Applied Mathematician (adjunct), UC Berkeley; Head of Research, Haven, Inc.
Babylonian Lottery

"Like all men in Babylon, I have been procounsel; like all, a slave"
Jorge Luis Borges, Lottery in Babylon

The lottery in Babylon begins as a simple game of chance. Tickets are sold, winners are drawn, a prize awarded. Over time, the game evolves. Punishments are doled out alongside prizes. Eventually the lottery becomes compulsory. Its cadence increases, until the outcomes of its drawings come to underpin everything. Mundane events and life turns are subject to the lottery’s "intensification of chance." Or perhaps, as Borges alludes, it is the lottery's explanatory power that grows, as well as that of its shadowy operator the Company, until all occurrences are recast in light of its odds.

Babylonian Lottery is a term borrowed from literature, for which no scientific term exists. It describes the slow encroachment of programmatic chance, or what we like to refer to today as "algorithms."

Today as in Babylon, we feel the weight of these algorithms. They amplify our product choices and news recommendations; they're embedded in our financial markets. While we may not have direct experience building algorithms or for that matter understand their reach—just as the Babylonians never saw the Company—we believe them to be all-encompassing.

Algorithms as rules for computation are nothing new. What is new is the sudden cognizance of their scope. At least three things can be said of the Babylonian lottery, and of our own:

First, the Babylonian lottery increases in complexity and reach over time. Similarly, our algorithms have evolved from deterministic to probabilistic, broadening in scope and incorporating randomness and noisy social signals. A probabilistic computation feels somehow mightier than a deterministic one; we can know it in expectation but not exactly.

Second, while in the beginning all Babylonians understand how the lottery works, over time fewer and fewer do. Similarly, today's algorithms are increasingly specialized. Few can both understand a computational system from first principles and make a meaningful contribution at its bleeding edge.

Third, the Babylonians for some time brushed under the rug the encroachment of the lottery. Because an “intensification of chance” conflicts with our mythologies of self-made meritocracy, we too ignore the impact of algorithms for as long as possible.

So here we are, with algorithms encroaching, few who understand them, and finally waking up. How do we avoid the fate of the Babylonians?

Unfortunately, those of us in the centers of algorithm-creation are barking up the wrong tree. That algorithms are not neutral, but in many cases codify bias or chance, isn’t news to anyone who’s worked with them. But as this codification becomes common knowledge, we look for a culprit.

Some point to the "lack of empathy" of algorithms and their creators. The solution, they suggest, is a set of more empathetic algorithms to subjugate the dispassionate ones. To combat digital distraction, they'd throttle email on Sundays and build apps for meditation. Instead of recommender systems that reveal what you most want to hear, they'd inject a set of countervailing views.

The irony is that these manufactured gestures only intensify the hold of a Babylonian lottery.

We can no more undo the complexity of such lotteries as we can step back in time. We've swept our lottery's impact under the rug of our mythologies for a good while. Its complexity means no one in a position to alter it also understands its workings in totality. And we've seen the futility of building new algorithms to subvert the old.

So what do we do? How could the Babylonians have short-circuited the lottery?

For Borges' Babylonian narrator there are only two ways out. The first is physical departure. We encounter the narrator, in shackles, abroad a ship bound out of town. Whether his escape is a sign of the lottery's shortcoming or a testament to its latitude remains ambiguous.

The second, less equivocal way out of Babylon is found, as we see, in storytelling. By recounting the lottery's evolution, the narrator replaces enumeration with description. A story, like a code of ethics, is unlike any algorithm. Algorithms are rules for determining outcomes. Stories are guides to decision-making along the way.

Telling the tale ensures that the next instantiation of the lottery isn’t merely a newly parameterized version of the old. A story teaches us to make new mistakes rather than recursively repeating the old. It reminds us that the reach of algorithms is perhaps more limited than we fear. By beginning with rather than arriving at meaning, a story can overcome the determinism of chance.

Babylon as a physical place, of course, fell apart. Babylon as a story endures.