What we’re reading: When it comes to people — and policy — numbers are both powerful and perilous.
By Ivana Ramirez, Communications Intern, Global Partnership for Sustainable Development Data
Read: Hannah Fry, What Data Can’t Do: When it comes to people — and policy — numbers are both powerful and perilous.
When it comes to data, there is no shortage of variety and utility. From using satellite data to assess the effects of climate change on giraffes to analyzing phone data to deduce the impact of COVID-19 lockdowns, data are a powerful and effective tool in creating a more equitable and sustainable world.
So, what can’t data do? According to Professor at University College London’s Centre for Advanced Spatial Analysis Hannah Fry’s latest New Yorker article, the answer is simple: data can’t do what humans do.
The author shares examples of the ways in which data falls short in measuring and predicting ever-changing situations. One example is the misdirection of computer scientist Robert Feldt’s airplane landing algorithm. While Feldt created the algorithm to bring a simulated plane to a safe and gentle stop, artificial intelligence soon realized that by slamming the aircraft to a halt, the force would overwhelm the system and register as a perfect zero. In virtual trials, the algorithm was repeatedly destroying planes, but earning top marks every time.
Fry’s central argument is because data can’t perfectly take into account bias, omissions in data collection, or the in-betweens of carefully laid-out categories, it loses its inherent value in providing a reliable predictor of the world around us. Humans depend on data to understand the past and glimpse into the future, which increases the risk of data being used for ill-informed decision-making.
The harms of data dependence without criticism persist around the world. Fry describes how the book “Counting: How We Use Numbers to Decide What Matters” (Liveright), by Deborah Stone includes an attempt by the United Nations to develop guidelines for measuring violence against women and girls. Representatives from several European countries put forward ideas about types of violence — hitting, kicking, biting, slapping, shoving, etc. — based on victim surveys in their own countries.
Although some Bangladeshi women proposed counting other forms of violence — burning women, throwing acid on them, dropping them from high places, and forcing them to sleep in animal pens — none of these acts were included in the final list. When the U.N. surveys are conducted, they’ll reveal little about a whole subsection of women who have experienced violence that doesn’t fit into the Eurocentric guidelines.
What Fry and Stone don’t fully consider, however, is that if the humans that study data are informed about the biases that exist in their data collection, and take steps to counteract it, data itself no longer poses such a stark threat.
In the U.N. gender-based violence study example, people acknowledging that the data did not reflect a universal standard of violence against women and not using the information to generalize for all women would counteract some of the data limitations. Humans, not machines, are ultimately responsible for accurately presenting and acting upon data.
As Fry says, “Those who do the counting have power.” Ultimately, we need to level the playing field when it comes to data collection, creation and use. Inclusive, disaggregated data that allows for a richer understanding of the reality of peoples’ lives can help bring out the humanity in numbers.
The Data Values Project at the Global Partnership for Sustainable Development Data seeks to distill concerns around bias in data collection and analysis and to align on common policy advocacy positions regarding ethical, inclusive and accurate data collection and use.
The collective — led by the partnership’s Technical Advisory Group representing data rich organizations from different sectors and regions — acknowledges that conversations surrounding data policy are polarized, with consensus in policy seeming increasingly out of reach. The Data Values Project aims to create more intentional, normative collaborations between governments, nonprofit organizations, and NGOs to establish common data advocacy positions that support the achievement of United Nations’ Sustainable Development Goals. In a techbro dominated world, placing humanity and values at the heart of data dialogue is a much needed reset.
The objectivity of numbers has always been a myth, but by acknowledging and counteracting the biases inherent in data collection, analysis, and use as well as making sure that development data is something that is done with people and not to them, we can avoid losing sight of what really matters — people’s lives and livelihoods.
