This is a blog post from
Has the internet moved on from being a mere ‘virtual’ space divorced from our ‘real’ world to emerging as a space in which the activities of multiple users can help us predict and analyse socio-political behaviour? However nuanced the answer to this question may be, it is proving to be a potentially embarrassing situation for quite a few prominent media outlets. This state of affairs was precipitated recently by a rather singular bit of information from Google Trends: ‘What is the EU’ was the second-most oft-asked Google query in the UK the day after the Referendum results were announced, followed in the third place by the query ‘What will happen if Britain leaves the EU’. Google Trends, as is well-known, is a tool that allows one to find indexed information on search activity that takes place on Google on any subject, over a period of time, and in a specified geographical region.
Xenophobia and the ‘ignorant masses ’
Owing to the certainty with which these tidbits about the referendum were circulated, they soon dovetailed into the larger narrative of how a certain section of xenophobic, ill-informed British voters had imperiled both Britain and Europe. This narrative was in circulation in reputed news and media outlets like The Washington Post , National Public Radio ( NPR ), fortune.com , etc. Even the celebrated talk-show host John Oliver featured it in one of his diatribes.
The narrative, however, has now been >contested by software developer and data analyst Danny Page who, >along with Remy Smith , a Political Science student from Northwestern University, Illinois, has shown that the number of people Googling ‘What is the EU?’ in the days after the referendum was actually under a 1,000. The number of people Googling the same question before the elections in May 2016 was around 231 everyday; the spike noted in Google Trends was a result of this low base-line. Page further pointed out that there was no contextualisation of the data, as it did not show if the voters were from the Leave or Remain camp, or eligible for voting, etc.
>@rohitguptahpf>@bencasselman even if you consider the data valid (such as the dates being accurate) the UK has 65m people, 17m voted remain
— Daniel P. McGrady (@dmix) >June 28, 2016
It must, however, be noted at this juncture that there were events and circumstances that might reinforce this narrative, given the rise in the attacks on immigrants and peoples-of-colour in Britain after the referendum, the varied levels of populist anti-immigration rhetoric championed by the Brexit leaders (and many other parties in Europe), the murder of British Labour MP Jo Cox, the uncertain future of Britain’s already devastated economy and the attendant impoverishment and anger among its working classes (especially white), etc. This turn of events, however, does bring to our attention the need for exercising caution while making analytical generalisations about socio-political phenomena using internet data.
Search engines as analytic tools
^ Google searches reveal a lot about the human being users. But it's important to be nuanced when interpreting the analytics of such searches.
Using Google Trends has proven to be helpful in accurate analysis in other cases. In a paper published in the journal Nature in 2009, four researches from Google along with their colleague from the American Center for Disease Control and Prevention proposed a system whereby it became possible to pinpoint locations of outbreaks of influenza by correlating Google-search queries with other variables. Such approaches have also been tried using the recorded number of visits to particular web-sites and Yahoo search results. For many noted media scholars like Dr. Richard Rogers of the University of Amsterdam, such approaches could spawn new methods of socio-political analysis where empirical data on online-search habits / interests can help researchers in saying something meaningful about the world at large.
The adoption of such a method of socio-political analysis seems appropriate in an age where Xerox Corporation scientist Mark Weiser’s vision of ‘ubiquitous computing’ — where computers disappear into the ‘fabric of everyday life’ — increasingly comes true. Internet-enabled computing devices are becoming a rather unremarkable part of our everyday lives, much as Weiser had envisioned. Given this state of affairs, it becomes increasingly meaningless to treat the internet as a ‘cyber-space’ separate from the ‘real world’, the prevalent way of thinking for the greater part of the period since the Internet came into existence. However, the development of this method is still at a nascent stage, and that makes the cavalier interpretations such as those made about the Brexit impinge on the credibility of future attempts at such analysis.
The effectiveness of using search-engine queries to build analyses and narratives can be understood by the fact that search engines remain the primary platform for accessing information and content online for most internet-users across the world. But the political implications of search engines hardly end there. The fact that traces of our search queries are maintained in databases is already a source of widespread concern. The selection of results that our Google search queries generate topmost on our screens is decided by algorithms and that has been a source of controversy since the days of Larry Page’s PageRank, Google’s earliest algorithm that ranked websites according to their perceived ‘importance’.
Now that there are tools available to capture ‘trends’ on Facebook and Twitter with geographical and temporal pin-pointing capabilities more attempts at such analysis can be expected in the future.
Such concerns have only increased over the decades as, now, the search engine is no longer solely responsible for inclusion or exclusion of results to a query; in a process called ‘registrational interactivity’, the algorithms display results based on the past preferences of the users, thus shifting the responsibility for the selection of search results onto the users themselves.
In spite of the dubiousness that now clouds the Brexit Google Trends affair, it does reveal the contested nature of the current political environment, where terms like ‘people’, ‘nation’, ‘outsider’, ‘democracy’ etc. are being re-defined in the media and elsewhere. Now that there are tools available to capture ‘trends’ on Facebook and Twitter with geographical and temporal pin-pointing capabilities more attempts at such analysis can be expected in the future. Even though these tools provide real-time data about political behaviour and insights into what goes on in people’s heads at a particular point in time (say before, on, and after the election day), they need to be interpreted in their proper context, with the caveat that such trends may only provide a very limited and superficial understanding of complex socio-political realities particularly in low-income countries, which typically have low levels of internet penetration.
