The perils of Big Data: How crunching numbers can lead to moral blunders Posted on : Feb 18 - 2019

The last six months have been brutal for McKinsey & Company, the storied consultancy where I spent my first years after college. A barrage of negative headlines have accused the firm of raising the stature of authoritarian governments, working with ICE (ended after outcry from firm alumni) and failing to advise a client not to use a business partner engaged in bribery. Most recently, reporting revealed that the firm had advised opioid maker Purdue Pharma on how to “turbocharge” sales.

The stories share a common thread: none of them implicate McKinsey directly. Rather, they point to illegal and unethical behavior a few steps away — behavior that McKinsey overlooked or supported. McKinsey’s official response to one of the articles suggested that the firm was being held to a uniquely high bar, that they had been declared guilty by proximity.

In a sense this is true, but this line of thinking exposes a common flaw in business ethics. It’s remarkably easy to overlook massive moral transgressions when you are doing business from a distance, reviewing entries in a spreadsheet or numbers in a presentation.

My time at McKinsey inspired me to study the history of quantitative management and the ethical challenges it creates. What happens when firms like McKinsey parachute into new companies to crunch data and offer advice?

Surprisingly, the history of American and Atlantic slavery offers insight into this question. Running a slave plantation involved lots of data carefully entered into paper spreadsheets and reports that were passed along to absentee owners in England. From the comfort of counting rooms, plantation owners could review this data without having to think too hard about the people it represented.

Some planters received standardized reports every month from their sugar plantations in Jamaica and Barbados. These careful records tracked the daily tasks of the hundreds (sometimes thousands) of people they enslaved, all with an eye to maximizing profits. The accounts monitored the output of plantations as well as the “increase” and “decrease” of laborers, slaveholders’ chilling economic shorthand for births and deaths.

When you understand the context of these records — high mortality, punishing slave labor, racialized violence — the records are horrifying. But without that context, they erase as much as they reveal. They look like antiquarian versions of Excel spreadsheets. And, absent a moral perspective, the productivity enabled by data-driven analysis could be seen not as a marker of degradation but of progress.

Planters in the American South also entered data into early versions of spreadsheets. The most sophisticated among them monitored enslaved people’s productivity in gridded journals, collecting data on the pace of cotton production. They tracked cotton picking on an individual basis, weighing output as many as three times per day. The surviving account books from these plantations contain thousands of data points. Even as the data illuminated productivity, it obscured other aspects of plantation life. It hid the immense human costs of slavery. View More