Do Algorithms Dream of “Data” Without Bodies?

My colleague Craig Purshouse chaired a Work in Progress Paper at our Health Law and Regulation Unit. The interaction between health law and new technologies is a fertile area for thinking about the cross-cutting themes in law, technology and health policy. In a talk, Rebooting Personalized Health and Well-Being: Data Protection Challenges at the Mobile Health and Sensor Technologies as Key Drivers in Precision Medicine I had a brief opportunity to re-visit one of the themes - algorithms and their role in precision medicine. Some of the ideas generated from the talk by Professor Sir Munir Pirmohamed and questions from the floor will form the basis of a paper to be delivered at a Conference organised by Chris Till at the appropriately named Digital Health, Digital Capitalism

The title of my paper, "Do Algorithms Dream of “Data” Without Bodies?" may remind some of the excellent science fiction work by Phillip K Dick. Thereafter, the focus diverges.

Abstract

L'esprit géometrique is ushering a new wave of method and thought in the way the avalanche of raw big data is to be harnessed to promote innovations and efficiencies in the health care industry. Algorithms are technological devices of inscription, which are themselves the direct product of not only ideological, political, technical and institutional forces but cultural and technological convergences between biosensor technologies and evolving human machine interactions. How do we begin to engage data protection rules when confronted with questions regarding the ethical implications of individuals being re-imagined through bodies of data within the formal ontology of big data and algorithms? By focusing on the lexicon of classification, standardization and reification of persons, we can begin to understand the salience and primacy of the organizational logic of algorithms. The failure to recognize the indivisibility of data and its subjects is not a malaise that is to be consigned to the annals of 18th century orthodoxy, which elevated the rationality of particular forms of actuarial calculative devices. We see not dissimilar intuitions in data protection rules as policymakers continue to focus on the analytical distinctions between health information of bodies and bodies of health information when their efforts would be rewarded instead by reflecting on the ontologies of algorithms, which elide the justice claims of individuals to a right to the protection of their personal data. In an age of permissionless innovation, we may in our haste to embrace the benefits offered by techniques of analysis, underestimate the fact that much of what constitutes knowledge of individuals, persons and things may be shaped by the “ontology of capture” and “grammar of action” of algorithms.

Here is a taster of what is likely to follow from the Introduction to the paper:

L'esprit géometrique is ushering a new wave of method and thought in the way the avalanche of raw big data is to be harnessed to promote innovations and efficiencies in the health care industry. Just as the discovery of the rules of inheritance heralded a new era in the science of genomics, we now seem poised to make similar breakthroughs in health care through the use of big data and algorithms. Algorithms are technological devices of inscription, which are themselves the direct product of not only ideological, political, technical and institutional forces but cultural and technological convergences between biosensor technologies and evolving human machine interactions. Algorithms will now be deployed to create meanings and produce relevance from volumes of raw data. N=all, is the trope that big data seems to suggest – though intuitively we will still have to distinguish between ‘signals’ and ‘noise’. It is however useful to recall not dissimilar expectations and challenges of those who put much faith in the technological inscriptions provided by the avalanche of printed numbers to de-code individuals and society during the Enlightenment. In his essay, “Making Up People”, Ian Hacking describes how processes for inscribing vast amounts of information enabled the State and elite professional and commercial institutions to transcend the limits to human vision and provided considerable opportunities for making breakthroughs, particularly in the realms of science, education, health, business and public order. These modes of collection, inscription and sense making, Hacking suggests, constructed a “space of possibilities for personhood.” Hacking draws on the tasks undertaken by doctors and statisticians in respect of the data relating to health information of bodies to illustrate what is at stake, when projections from data diverge, depending on the functional goals being pursued. He suggests that care should be taken when pronouncing on the constitution of subjects based on projections of raw data. How do we begin to engage data protection rules when confronted with questions regarding the ethical implications of individuals being reimagined through bodies of data within the formal ontology of big data and algorithms? By focusing on the lexicon of classification, standardization and reification of persons, we can begin to understand the salience and primacy of the organizational logic of algorithms. The failure to recognize the indivisibility of data and its subjects is not a malaise that is to be consigned to the annals of 18th century orthodoxy, which elevated the rationality of particular forms of calculative devices. We see not dissimilar intuitions in data protection rules as policymakers continue to focus on the analytical distinctions between health information of bodies and bodies of health information when their efforts would be rewarded instead by reflecting on the ontologies of algorithms, which elide the justice claims of individuals to a right to the protection of their personal data. We can paraphrase the question implicit in the article's title, “What do algorithms want?” By doing this, I want to raise the question whether we understand the techniques of algorithms enough to balance the potential benefits of data driven technology with safeguards for the protection of privacy and personal data. In an age of permissionless innovation, we may in our haste to embrace techniques of analysis, underestimate the fact that much of what constitutes knowledge of individuals, persons and things may be shaped by the “ontology of capture” and “grammar of action” of algorithms.