Data never got its driver’s license

When I started my professional career in 2011, “data driven” was a hot concept. This was when Facebook was still primarily a way to keep up with friends from high school and college, Wikipedia wasn’t yet considered a cite-able source in academic papers, Tableau was just getting picked up by large management consulting firms, and as a society, we were only just beginning to grasp the scale and speed of digital data being generated.

I was young and impressionable and found myself eagerly using the phrase. As a chemical engineer by training who had the privilege of choosing a completely unrelated career path, I found myself in spaces where despite just being out of college, I was often quickly seen as an “expert” in data — whatever that meant. Within two years, I was training colleagues on the idea of data-driven policy-setting for governments. I made some really slick slide decks. I was trucking 100mph down Data Driven Highway, singing at the top of my lungs with no end in sight.

In hindsight, it’s hilariously sad that I missed the warning signs and huge crashes on the side of the road. Crashes like:

• The global financial crisis of 2007, which occurred despite a plethora of data available on the sub-prime mortgages (learn more: The Big Short)
• The codification of mass discrimination through algorithms that simultaneously have increased the environmental cost of data analysis (learn more: Be Careful What You Code For, by danah boyd of Data & Society; the work of Timnet Gebru)
• Massive wealth inequality incentivized by measuring economic success solely on the basis of GDP, a completely outdated and simplistic metric that excludes valuation of social goods (learn more: Mariana Mazzucato on what is economic value; Bhutan’s Gross National Happiness)

Truth is, data was never driving. Data never got its driver’s license. Humans do the driving, because humans define and collect the data, humans choose how to “clean” the data for quality control, humans choose the analysis techniques and confidence thresholds for statistical tests, humans interpret the results, and humans decide what to do with them. It’s humans. Humans all the way down.

These days, I’m cruising in a hybrid compact car at a steady 50mph, checking my rear-view mirror constantly and making sure to keep the radio on to hear other human voices. I listen when my stomach grumbles or my heart aches and ask myself why. I think about responsible decision-making as the result of three key ingredients: quantitative data, qualitative stories, and quintessential instinct. And sometimes, I just get out of the car because driving for drivings sake isn’t what gets me up in the morning.

Published March 21, 2021