Decolonialising Data

After lunch at Open Data Camp 7, consultant Edafe Onerhime explained why she had wanted to raise the issue of ‘decolonialising data’.

She is Nigerian as well as British, she explained, and her research had brought her into contact with the issue of “white by default.” As an example, she said, the census puts ‘white, British’ at the top of its identity options; which affects the data that is collected, because people have to scroll down to find other options.

At the same time, she said, a lot of technology relies on data collection and analysis that is carried out in India and countries and Africa. For example, a lot of tagging of pictures for what is billed as AI is done there. The taggers may not receive much money; yet their work is consumed by affluent, Western consumers.

So, the question: “How can we remove the effects of colonisation on data collection and use?” Campers at the session felt the first issue is to recognise that there is a problem, and the second is to get more voices at the table. Then, the assumptions underlying the collection and use of data need to be interrogated.

One participant pointed out that some decisions, such as the order of categories, may have been made to accommodate technology constraints that no longer hold. But another pointed out that some, perhaps including those decentralised work models, might be built into the business decisions of the often large corporations that have made them.

“There is no point pretending that technology can be neutral, because it can’t,” one camper argued; but that just makes it even more important to acknowledge other perspectives. In doing so, the session agreed, it was important not to make decolonialisation a “tick box” exercise; for example by putting one BAME person on a board or project and thinking job done.

Who benefits? Who is harmed?

The session then turned back to the way menus that collect data can affect the data collected. Often, data will be collected the way it is collected because “that is the way it has always been done.” But that’s not a reason to continue doing it that way.

Meanwhile, a common justification for putting ‘white, British’ at the top of something like a census menu is that this is the category that will be chosen by most people, at least in the UK. But, participants pointed out that most people would surely pick this option, if they had to look further down the menu. Canada presents its identity options in alphabetical order.

Users who are not ‘white, British’ may also be given a limited range of options to express their identity: such as ‘African’. So developers should think about whether, in trying to standardise data by using categories, they are providing enough categories to capture people’s lived experience; and also go and ask.

In all cases, Edafe said, the critical thing was to interrogate the decisions being made. “I think the question we have to ask is who does that benefit and who does that harm. Particularly when we want systemic change, then we have to think about how data affects the system.”

Anyone interested in the issues raised by the session can comment on the liveblog, or follow Edafe on Twitter, or email her:

[Session Notes]

One thought on “Decolonialising Data

Comments are closed.