01. What is Data-Driven...
To start with, some context. Reconfiguring data practices for ADM is a core goal of the workstream of research I lead within the Data Program of the ARC Centre of Excellence for Automated Decision Making and Society (ADM+S). It's a group of projects that draw on the ethical frameworks, principles and approaches developed by my colleagues and others to address data practices and data capability in situ, with the organisations and communities that can benefit from them the most (and are also most likely to be harmed or disadvantaged).
Data-Driven… is a place to think through and report on the progress of that work.
Data-driven is a programming paradigm, where statements are oriented toward data and its processing rather than steps and instructions or many other programming models. It is underpinned by networked digital technologies and data generating systems and hence by Big Data's supposed volume and velocity of data flows (supposed, because the claims of Big Data are never clear-cut). It is also a mantra that has been weilded sometimes blindly by governments and corporations in the fallout from the hype around Big Data. It's in many ways the business end of datafication, and the practical driver of automation.
Some accounts of 'data-driven' focus on the status of data as facts and evidence and its subsequent logical usefulness over 'intuition' or even creative modes of operation. Behind the hyphenated phrase data-driven lies an unannounced set of outputs (data-driven x) and an associated model of judgment, or rather displacement of judgment from 'experts' to a computational system. Data driven 'decision-making', data-driven 'expert systems', data-driven science, engineering, medicine, clinical practice, governance, policy, population health, machine learning models, and so on.
Each application of the idea of data-driven x carries a deeply felt, but often silent ambivalence. Data-driven systems are at the base of a new evil technological-world order (the so called fourth industrial revolution), or offer an almost infinite set of mathematical solutions to any number of problems, and the means for creative (or disruptive / destructive) innovation.
There has to be a way between the critical dystopian and instrumental utopian accounts of the array of new digital technologies, the internet and its mega corporate tech platforms and developments in artificial intelligence (AI). I would argue that both sets of views are dangerous and counterproductive. They set up straw figures and all or nothing solutions.
Is it possible to shape and nudge these technologies toward better social outcomes? By better I mean better than if we didn't have them, better than they so often are. That seems a pretty weak perspective to set out from but at least it's a creative and generative one. There is a chance to take the middle ground and expand it, make it seem common sense. The only way forward is to ensure that the 'do no harm' goals of digital, networked and automated tech holds true at all cost.
There are some key aspects of the data-driven ethos that I'm studying carefully and working on.
First, expertise is being deeply re-configured by data-driven systems, by advances in expert-systems, and by the tensions and negotiations around the involvement of the humans in the loop in machine learning and data science techniques.
Second, there is a newfound interest in data literacy as a response to datafication that might help to empower citizens and revitalise the knowledge and information professions.
Third, accountability, the foundations of both state and corporate governance, is dramatically challenged and seductively enlivened by the ethos of data driven.
These are themes that I'll be addressing through practical challenges, observations, partnerships and activities in the ADM+S projects, and will be reporting on here.