data

4 I THE DATA REVOLUTION

revolution' is underway- referring not only to the growing value o f data and how they are reshaping society, but also to the nature and production o f data.

While there are thousands of articles and books devoted to the philosophy, politics and praxis of information and knowledge, it is only in the last decade that there has been sustained critical reflection on the nature of data, their production and use. In the past, when attention was paid to data it was usually to consider in a largely technical sense how they should be generated and analysed, or how they could be leveraged into insights and value. Little consideration was given to the nature of data conceptually, philosophically and politically, or their contextual and contingent production, circula- tion, usage and effects across all aspects of daily life. The principal aim o f this book is to consider data and the data revolution from a critical perspective: to examine the nature, production and politics of data and how best to make sense o f them, their uses and con- sequences. To supply an initial conceptual platform, this chapter examines the forms and nature o f data.

W h a t a r e data?

The Oxford English Dictio11ary defines data:

1. As a count noun: an item of information; a datum; a set of data. 2. As a mass noun.

a. Related items of (chiefly numerical) information considered collectively. typically obtained by scientific work and used for reference, analysis, or calculation.

b. Computing. Quantities, characters, or symbols on which operations are performed by a computer, considered collectively. Also (in non-technical contexts): information in digital form.

This definition reveals data to be representative pieces of information about phenomena and the input for (and output from) computational processes. Data reflect some aspect of the world (e.g. a person's age, height, weight, colour, blood pressure, opinion, habits, loca- tion, etc.) or the results of an experiment (a controlled condition for determining some- thing about phenomena) captured through some form of observation or measurement ( e.g. a scientific instrument, sensor, camera, survey, etc.). They can also be derived in nature (e.g. data that are produced from other data, such as percentage change over time calcu- lated by comparing data from two dates), generated indirectly as the exhaust of another process (e.g. a database of social media posts), and produced through inference, prediction and simulation. Data can take a number of forms - numbers, characters, symbols, images, sounds, electromagnetic waves, bits - and be recorded and stored in analogue or digital form. Good-quality data are discrete and intelligible (each datum is individual, separate and separable, and clearly defined), aggregative (can be built into sets), have associated metadata (data about data), and can be linked to other datasets to proVide insights not available from a single dataset (Rosenberg 2013).

S. mn!ogiczlly the n-md m:: - = - . " < : . data are elements tml:G:

=i::::,,.'c, er, data refer m tim: I _ 1 ·,-'f llm, e::cpetimentsan:if

:::.,-¼ S·-"'1'.i <'S data are ··.:::cs=:::n:tsof data th2t hai'2: b : 7 - - .m-d Dodge 2011. A s !

d .; cnm msJ 1!0 --.:± 't?e..&. ..-m;ch i".._as --: · -:::=r- er sa -rl -&cm r:s::.=

:: -_. -:.:mg, tllis 00,,S 5i: . · _ , . . !:.:s ::eoome so = -· 4 m h e r f i r m a

& : E 'i\""!!i:re . . . . : : : ; 'De = c t s c i me t a m , m • · " " - - - q ; h m-eas:x.

- ·-.- --;.. _ ••ml n c llZ\""cd::? = ::-=-,,,;.- h a . e -"'"t c;;:iwla:rs h<rre n o ! a

._,.._, rust used i..-i f r= and the de-.cltJ¥:I - e : g 2013). In pa,."1fco

evidence than { : : ::i:::a..."C..ie!ll:aJt and £ e 0 0 • ..; .::::x. cse o f formalised p::-

. • «. s . analysis (such 2 S :!: yticaJ an.d t l y o f the m - analysis or inteLJ

: : _ . . . . : e : g 2013). And t h ·r • : ;,-;e meaningful m:

....!. >_,_ d their meant - "'.::: es.p:,es.xtl i n any

O b i .:::::i. c:sc o f mrletly. ro:, - _ --- =i:.-r::.a,-alauthn t:,

.:::..:. cmill d.i ::::z;: 1998; Gar.:.: -===· ·_::.:::"' r e z n i n g o f dat::

I I

16 THE DATA REVOLUTiO:".;

from a sensor, choices have to be made Rgatding the sensor specification and quality, its calibration, its siting, its sampling rate. how the data are recorded and analysed, how to treat errors and gaps, and so on. These choices are made and framed within an operational context, shaped by prevalent knowledge, established practices, existing systems, cultural lenses and intended uses (Bell 2015; Loukissas 2018). Moreover, interpreting those sensor data meaningfully 'requires an understanding of the instrument - for example, what do the sensors detect, under what conditions, at what frequency of observation, and with what type of calibration?' (Borgman 2007: 183). Data then do not simply represent aspects of the world; they are partial constructions about the world (Desrosieres 1998; Poovey 1998). Or, as Borgman (2015: 17) puts it, '[d]ata are neither truth or reality', though they are used to assert truth and reality.

From this critical perspective, scientific knowledge is understood as being 'produced, rather than innocently "discovered'" (Gitelman and Jackson 2013: 4), its supposed neu- trality and objectivity a discursive fiction (Ribes and Jackson 2013: 165). Instead, how data are ontologically defined and delimited is cast not as a value-free, technical process, but a normative, ideological and ethical one that has consequence for subsequent analy- sis, interpretation and action (Bowker and Star 1999; Reigeluth 2014; Markham 2017a). The production of data is a social practice, conducted tluough structured and structuring fields (e.g. methods, concepts, expertise, institutions) that are shaped by and contribute to configurations of power and knowledge (Ruppert et al. 2017). At the same time, data are also open to 'the unplanned, unexpected, and accidental', moulded by happenstance (BoeUstorff 2013), as well as guesswork, hunches, wrangling, and compromise (Neff et al. 2017). Indeed, data work (e.g. collecting, processing, analysing) rarely operates smoothly and tweaks, badges and repairs to achieve a working outcome are the norm not the excep- tion (Pink et al. 2018b). As Gitelman and Jackson (2013: 2) put it: 'raw data is an oxymoron'; 'data are always already "cooked" and never entirely "raw'". Moreover, data do not follow a preordained recipe (Boellstorff 2013) but are 'lively' in their cooking and consumption (Lupton 2016), and contain noise, errors, biases and gaps. Likewise, a dataset might not be uniform or consistent, consisting of an amalgam of data produced from varying sources (Tanweer et al. 2016).

Further, data are not immutable, wedded to a particular form or unchanging over time and circumstance, rather they need to be maintained and can take shift in character across media, platforms and use (Leonelli et al. 2017). As Markham (2013) observes, part of the issue is that data are understood as things that can be harvested, rather than as a pro-• cess that is continually in the process of taking place. In this sense, data are generally approached as antic (discrete, fixed objects) rather than ontogenetic and emergent (always ir1 a state of becoming) (I<itchin and Dodge 2011; Markham 2017a). As discussed in Chapter 2, this process of becomirlg is thoroughly intertwined and inseparable from the assemblage of people, institutions, instruments, infrastructure, discourses, regulations, laws, standards and finance that shape and are shaped by data. Rather than existing as somehow separate from world they represent, data are co-emergent with, and co-constitutive of, their assemblage (Tanweer et al. 2016; Pink et al. 2018b).

::::s. J ? n is _ : _ - = .:-,rn c;oc?ally pro:ma:rl,. c : :-:-,,.: .ct t l 2:016). D a g c

t:e:.} and phY2G!l mn:::i ::::=· ;;-=-----,.,:mo ;,ndJ 1l!J - :::r ... , ; s ,of a "'<! o i f u . .:....:::::. ;'.°'-'-cib1C CO: ·:::::: z = =e mnc!ed and . . . - t i

fnfrastructmesm:. - at. 20!6}. P-.Elij-c=.121:: .::::: :::_-=-:!ii:ing er ccletro:a '1: -i=c;:s:::mas tned..""CZ_:

;:::..r- ::r aai 2D1Sb). s . c , rrr:,

- z a ? £ ' s ami mte.p dzll =ewcr:.W. but ti! 2 5 (

Z . 7 _ c : • " t O! action 2!!ii .:..:::.a::::::b:: ,E:::J irx::::rn:m-" ts:mm z - h , e l p t i :

""" man mmms ! =- httrai = .. ,fW..S..= fm pe : .... '"7--.a.:s a r t reil ; :::.:._--:c .. ruFn - · ·'!'::f':!: X-3-'- iilre TI:.y f::.:t

:. - g:::es::nZJ!& a:.tl. - - - - s n me.=i· =- t:dm:ral SJS'!'e .::z .=::.:: : : i fn .. :::-· -: ->;:,"(IR!:, z,:;::

.:: = e z .... ...=:.-:- ..,..-PS[;,,,. zcra:ssz.1 . .:-= --,....::::; mme

. ·· :::ctfs y .=: •: •=- z-,,i ·.-:.:.=::::. =C•-e&:efre-ct:so · . .:.:..::c.: '::' ,!o:ct!l:;

,::-,-,•--::--- m E -::..: = : : = - m o o . : : : c ; ; . ·.:::..;;: =: - ::"«! B,=,.]) 1_i2lf. .'. - , . . . . . -'"-!'S !"'?--fn; _ -;:.::;- -=r:!m ffi; 1mw