25 February 2014
Lessons from science: Digital Humanities, History of Science and Technology, and juking the stats
One of the joys of being both part of the Digital Humanities community and an early riser is brushing one’s teeth at 6.30 am, checking one’s email, and seeing the days Humanist Discussion Group messages dribble into one’s inbox. Humanist was established by Willard McCarty in 1987 and each morning Willard compiles and arranges the previous day’s correspondence and pushes them out to the group. On more than one occasion, such is the usual punctuality of his efforts, have I glanced at a Humanist-free inbox and considered, albeit briefly, Willard's health or safety, only to realise that he is probably just in another time zone...
At Digital Humanities 2013 Willard was bestowed with the sobriquet the Obi-Wan Kenobi of Digital Humanities - the 'Father' accolade having already been granted to the person whose name was on the lifetime achievement award Willard was receiving, Roberto Busa. Last summer, as part of their ‘Being|Human’ festival, King's College London gave Willard a forum to repeat his address. In it, and among (many) other things, Willard dwelt at length on the relationship(s) between information technology and computation and society and culture, the co-evolution of man and machine, and how humanities scholars have used machines in their research.
Many humanities scholars will be forgiven for not identifying with this aspect of DH, that DH is the study of humanistic phenomena using digital technologies and the of the role of digital technologies within humanistic phenomena; or as Kathleen Fitzpatrick once put it, DH is:
A nexus of fields within which scholars use computing technologies to investigate the kinds of questions that are traditional to the humanities, or […] ask traditional kinds of humanities-oriented questions about computing technologies.
The intersections between this latter category of work and the History of Science and Technology (hereafter HST) are obvious; indeed, during a brief discussion after Willard's lecture I was delighted to find him most enthusiastic toward the then still relatively recent move to King’s from Imperial College of the renowned Centre for the History of Science, Technology and Medicine.
Which brings me to my point: how can the other side of DH, the DH as using digital research methods to explore phenomena in the humanities, benefit from close(r) dialogue with HST? Franco Moretti's most recent paper – ‘Operationalizing’ – uses the work of another path finding Obi Wan Kenobi of scholarship, Thomas Kuhn, to frame its argument. So like Moretti I went back to HST, back to Kuhn, and in particular back to his 1961 essay ‘The Function of Measurement in Modern Physical Science.’ Here Kuhn makes three important claims whose relevance to digital research, especially quantitative work, should be clear:
First, quantitative work comes from qualitative work. As Kuhn writes: ‘qualitative research, both empirical and theoretical, is normally prerequisite to fruitful quantification of a given [scientific] research field.’ (185) Quantitative methods are not inherent to scientific research but rather were introduced into sciences ‘that had previously proceeded without major assistance from them.’ (162) In this light ‘quantitative facts cease to seem simply “the given.” They must be fought for and with.’ (171) Quantification then is a human act. ‘Numbers,’ Kuhn writes, ‘gathered without some knowledge of the regularity to be expected almost never speak for themselves. Almost certainly they remain just numbers.’ (175) Or as Louis Pasteur once put it: “In the field of observation, chance favours only the prepared mind.” (179)
Second, data is always approximate. Why? Because measurement is never ideal, regularly insecure, and often at limits of what is possible. As a consequence, scientists rely on “reasonable agreement” (165-166) about data to move forward, agreement that accepts measurement, at least in early experimentation, as often deriving from investigations of an instrument as much as investigations with it. (188)
Third, scientists adapt measurement to fit theory. Although ‘overly close agreement [in numerical results] is usually taken as presumptive evidence of data manipulation,’ (165) at the same time ‘in scientific practice, as seen through the journal literature, the scientist often seems,’ writes Kuhn, ‘to be struggling with facts, trying to force them into conformity with a theory he does not doubt.’ (171) This goes back to measurement, for Kuhn continues: ‘one of the tests for reliability of existing instruments and manipulative techniques must inevitably be their ability to give results that compare favourably with existing theory. In some parts of natural science, the adequacy of experimental technique can be judged only in this way.’ (172) Here Kuhn goes on to describe Dalton’s Law and how as a consequence of extant data not conforming to the law, experimentation with and improvement of instrumentation followed so that data was brought in line with the law; a fascinating example of the interrelation between data, experimentation, theory and research. (173)
There is much here for humanists – digital or otherwise – to chew on, though perhaps it is at this last claim that we humanities scholars reach a crucial point of departure: for a qualitative-led approach to measurement and data collection, which accepts adapting instrumentation to get the data the theory requires, relies on there being reasonable agreement over theory or the possible theoretical options. And whilst in the case of science there are, it would appear, a discrete number of options, at glance at the humanities reveals a seemingly infinite number of possible nuance laden scenarios and explanations for any one given field of research. There does not, in short, appear in the humanities to be a stable or discrete set of qualitatives from which quantitative measurement can fruitfully develop (and if so, they may be too broad to be of use); and if a humanist adapted their measurement of data to fit their qualitative work, either empirical or theoretical, they would likely – quite rightly – be accused of juking the stats.
Or would such criticism be unfair? Could it not be that as quantitative work in the humanities can take the appearance of representing certainty by virtue of numbers and stats being somehow ‘scientific’ (a relationship between numbers, truth and science Kuhn shows to be false), interpretation of measured, derived, quantified data highlights – somewhat uncomfortably – the extent to which humanities scholars fit their choice of sources and their interpretations of them to their theories and assumptions? Whilst research published in the humanities, and in particular research based on quantitative work, is expected to describe what sources were used, why they were chosen, and how they were used (certainly for historians, this is week one undergraduate stuff…), few examples of qualitative research in the humanities are, from personal experience, as open and honest about their approach to sources as Rens Bod’s recent A New History of the Humanities. In a frank and somehow liberating passage of his introduction, Bod writes:
My decision to focus on principles and patterns will, however, often lead to surprising choices. Many a famous humanist, historian, or philologist will be mentioned only briefly – if at all – while other scholars are dealt with at length. More than once I will describe a well-known work with a single sentence, not because I consider it unimportant or not influential, but because it did not contribute much to the quest for principles and patterns (10)
Or, in other words, based on his theory that the history of the humanities could be characterized by a quest for principles and patterns, Bod chose to focus on humanities research that sought principles and patterns. And this he admits.
Anyhow, I digress. The point is that the HST, with the example here being research into the role of measurement and quantification in the sciences, can offer something to DH, to those humanities scholars using digital methods and data to do research: it can offer a critical apparatus for thinking through the relationship between quantitative work and established patterns of qualitative research; having considered the role of experimentation and the investigation of instrumentation in the sciences, it offers a parallel to the ‘productive play’ that by necessity surrounds much research that uses data and digital methods to explore humanistic questions; and it offers a narrative digital humanists can bounce off to erode the lingering fiction – if not in our minds, then in the minds of others – of ‘data driven’ research being a category of humanistic enquiry: for HST reminds us that in both scientific and humanistic research sources do not speak for themselves, they are always approached with qualitative baggage in mind, and the combination to varying degrees of theory and source material, be those sources data or archival material, is what defines scholarship.
I am guilty of cherry-picking Kuhn. And there is of course a rich tradition of debate, and disagreement, that followed both ‘The Function of Measurement’ and his later, more well known The Structure of Scientific Revolutions (1962) that I have ignored. Nonetheless ‘The Function of Measurement’ suggests, to me at least, that HST has spent enough time researching the use of technology for digital humanists who use technology – and indeed those digital humanists who research the use of technology – to take note. With this in mind having Bruno Latour, author Science in Action – an enormously influential text in HST –, open Digital Humanities 2014 seems to me a very shrewd move and will hopefully prove an ideal compliment to Willard's soon to be published Busa Prize lecture.
And if you want to see some science in action this Spring – hopefully with no juking of the stats in evidence! – the exhibition 'Beautiful Science: Picturing Data, Inspiring Insight' is open in the Folio Society Gallery at the British Library until 26 May 2014.