Internet Geographer


Posts tagged google
Digital Hegemonies: The Localness of Search Engine Results

I have a new paper out in the Annals of the American Association of Geographers with Andrea Ballatore and Shilad Sen. In it, we ask (and empirically answer) questions about the the local-ness versus foreign-ness of content that Google serves up to people around the world. You can access the full paper below. 

Ballatore, A., Graham, M., and Sen, S. 2017. Digital Hegemonies: The Localness of Search Engine Results. Annals of the American Association of Geographers.  DOI:10.1080/24694452.2017.1308240.

Please also note that I'm currently hiring a Researcher in Digital Geography if you'd like to work with me on similar topics!

Summary (taken from the paper's conclusions):

This investigation of the geography of Google Search results shows that wealthy and well-connected countries tend to have much more locally produced content that is visible about them than poor and poorly connected countries. Even cities located in countries with huge populations such as Lagos show a tendency toward having relatively little local content about them in Google Search results. This means that a user in the United States or Germany searching for cities is far more likely to be given access to locally produced content than a Tanzanian or Cambodian.

In our empirical study, the results of only eight countries in Africa (and four low-income countries: Tajikistan, Madagascar, Burkina Faso, and Tanzania) have a majority of content that is locally produced. This gives rise to a form of digital hegemony, whereby producers in a few countries get to define what is read by others. The United States, in particular, is a dominant content-producing force, even when excluding Wikipedia, which is a highly visible U.S.-based but globally assembled resource (Figures 10 and 11). In the results for sixty-one countries, the United States supplies over half of the first page content on Google. This means that not only are U.S. Internet users surrounded by an extremely locally produced Internet, but U.S.-produced content is highly visible in much of the rest of the world. This does not necessarily mean that the United States is an informational hegemon everywhere in the world, however. France has a somewhat smaller sphere of influence, mainly limited to countries in Africa, whereas Russia produces a visible effect only on results about Kyrgyzstan (see Figure 10).

It is important to note that, because our data set focused only on capital cities, caution should be taken when extending the results to higher spatial granularities. Our localness indicator does not take into account the actual interaction of users with search results and the variety of devices and media across which individuals currently access search engines. Despite the precautions that we took to access representative samples of search results, some noise is still present and some results might show high volatility. Much more empirical work is needed to study finer patterns within countries and to build more accurate models to investigate the consumption of geographic information on search engines in different geographic locales.

More broadly, the point remains that most countries in the Global South continue to be defined by a diverse range of sources originating from a diverse range of places. The issue here is not that Internet users are exposed to a diverse range of sources from a diverse range of places—indeed, as Pariser (2011) noted, there are significant concerns for people and media ecosystems that lack access to such diversity. The issue is rather that that diversity itself has a particular bias and those sources tend to be almost entirely from the Global North; very few of the sources come from anywhere in the Global South. For instance, although the search results for Google's Ghanaian page for its capital “Accra” include pages from six countries, five of them are firmly located in the Global North.

When looking at countries in the Global North, the results for Denmark's capital are similarly diverse, with five out of six source countries also being located in the Global North. By contrast, a country like the United States suffers from the inverse problem: having almost no exposure to geographic representations made by nonlocals.

The key question, then, is why. What explains this informational hegemony, or the dominance of the Global North in producing digital representations about not just themselves but also about much of the Global South? Interestingly, our explanatory models indicate that network connectivity and economic development in a country are not enough to make content about that place more local in Google Search results. The presence of a strong publishing industry, using SciMago publication data as proxy, is the strongest predictor of the production of visible online content. The importance of the h-index in the model also shows that the impact of scientific publications is a better predictor of localness than the mere number of publications. Thus, we suggest that socioeconomic systems that produce high-quality research also tend to produce highly visible online content. There are no countries in the Global South that score well on such metrics, and there are consequently no countries in the Global South that play a major role in constructing contemporary Internet geographies.

Having taken a first step in this direction, more quantitative and qualitative research is needed to better understand why exactly scientific knowledge production explains so much of the variance in Google's local digital representations. More relational variables and different spatial granularities will have to be considered. Until then, though, we hope that the finding that wealth and network connectivity alone are not sufficient factors is worth demonstrating, especially for Internet activists who hope to bring about more genuinely participatory and representative digital environments. This point increasingly matters because places are ever more defined by their digital presences, and the ways in which places are represented digitally increasingly shape how people understand and reproduce those very places (Graham, De Sabbata, and Zook 2015). Google plays an enormous role in constructing these digital representations of places. Because of their dominant role in mediating a majority of the world's Internet use and the fact that few people ever explore beyond a first page of search results, they essentially determine which digital augmentations of place are made visible or invisible, with tangible effects in the physical world.

This article demonstrated that Google Search results are actively reproducing new forms of informational hegemony around the globe. A few countries in the Global North play an inordinately large role in defining the digital augmentations of the Global South. Google's methods for ranking and representing are notoriously opaque (Vaidhyanathan 2011; Graham et al. 2014), but we do know that two key factors come into play. First, much of the reason of the lack of local voice in the Global South is likely simply because the production of Internet content happens at a much lower rate compared to that in the Global North (Graham, De Sabbata, and Zook 2015). Second, since the company's creation in 1998, Google's algorithms have tended to favor highly central Web content: Pages linked to by a lot of other pages are prioritized, and those largely ignored are demoted in the rankings. This creates a worrying situation whereby it becomes difficult for those on the information peripheries to break out of their digital marginality.

Building on earlier research looking at the geographies of information, this article has analyzed not just where digital content comes from but how it is ranked in the world's most powerful digital mediator. Much more will need to be done to understand not just the ways in which people are afforded voice about their own communities and countries but also the myriad factors that serve to amplify or constrain it. Until then, we hope that other research can use this article as a beginning to ask not just why some parts of the world are denied locally produced representations but how we might bring about more representative and participatory digital augmentations of place.

Google and its responsibility for search results

BBC News has just published a piece in which they tackle the fact that Google has been prominently displaying Holocaust denial content

Shockingly, they also uncover other examples such as the screenshot above (apologies for reproducing it) in which Google's knowledge graph is displaying concerning and highly offensive content. 

In the BBC piece, I argue both that algorithms are not neutral, and that Google occupies a position of immense power because of the huge amount of digital content that it mediates. As such, they need to take responsibility for these results. They can't just point to the algorithm and say 'well, this wasn't our intent' (an argument that I make more directly in this recent piece: 'Let’s make platform capitalism more accountable').

For people interested in the provenance of some of Google's more questionable search results, please read these recent articles that Heather Ford and I have been working on:

Ford, H., and Graham, M. 2016. Semantic Cities: Coded Geopolitics and the Rise of the Semantic Web. In Code and the City. eds. Kitchin, R., and Perng, S-Y. London: Routledge. 200-214.

Ford, H. and Graham, M. 2016. Provenance, Power, and Place: Linked Data and Opaque Digital geographies. Environment and Planning D: Society and Space. doi:10.1177/0263775816668857 (pre-publication version here).

Graham, M. 2015. Why Does Google Say Jerusalem is the Capital of Israel? Nov 30, 2015

See also, my earlier post on this topic: 'On how Google frames, shapes and distorts how we see the world'

Semantic Cities: Coded Geopolitics and the Rise of the Semantic Web.

In order to understand how the city’s contested political contexts are embedded into its digital layers, we traced how the city is represented on online platforms that house facts about much of the world. We did this by analyzing representations of Jerusalem across the Arabic, Hebrew and English versions of Wikipedia (working with a translator on the Arabic and Hebrew versions), as well as on the platforms of Wikidata, Freebase and Google. As our cities become increasingly digital, and as the digital becomes increasingly governed by the logics of the semantic web, there are important questions to ask about how these new alignments of code and content shape how cities are presented, experienced, and brought into being. What we found is a paradoxical situation whereby, through connecting datasets, semantic web initiatives detach localized information from the contexts of its creation. By divorcing content from its contexts, this process establishes new contexts in which necessarily political decisions are being made with far reaching consequences.

This is a topic of a new chapter (that I wrote with Heather Ford) that just arrived on my desk this morning. You can download the piece here:

Ford, H., and Graham, M. 2016. Semantic Cities: Coded Geopolitics and the Rise of the Semantic Web. In Code and the City. eds. Kitchin, R., and Perng, S-Y. London: Routledge. 200-214.

Otherwise, here’s a shorter version I wrote in Slate:

Graham, M. 2015. Why Does Google Say Jerusalem is the Capital of Israel Nov 30, 2015

We also have an earlier blog and webcast on the topic (and here's Washington Post’s coverage of our work). 


New publication: Mapping Zombies

Not often do I get to write about zombies, internet geography, German Wikipedia articles, cats, and all in the same chapter. But that is precisely what I got to do in a piece on “mapping zombies” that I carried out with Taylor Shelton and Matt Zook.

Feel free to download the chapter below:

Graham, M., Shelton, T., and Zook, M. 2013. Mapping Zombies: A Guide for Pre-Apocalyptic Analysis and Post-Apocalyptic Survival. In Zombies in the Academy: Living Death in Higher Education. Eds. Whelan, A., Walker, R., and Moore, C. Chicago: University of Chicago Press.


Zombies exist, though perhaps not in an entirely literal sense. But the existence, even the outright prevalence, of zombies in the collective social imaginary gives them a ‘realness’, even though a zombie apocalypse has yet to happen. The zombie trope exists as a means through which society can playfully, if somewhat grimly and gruesomely, discover the intricacies of humanity’s relationship with nature and the socially constructed world that emerges from it. In this chapter, we present an analysis of the prevalence of zombies and zombie-related terminology within the geographically grounded parts of cyberspace, known as the geoweb (see also Haklay et al. 2008 and Graham 2010). Just as zombies provide a means to explore, imagine and reconstruct the world around us, so too do the socio-technical practices of the geoweb provide a means for better understanding human society (Shelton et al. 2013; Graham and Zook 2011; Zook et al. 2010; Zook and Graham 2007). In short, looking for and mapping geo-coded references to zombies on the web provides insight on the memes, mechanisms and the macabre of the modern world. Using a series of maps that visualize the virtual geographies of zombies, this chapter seeks to comprehend the ways in which both zombies and the geoweb are simultaneously reflective of and employed in producing new understandings of our world.