The End of Random

Digitization is eradicating randomness. It’s a bold conjecture. But one manifesting itself in myriad ways.

If you’ve been following the news over the past couple of weeks, you’ve surely read more than a few articles about United flight 3411 from Chicago to Louisville. When videos of Dr. David Dao being dragged off United Airlines flight 3411 hit social media on it sparked outrage. Statements from United and United CEO Oscar Munoz fueled the fire of a quickly escalating PR crisis.

Initial reports of the incident came from passengers on the flight. Audra Bridges, who was one of those passengers, reported the airline needed four additional seats so United initially offered $400 and a free hotel to incentive volunteers. When no one volunteered, the offer was doubled to $800 and with still no volunteers, the airline selected four passengers to involuntarily leave the flight. Bridges reported Dao was “told he had been selected randomly to be taken off the flight.” But as USA Today noted, Dao and the other individuals denied boarding were likely not chosen at random. Rather, a computer algorithm probably chose these individuals from the full list of passengers on the flight. They were targeted based upon their status with the airline, the fare they paid for their tickets, and perhaps a host of other variables. The Courier-Journal, who spoke with Bridges after the incident, reported “a manager came aboard the plane and said a computer would select four people to be taken off the flight. One couple was selected first and left the airplane” and then Dao was confronted. We all know how the rest of this part of the story ends. But it’s another element of the story I want to focus on here. 

As digitization infiltrates increasing swaths of our daily lives, we will likely see algorithms play a more defining role in our actions and activities. Algorithms often operate in the background so we will see the manifestation of these algorithms without always seeing their presence explicitly. Cultural and social norms will be upended as heuristics give way to programmatic rule. 

Algorithms are designed based upon a perception of how the world works and how it should work. Algorithms are being used to order your Facebook and Instagram feeds, pick which advertisements are served up to you and are increasingly redefining real world issues like lending decisions, prison sentencing, and police force resource allocation.

Increasingly activities that are digitized are being governed by algorithms and the implications vary. Because algorithms often learn from input data there is risk of bias transfer if the underlying data are biased. This could include income, race, geography, or any host of other underlying variables whether the variables are explicitly targeted or just correlated with something that was implicitly taken into account in past decisions that are now being utilized to train algorithms. 

While we might not recognize it, digitization is replacing randomness with codified, programmatic approaches to decision making. We need to increasingly recognize the role algorithms are now playing in decisions that might have historically been random and take a more proactive approach to ensure the outcomes being served up are the societal outcomes we desire. Random is disappearing and we have a choice in what replaces it.