National Tracking Ecologies
Anne Helmond, Carolin Gerlitz, Esther Weltevrede, Gabriele Colombo, Astrid Maria Bigoni, Bev Skeggs, Marc Stumpel, Marta Severo, Sarietha Engelbrecht, Lonneke van der Velden
The team came from a range of different interests, including understanding the mechanics of connectivity on the web, making visible the invisible (eg cloud computing), asking what where, when and how trackers operated, the dynamics of online data, dynamics of social web economies, activism. There was a general interest in the ‘social’ of social media, hence the widget provided a performative object for investigation into the forms of engagement that are enabled.
The aim of this project is to map the encounters with trackers by Ghostery-users per country. By browsing, internet users communicate with third party servers through trackers that are embedded in websites. Ghostery is an anti-tracking browser extension www.ghostery.com that allows users to identify the invisible web, devices that track user activities online and the services associated to them. Ghostrank collects anonymized data about the presence of trackers from Ghostery users that opt-in to share their encounters with Ghostery’s database. The dataset consists of all the trackers loaded by Ghostrank users in the month of May 2013, including the frequency per country. In the process of anonymizing the data, the IP addresses of the Ghostrank users were geolocated and discarded. Ghostrank reports on tracker data on over 26 million sites, and has eight million monthly users worldwide 1
]. We are trying to see how trackers relate to each other and their environment. In doing so we mapped national tracking ecologies (Fuller 2005:2)
Original research question: Do national webs have different tracking ecologies? What is the extensiveness of tracking and trackers per country?
Revised research question: Which trackers do users encounter & participate in when surfing from specific countries?
We used the DMI Tracker Tracker tool and Gephi: https://gephi.org/
. The Tracker Tracker tool was conceived at the Digital Methods Winterschool 2012 in January. It is built on top of the anti-tracking plugin www.ghostery.com and allows one to identify the invisible web, devices that track user activities online and the services associated to them. Our data set, ‘Ghostrank’, was courteously provided by Ghostery. The data contains information on the trackers that users encounter, based on the country from where the user is located. The data set contained 44 countries and was based on trackers encountered in the month of May 2013. Visualising the network of trackers Firstly, the data was arranged in Open Refine. The columns were renamed and then exported as a .csv file. The column are count, country, the name of the tracker and which category the tracker is (analytics, ad, widget, tracker). With excel pivot table function, we calculated the total count distribution of trackers by country. [image: pivotTable.jpg] Then, this .cvs file was converted into a bipartite graph for Gephi (a .gexf file) by the Table2Net
tool. Then, we opened this file in Gephi. In our directed graph, nodes are trackers and countries and edges represent when users in a country have encountered a tracker. We visualised the graph with the ForceAtlas
2 algorithm. We tried different gravity/scaling combinations yet the graph was not so readable because all trackers are generally connected to all countries: as a result, the graph is strongly interconnected and no cluster or group can be easily separated and recognized. So we decided to work with ranking. The size of the nodes is proportional to the total count. [image of the graph] The top ten visualisation In order to visualise the top 10, the top ten trackers were pasted into a single spreadsheet. There the data was normalised based on the total count of trackers encountered in the country. This was then added together to see the percentage size of the top ten trackers in each country. We triangulated lists of the trackers in countries to find the trackers that were unique to a country. This was done in two steps: first a group of 5-6 countries were triangulated, then all groups were triangulated. The triangulation was done using the Triangulation tool.
What are the unique trackers specific to each country?
Where are the companies that operate these trackers located remains unanswered for now.
- Which types of trackers are most prevalent in each country?
- What is the relative value of the top trackers in relation to all trackers in the country field
Helmond, Gerlitz, Weltevrede, et al. (2013). 'National Tracking Ecologies.' Digital Methods Winter School 2013, University of Amsterdam. https://digitalmethods.net/Dmi/NationalTrackingEcologies
Fuller, Matthew. 2005. Media Ecologies: Materialist Energies In Art And Technoculture. Cambridge: MIT Press.