Deprogramming the Apocalypse

Then everyone was plural: data. Everything
served and being served on metal servers.
It was never our intent to punctuate the sentences
of others. But now it’s late. Too late to unstate our
​importance. And besides the crickets died for this.
—Zoë Hitzig, "How We Programmed The Apocalypse" 

In the early 2000s, Boston Public Schools had a problem. 

Economists in the fields of game theory and mechanism design pointed out that the algorithm the city used to assign students to its public schools allowed families to game the system and improve their chances of getting into one of their top-choice schools. Moreover, contrary to the city’s goal of promoting equity, the families who were most likely to employ these strategies—and get assigned to their preferred schools—tended to be more affluent.

The economists proposed a fix to the algorithm that would make it—and the process of school assignment—“strategy-proof.” In 2005 the Boston Public Schools (BPS) implemented the algorithm.

It worked. Families were no longer able to improve their chances of getting into a top-choice school. The new algorithm was praised as an example of the way that economic models and mechanisms could solve real-world problems.

The economic theorist and poet Zoë Hitzig sees the BPS story differently. A PhD student in economics whose work incorporates moral philosophy and theories of justice, Hitzig says that when institutions like the BPS choose one algorithm over another, they aren’t merely making a technical fix like a civil engineer improving a bridge; they’re making a normative judgment, choosing one set of values over another. The tradeoffs involved in that choice are rarely articulated or submitted to a public process. The designers and implementers are usually not accountable to the citizens impacted by their decisions. Hitzig sees this situation as an example of “algorithmic life.” In her academic work and her poetry, she lays bare its costs and offers new ways of making decisions and distributing resources in a democratic society.

School assignment in Boston in 2005 is a stark example of how our lives are shaped by algorithms.

Bringing Democracy to Algorithmic Life

Image

Zoë Hitzig is a PhD student in economics at the Graduate School of Arts and Sciences (GSAS), Harvard University.

Hitzig studied the case of student assignment in the Boston Public School system to grapple with the gap between ideals of equality and the mechanisms designed to achieve that end. The result was “The Normative Gap: Mechanism Design and Ideal Theories of Justice,” a 2020 paper published in the journal Economics & Philosophy. She explains that parents circumvented the intent of the BPS assignment system in the early 2000s by misrepresenting their top-choice school.

“Let’s say I had a kid and our top choice school was highly sought,” Hitzig says. “Lots of students want to go there. But I also like another less popular school. So, I list that school as my first choice even though it’s not because it increases the chance that my child will get in. The cost is that they don’t have a shot at getting into our true top choice school but that was already unlikely.”

The strategizing conflicted with the BPS’s goals of equitable assignment. At the time, researchers found that families in wealthy neighborhoods were more likely to connect and strategize about school choice than were less-affluent families. The economists Atila Abdulkadiroglu, Tayfun Sönmez, Al Roth, and Parag Pathak proposed making the algorithm strategy-proof through “deferred acceptance.”

School assignment in Boston in 2005 is a stark example of how our lives are shaped by algorithms.
-Zoë Hitzig

“With deferred acceptance, schools still fill their seats with those who list them as their top choice,” Hitzig says. “But the assignment system doesn’t end immediately: A student who lists a school as their second choice, for instance, but has higher priority at that school, perhaps because they live within walking distance, could bump another who’d been tentatively accepted after listing that school as their top choice. The bottom line is that you don’t improve your chances of getting assigned to your second-choice school by misrepresenting your top choice.”

The modified algorithm was hailed as a success when the BPS implemented it in 2005, but Hitzig says that the new system had its own flaws. Deferral was a less efficient way to assign students than immediate acceptance. Having made the system strategy-proof, there was evidence that not all families understood the change or that it was in their best interests to write down their true preferences, even after they were instructed to. Finally, given that top schools often sat in wealthy neighborhoods, the system might still perpetuate other forms of inequality by giving priority to students who lived within a school’s walk zone. Indeed, citing equity and transparency concerns, BPS eliminated walk zones entirely in a later reform of the assignment mechanism.

Hitzig says her intent in pointing out these flaws is not to condemn the BPS;  it’s to demonstrate that its decision to implement the new algorithm was a normative choice—one with trade-offs—as much as a technical one. “Economics can come up with a formal language that becomes a technical property of the algorithm that leaders and policymakers choose,” she says. “You can say, ‘Look, this algorithm has this particular property. If you think that it represents the kind of equity that you’re after, then you should use this algorithm instead of another.”

Moreover, Hitzig argues that the values embedded in an algorithm reflect those of the people who design and implement it. More often than not, she says, those people are neither elected nor directly accountable to the citizens their decisions affect.

“School assignment in Boston in 2005 is a stark example of how our lives are shaped by algorithms,” she says. “The people who create and execute them have low visibility and transparency. Their interests are not necessarily aligned with those of the people in the community. Their notions of equity are often limited. It’s a step away from the kind of decentralized process that gives citizens a voice in the way decisions are made.”

A Better Way to Finance Public Goods

Giving people more of a voice in the way funding decisions are made is the focus of Hitzig’s recent work. With E. Glen Weyl, an economist at Microsoft, and Vitalik Buterin, co-founder of the cryptocurrency/blockchain ecosystem Ethereum, Hitzig created the “quadratic finance” (QF) mechanism for funding public goods, defined by the team as “any activity with increasing returns in the sense that the socially efficient price to charge for the activity is significantly below the average cost of creating the good.” Think of your local public broadcaster. The cost of producing accurate news and information is significant, but the “socially efficient price” for that good in a democracy is “free.” Hence the fundraising drives where contributions are frequently matched dollar-for-dollar by funds raised from a small number of large donors.

The problem with this model, Hitzig, Weyl, and Buterin argue in a paper for the journal Management Science, is that the public broadcaster has to compete with a thousand other causes—from the local homeless shelter to the ballet. The ones that raise the most money—and attract the greatest matching funds—are not necessarily the ones with the greatest popular support or alignment with the common good. Often, as the authors write, the causes represent “status motivations and parochial, even exclusionary interests.”

With QF, though, projects with the greatest popular support would also get proportionally greater financial support. In this system, philanthropists might raise a large fund and invite the public to contribute to any one of several causes. A project that attracts many small donations—your local homeless shelter, for instance—might receive in matching funds multiples of each dollar raised. A project that attracts a few large donations, on the other hand, would still get matching funds, but perhaps less than a dollar-for-dollar match.

“QF speaks to the democratization of governance systems,” Hitzig says. “And these ideas have been influential already even though they are young. The system we proposed has been used to democratically allocate millions of dollars to open-source software projects in the blockchain space and was also used to allocate emergency stimulus funds during the pandemic in the city of Boulder, Colorado. It’s a concrete example of a new way of organizing ourselves.”

One of Hitzig’s advisors, Jerry R. Green, David A. Wells Professor of Political Economy and John Leverett Professor in the University, says that by integrating ethical and moral concerns with classical economic models, Hitzig is “expanding the scope of economics.”

Her work dovetails completely with current trends in economic theory and she is perfectly positioned to push these developments to the next level.
-Jerry Green

“Zoë and her co-authors explore the extension of market design to include concerns such as the fairness of outcomes, the preservation of privacy, and the potential for collusion,” he says. “They explore motivations and constraints that should be avoided by design mechanisms, some between sets of winning bidders and some between the seller of the goods and losing bidders. Her work dovetails completely with current trends in economic theory and she is perfectly positioned to push these developments to the next level.”

A Shared Predicament

As her academic career has developed, so has Hitzig’s poetry, which has appeared in The New Yorker, London Review of Books, Paris Review, Lana Turner, and Harper’s, among other publications. In 2020, she released her first book of poetry, Mezzanine, to critical acclaim. She says that her creative writing and her work in economics “share a predicament”: how to capture and value the aspects of life that cannot be quantified.

Hitzig is drawn to the beautiful, the terrifying, and the mysterious—as well as to the costs of imposing a narrowly defined type of rationality on all aspects of human existence. That fascination is what led her to include in poetic form an epigraph in her book, an excerpt from the UC Berkeley Professor Emerita Carolyn Merchant’s seminal 1980 work, The Death of Nature which speaks of “autonomous” and “non-autonomous” (human) machines fundamental to a “new value system” based on power and order. “I’m very glad we have calculus,” Hitzig says. “I’m very glad we have understandings of the human body that allow us to live long healthy lives. But Merchant’s book reminds me that the Enlightenment came at a cost. I made that excerpt the epigraph of Mezzanine because I try to capture similar ideas in my poems.”

Poetry is also another way that Hitzig interrogates the influence of economic and technical systems on our lives. Although she did not write it with economic concepts in mind, “How We Programmed the Apocalypse” echoes the “performativity thesis” that Hitzig invoked in her work on school choice algorithms: the notion that economic narratives, when adopted and repeated by society, become self-fulfilling prophecies, engendering the types of behavior that they claim to describe.

We can't continue consuming and spending on things that just end up in landfills and burn fossil fuels.
-Zoë Hitzig

“You can hear in the title of that poem the idea of performance,” she says. “If you’re in a system that tells you long enough that this is how you’re supposed to be, at some point that narrative becomes performed. Even if people weren’t acting like ‘rational economic agents’ to begin with, if that’s how to get ahead, people will start behaving that way.”

For Hitzig, the results of this transformation into homo economicus are consumer capitalism, indifference to human suffering, and climatological catastrophe, themes she explores in the poem “The War Gone Wrong Room”:

More for your dollar? More for the
Dollar. More exotic juice in the
juice boxes to trade with the bullies.
More juice to burn into smoke to soot up
the ceiling fan. More keypads to catch
our twiddling thumbs. More tangling
bodies in nets.More nets filling with
corpses. More decals for ships and
chests. More juice to bargain for, then
burn. More fire blankets on our lungs.

“We can’t continue consuming and spending on things that just end up in landfills and burn fossil fuels,” she says. “I can’t rewrite the narrative about what’s happening in the atmosphere, but I can play a small part in rewriting the narrative of how we can relate to each other and the planet.”

As she nears graduation, Hitzig hopes to find a teaching or postdoctoral position that will enable her to continue her theoretical work in economics. Meanwhile, her second book of poems is nearly complete. In the long run, she hopes to help leaders envision new models, values, philosophies, and narratives and integrate them into a new social system that works for more people.

“We’ve lived for many years under the narrative of neoliberalism,” she says. “Anything’s possible if you work hard. Individuals who are free to act in their own self-interest will always produce the best outcomes for society. Social insurance can fill in the gaps. But so many of those pieces are crumbling. Until we have a new way of thinking about how we live together, it will be very difficult to enact real change. I hope that the work I do can contribute to that new way of thinking.”

Photos by Mark Ostow