query “rights” (14.07.09)
Google.com with query “derechos” (15.07.09)
Google.com.mx with query “derechos” (15.07.09)
Google.com.br with query “direitos” (15.07.09)
Google.ar with query “derechos” (15.07.09)
Google.pe with query “derechos” (15.07.09)

Table 1. Local Googles and the term “rights” in the respective local languages

Reading and interpreting the results, result page analysts had to take into account two things. Firstly, the prominence of Wikipedia in most queries’ top results. By comparing results for a selection of queries over time, search engine critic Nicholar Carr (2008) found that the symbiotic relationship between Google and Wikipedia is increasing. Starting in 2007 on a position between one and nine, in 2009 Wikipedia has reached the number one position on all test queries. Although arguably Wikipedia scores highest because it contains most links, in this case study, as we tried to find local rights, Wikipedia articles were excluded from the analysis.

Secondly, when analyzing the returned sources, a second source appeared in a number of result pages. Rightsforartists.com appeared among the top results because of its title R.I.G.H.T.S.. Google takes the title into account when calculating the relevance of a source for a particular query. Considering this as a Google artifact, this result was excluded from analysis as well.

Given these two editorial decisions, is it possible to study the social with Google? Countries could be argued to have diverging concerns, in view of Google results (Figure 1). For example, ‘everyman’s rights’ in Finland, ‘prostitutes’ rights’ in the Netherlands, ‘computer programmers’ rights’ in Japan and the ‘right to oblivion’ in Italy (the right to have personal data deleted) are unique to the respective countries. Given the commitment to use Google as a data source for socio-cultural research, as well as to preserve the cultural distinctiveness of the rights found, classical social science methods (such as categorizing findings for easier comparison) as well as attempts to rephrase or correct the phrasing of types of rights (e.g. regarding LGBT rights in the United States and homosexual rights in Hong Kong as equivalents) were resisted. Also, given the limited sample of countries and the selection method, the most widely shared rights across countries are not the subjects of analysis (Rogers et al, 2009).

In this case study we established that differences between the local Googles do exist. The shared as well as the culturally specific may be read in Google results. Moreover, the study demonstrates that Google result pages may be read as a text in order to make claims about local cultures. In the following case study, we applied an extra methodological step by looking how local sources actually are.

Figure 1: Selection Rights Types. Visualization by Vera Bekema and Anne Helmond

This is a project of the Digital Methods Initiative Summer School. Go to project page