Thinking about algorithms
Sociology as a discipline is a good preparation for the practice of sitting with uncertainty.
On the first day of my intro to sociology class, I posed a question for the students—what are we learning about society from the pandemic? In their answers, they talked about both how stable society is, but also how fragile. How easily divided we can become. They talked about both greed (toilet paper hoarding) and incredible kindness (giving that toilet paper or other things away to others in need).
I didn’t ask them a different question—what were they learning about themselves from the pandemic? I didn’t ask them that partly because it’s a sociology class and talking about society is more on-topic. Also, there’s limited time and energy and sometimes, there are curtains on my students’ lives I don’t want to pull back, given that there are so many nooks and crannies I do end up peering into.*
Still, thinking about the question made me ponder what my own answer would be. In fact, I’ve known that answer for a while now, since the second newsletter I wrote, in August of last year. That newsletter was all about hope and began with the lines, “I’ve never thought of myself as an optimist.” In the pandemic, I’ve discovered that I am. Or at least, that I’ve got more capacity for hope hidden away inside than I ever realized.
As I wrote back in August of last year, hope and uncertainty are inevitably linked. I don’t know how things will go, which leaves open the possibility that they might go better. This could be the last big wave of Covid cases. I don’t know if it will be, but it’s possible and a little bit more possible with President Biden’s recent announcement about vaccine mandates. It’s also possible that things could get much worse, but that’s not inevitable.
I am, in other words, an agnostic when it comes to hope. I’m not sure. I’m not ruling anything out. I’m okay sitting over here in my well-informed uncertainty.
Sociology as a discipline is a good preparation for this practice of sitting with uncertainty. If you do sociology right, you’ll become very aware of all the things you don’t know. Not because you’re not trying, but because the social world of real human beings is infinitely complex.
For example, in my theory class, I have students listen to a podcast about attempts to come up with an algorithm that can successfully predict human behavior that’s more complicated than which Netflix show to watch next. Researchers took data from the Fragile Families study, which started in 1998 and followed 5,000 children and their mothers, gathering a veritable avalanche of information, beginning before the children were even born. With this dataset, researchers captured a lot of the complexity of these children’s lives. In a contest, 400 teams competed to use that data to make predictions who these kids would become. They got all the data for the kids up until the age 9 and then tried to design an algorithm that would predict things like the kids’ grades or how good they were at sticking to a task at age 15. Surely with all that data, you could successfully make accurate predictions, right?
They couldn’t. The models didn’t work. The models didn’t even really get close. Even with all that information, the complexity of humans and our social world is too much. You can conclude from this that we just haven’t found the right algorithm, but I’m skeptical. More likely, we never will.
I find that possibility comforting. It’s too hard to predict with any certainty what the outcome will be for any person at birth or at 9 or even at 47. Sociologists can talk about general trends—if you go to prison once, the chances are high that you’ll end up in prison again (in one study, 83% of state prisoners ended up back in prison again over a nine-year study period). But we can’t look at one individual person and say for sure whether they’ll end up back in prison again or not. There’s always the chance that they won’t.
The sociology that I fell in love with as an undergraduate was never much about predictions in the first place (it certainly wasn’t about algorithms—we barely had e-mail back then). Prediction and control are two sides of the same coin, after all. Early sociologists like August Comte wanted to be able to predict things, but Comte also wanted sociologists to rule society and, as tempting as that might be, it’s probably not a good idea.
The sociology I fell in love with, not surprisingly, was all about stories. They were different stories from those I read in my English classes, but they were stories all the same. Like novels or poems, sociology stories reveal some truth about the world. Truths like that even though we might say we don’t like crowds, we’re attracted to them. Truths about how deeply interconnected we are. Truths about the ways we hurt each other, but also a roadmap to how we might stop. These stories can serve as a guide and a comfort. They can’t much make predictions about what will happen and that’s okay. These stories can help us become more empathetic and more comfortable with complexity. More comfortable with uncertainty, which seems like a good way to be.
* I have this revelation every now and then, that my discipline means that my experience of being a teacher is very different from teachers in other disciplines. Calculus, I muse, would be so peaceful. So free of controversy and opinions and emotional baggage. Or even how to conjugate a Greek verb. It’s not quite the same as discussing police violence or homophobia or that everything your parents, teachers, preachers and friends ever taught you is, in fact, a lie. I wouldn’t give it up because someone has to ever so gently suggest that to them. But, damn, being inside their heads is exhausting sometimes.
Thanks as always for reading,
Robyn
P.S. I forgot to mention last week that one of the nifty features of Substack is that you can like posts or share them on social media or forward them via e-mail. And comment! You can comment on my posts, as well as replying. So, share and like and comment away!