When I tell people I work for an insurance company, they nod politely and change the subject. This is exactly as it should be. What I actually do is adjust the expected lifespan estimates in our actuarial models when someone does something sufficiently stupid or sufficiently beautiful that our algorithms flag them for manual review.
Today's case: A man named Marcus Chen who, at age 34, has just donated a kidney to a stranger he met on a Reddit forum about ferret rescue.
The kidney itself is easy. Minus 0.7 years, standard single-organ donation adjustment, though we add back 0.3 years because people who donate organs tend to take better care of themselves afterwards—a psychological effect our models picked up after analyzing 50,000 donors. Net: -0.4 years.
But here's where it gets interesting. Marcus met this stranger, Jennifer, because he spent approximately 847 hours over the past two years helping coordinate ferret rescues across the Pacific Northwest. Our behavioral models have him flagged as "pathologically altruistic"—a technical term we use for people who reliably choose others' welfare over their own in ways that should, by any reasonable model of rational self-interest, not exist.
I pull up Marcus's full file. Standard middle-class upbringing. Degree in electrical engineering from a state school. Works as a software developer at a company that makes scheduling software for dentists—the kind of monumentally boring job that makes my own seem glamorous by comparison. Never married. No kids. Hobbies: ferret rescue, apparently, and playing the cello badly in a community orchestra.
Here's what I'm supposed to do: I'm supposed to look at the kidney donation, apply the standard adjustment, and move on. Maybe add a small risk premium because people who do one crazy altruistic thing often do more.
Here's what I actually do: I stare at Marcus's photo—he has the kind face of someone who apologizes when you bump into them—and I think about the fact that somewhere in Seattle, a woman named Jennifer is not dying of kidney disease because Marcus decided that a stranger's life was worth a piece of his own body.
And I think: what the fuck are we even doing here?
My manager, Robert, has worked in insurance for 31 years. He has the specific kind of deadness in his eyes that comes from spending three decades reducing human experience to probability distributions. I once saw him eat a sandwich during a meeting where we were discussing how to price policies for people with terminal diagnoses. The sandwich was egg salad. This detail has haunted me.
"Something wrong with the Chen file?" Robert asks, appearing behind my desk with the supernatural silence of middle management.
"No, just... this guy gave a kidney to someone he barely knows."
Robert leans over, squints at my screen. "Pathological altruist. Tag it with the standard risk premium. These types usually do something else stupid within five years."
"Stupid," I repeat.
"Financially stupid," Robert clarifies, missing my point entirely or perhaps understanding it perfectly and choosing violence. "These people donate more money, volunteer more hours, take jobs that pay less because they're 'meaningful.' Higher divorce rates too—turns out most people don't actually want to be married to a saint. It's all in the models."
He's right, of course. The models are always right. That's the problem.
"Apply the premium. Move on. You've got forty-seven other cases to review today."
I became an actuary because I liked math and wanted job security and fundamentally misunderstood what it would feel like to spend my life putting price tags on human existence.
In school, it was all elegant probability theory and stochastic processes. The professor would put up a problem: "Calculate the expected present value of a life insurance policy for a 40-year-old non-smoking male." And you'd work through the integrals and feel the satisfaction of an answer that was correct, provably correct, mathematically correct.
They never mentioned that every one of those 40-year-old non-smoking males was someone's father, someone's husband, someone's mediocre cello-playing ferret-rescuing miracle of consciousness who happened to contain multitudes.
Or maybe they did mention it and I wasn't listening because I was too busy thinking about the elegant mathematics of mortality.
I apply Robert's recommended adjustment to Marcus Chen. Risk premium: +$127 annually on his life insurance, +$83 annually on his health insurance. Expected value calculations suggest we'll make an extra $8,200 over his remaining lifetime because people like him reliably do expensive, beautiful things.
The system accepts my inputs. Marcus's file closes. Somewhere in Seattle, he's probably recovering from surgery, maybe texting Jennifer to make sure she's okay, maybe playing the cello badly, maybe existing in that narrow space where life is hard and painful and utterly, unreasonably worthwhile.
I open the next file. Sarah Mitchell, age 29, flagged because she quit her consulting job to become a park ranger at a national park, taking a 70% pay cut. The models are concerned. The models want to know if she's going to do something else similarly irrational.
I look at her photo. She's smiling. Actually smiling, not the corporate headshot smile I see in most files, but the smile of someone who made a decision that makes no sense on paper and perfect sense everywhere else.
I should apply the standard adjustment. I should move on. I have forty-six other cases to review today.
Instead, I do something I've never done before. I open a new document. I title it "Notes on Cases That Made Me Feel Things." This is a catastrophically unprofessional thing to do. These notes will never be read by anyone, will never affect anything, will exist only as a small monument to my own sentimentality and weakness.
I write:
Marcus Chen donated a kidney to a stranger. The models say this makes him risky. The models say people like him cost us money. The models are probably right. But somewhere tonight, Jennifer is alive because Marcus decided that math shouldn't be the only thing that matters.
Sarah Mitchell chose beauty over money. The models are worried. I am worried about the models.
That night, I go home to my apartment—a place I chose because it was a good investment in an up-and-coming neighborhood, which is a fancy way of saying I participated in the economic displacement of people who couldn't afford the new rents. I heat up leftovers. I do not play the cello badly or rescue ferrets or do anything that might cause someone in some distant office to flag me as "pathologically altruistic."
I think about Marcus Chen, about how the models predict he'll probably donate more money than he should, volunteer more hours than is reasonable, maybe even donate his other kidney if someone needs it bad enough, though we don't have enough data on double donors to model that properly.
I think about how I spent today making Marcus's life measurably more expensive because he chose to be good.
I think about the fact that I make $127,000 per year—$127,000!—and I chose that number very carefully because it's what the models say an actuary with my experience should make, and I would feel foolish making less, would feel like I was failing some crucial optimization problem.
I think about Robert's egg salad sandwich.
I think about the old sky, though that's not quite right because the sky hasn't changed, I have.
The next morning, I submit my resignation. No plan, no other job lined up, nothing sensible or rational or optimized. Just a profound exhaustion with being the kind of person who makes the world slightly worse in small, mathematically rigorous ways.
Robert calls me into his office. "You're having a quarter-life crisis," he says. "It's common in the actuarial field. Take a vacation. Go to Thailand. Eat some street food. Get diarrhea. Come back refreshed."
"I don't think this is that," I say.
"You know what you remind me of?" Robert says, and I can already feel where this is going. "You remind me of this case I reviewed seven years ago. Guy quit his job in finance to become a teacher. Took a sixty percent pay cut. We applied the standard risk premium, of course. Know what happened?"
"He lived a rich and meaningful life teaching children and died with no regrets?"
"He got divorced within three years, declared bankruptcy within five, went back to finance within seven. The models were right. The models are always right."
"Maybe they're right about the wrong things," I say, and immediately feel like an idiot because that doesn't even make sense grammatically.
Robert sighs the sigh of a man who has seen many junior actuaries have feelings and knows this too shall pass. "Your resignation is accepted. HR will contact you about exit procedures. I hope you find whatever you're looking for."
I won't, probably. The models say people like me—people who make impulsive decisions based on emotion rather than optimization—tend to regret them within 2.7 years on average.
But here's the thing the models can't quite capture: regret isn't the worst thing. The worst thing is being the kind of person who never risks it.
I'm writing this from a coffee shop, unemployed for the first time since college, feeling that specific kind of terror that comes from realizing you just made your life much harder for reasons you can't quite articulate to your concerned parents who keep asking if you're "okay" in that tone that really means "have you lost your mind?"
But I keep thinking about Marcus Chen, about how the models predicted he'd do more stupid, beautiful things, and how that prediction was, in fact, the sanest thing about him.
I keep thinking about Sarah Mitchell, smiling in her park ranger photo, and how I hope she's happy, truly happy, even though our models give it only a 34% probability.
I keep thinking about all the cases I reviewed over the years, all the people I made slightly more expensive because they chose meaning over optimization, beauty over efficiency, love over actuarial tables.
And I think: maybe it's okay to be a bad bet sometimes. Maybe the point isn't to be predictable. Maybe the models are sophisticated enough to price human goodness but not sophisticated enough to understand why it matters.
I don't know what I'm going to do next. Maybe I'll be back in finance within seven years, divorced and bankrupt and exactly as predicted. Maybe I'll learn to play the cello badly. Maybe I'll rescue some ferrets.
Or maybe I'll just try to be the kind of person who, when reviewed by some future actuary, makes them stop and think and feel that small, dangerous thought: what the fuck are we even doing here?
If you're reading this, whoever you are, I hope you do something stupid and beautiful today. I hope you cost someone's model a little bit of accuracy. I hope you're exactly the kind of risk that no algorithm can properly price.
The models say this is all very foolish.
The models are probably right.
Fuck the models anyway.