Data feminism … introductory thoughts

The ‘Data Feminism’ approach is growing in prominence, with an MIT Press book of this title due for release in 2019. The authors of this book – Catherine D’Ignazio and Lauren Klein – have been sharing their chapers for community review in draft form here. Along with cognate approaches such as ‘data justice’ and ‘good data’, the focus of data feminism is about moving debates about data and society toward issues of structural power, oppression and inequality. As the authors put it at the begining of their introductory chapter:

“Intersectional feminism isn’t just about women nor even just about gender. Feminism is about power – who has it and who doesn’t. And in a world in which data is power, and that power is wielded unequally, data feminism can help us understand how it can be challenged and changed”

We shall we working with ideas from this book once it is published. For the meantime, here is a very useful table from Chapter 1.

Concepts Which Uphold “Imagined Objectivity” Because they locate the source of the problem in individuals or technical systems	Intersectional Feminist Concepts Which Strengthen Real Objectivity Because they acknowledge structural power differentials and work towards dismantling them
Ethics	Justice
Bias	Oppression
Fairness	Equity
Accountability	Co-liberation
Transparency	Reflexivity
Understanding algorithms	Understanding history, culture, and context

 

These are shifts in thinking that move us far from many of the ways that we tend to think about making data ‘fairer’. In particular, right-hand side column warns us against (mis)locating the source of data inequity in the behavior of individuals and/or in the outcomes of a technical system.

In this sense, we should not imagine that we can simply engineer an unbiased system or a transparent algorithm. Neither are the ‘biases’ or ‘un-fairnesses’ of data processes simply due to individual bad actors, rather than the wider structures or systems that they work within. For example, issues of what Safiya Noble identifies as ‘algorithmic oppression’ will not disappear if we ensure that non-racist programmers train their systems on racially-representative data-sets.

Instead, data feminism raises the argument that the problems that we see with current uses of data relate to how power and privilege operate in the present moment. As such, these are not issues that can be carefully avoided or neutralised. At best these are injustices that might be rebalanced in the future.