Open data, institutional data, personal data and social data can be gathered by a data broker, deemed to be adding value by creating unprecedented combinations. Some or all of the original data may not belong to the broker, but their control of data flows and ability to combine different sources takes the existing data points into something new that can be sold and resold. This new object may be seen as beneficial, where the data donor and/or society receives something in return, or what I call ādata foamā , where the new product or service has little or no benefit to anyone other than the seller. This chapter explains the concept of data foam , using the specific example of the increasing use of metrics in researcher evaluation.
The concept of the surveillant assemblage (Haggerty and Ericson 2000) or data double (Poster 1990) is a familiar one. People are aware that their data are profitable to corporations, for marketing and insurance purposes, crime prevention and control and myriad other uses. The privacy intrusion is seen as acceptable, if they receive something in return (Gordon 2014). This is a part of the price we pay for cheaper and more personalised products and services, and has come into its own with the rise of āfreemiumā apps for mobile devices and the seemingly almost compulsory engagement with platforms from Google and Facebook in order to engage with modern life. Our shadows are always with us.
There are so many data-related metaphors now: data flows , data journeys, data friction, data doubles, data ghosts. Why throw another into the fray? Why ādata foamā? Well, the various components of the āassemblageā can be agitated again and again to produce a new and shallow layer of dubious value on top of the seemingly reasonable use of data in public statistics or as payment for products and services. This agitated ephemeral layer, this āfoamā, is another product or service to be soldānot to solve a problem but to create a market or influence decision-making.
Is charging for value-added services really an unproblematic aspect of open data and āleakyā data (social media, institutional/work data, phone data, CCTV , ANPR, data others put out there on our behalf) (Bates 2012)? Who benefits? Whose labour is not compensated or valued? Are platforms really the problem? Is the financialisation of open data, that should be publicly owned and not necessarily exploited for profit (Bates and Goodale 2017), and personal data, that should not be weaponised against a user (Lyon 2003; G.J.D. Smith 2016), the price we have to pay to live and work in the twenty-first century?
Once, my wallet was stolen in a large store in an out-of-town shopping centre. There were no witnesses and, according to the duty manager, the area of the store where I had been standing when it happened was not covered by CCTVāit was an open area with no shelves and therefore no chance of employees or shoplifters damaging or stealing stock or fittings. The security camera in āpublicā areas does not exist to protect citizens, but to monitor employees and protect property. Workplace monitoring and productivity metrics are again not for the benefit of the surveilled. So combinations of these data sources, frothed up into something new, are used for decision-making that rarely benefits the data creator. It is exacerbating existing problems.
The new objects created from combining sources, this data āfoamā on top of the existing flows, can be used for monitoring, marketing, assessment and control. Cambridge Analyticaās algorithms worked with their unique combinations of Facebook data to influence the outcomes of the UK EU referendum vote (āBrexitā) and the 2016 US Presidential election. Elsevierās ābasket of metrics ā is used for researcher assessment, comparison and employment decisions. The call centre or warehouse performance dashboard and the Bradford Factor for measuring employee absenteeism are so old fashioned now that you can combine video tracking, social media and physical social interactions of employees and persuade them to do corporate wellness wearables such as Corporate Wellness 360 , which offers corporate wellness packages where staff are provided with devices that generate āsmart dataā and advanced analytics for employers.
Quantification in higher education made its biggest early strides on the student-facing side of the university, providing āricher informationā (Williamson 2016) to support ālearning and teachingā via learning analytics, recruitment and retention management and course and tutor evaluation (Hall 2016). It crept slowly into the work of the researcher, as the regular research evaluation exercises started to take hold in many countries, with academic tenure, promotion and recruitment committees also enjoying the āevidenceā provided by cold hard numbers (Besley and Peters 2009). The quality of research could suddenly be measured by the Impact Factor of the venue in which the outputs were published, the number of times they were cited, a star rating in the research excellence framework (REF) , and the ability of the researcher to bring in grant funding against targets. The backlash against such crude measures (Gruber 2014; Anonymous Academic 2015) has only encouraged the spawning of yet more metrics , bringing in quantification of impact via social media data and other sources (MartĆn-MartĆn et al. 2016) and touting the value of Lambertās (2003) ābasket of metricsā for researcher assessment (Clements et al. 2016b).
Metrics are not merely āneutralā statistics as all chapters in the current edited collection maintain. When multiple actors use a measurement, it becomes a visible artefact that can be compared with other artefactsāa material object that did not exist before (Pine and Liboiron 2015; Moore and Robinson 2016). The production and analysis of these artefacts is a profitable service, be it traditional bibliometrics , which serve disciplines such as the humanities very poorly (Thelwall and Delgado 2015; Stelmach and Von Wolff 2011), or the alternative article level metrics commercialised as A...