Two Political Short Stories
(a fictional 2017, as imagined in November 2016)
I cough nervously to break the awkward silence as we wait for the Chinese ICBM to kill us. "Don't blame me," I say, "I voted for Gary Johnson!"
Glares all around.
"Aaaand I live in California, and I'm not eligible for the vote-trading hack because I'm not a Clinton supporter," I clarify.
My insufficiently-requited love continues to glare, contempt gleaming in her eyes. "People have been explaining the idea by talking about Clinton supporters in safe states, but the case for vote-trading doesn't depend on that," she says. "As long as you care more about defeating Trump than supporting Johnson, you should still buy a Clinton vote in a swing state in exchange for your California vote; it doesn't matter what you would have done with your California vote otherwise."
"I don't think that works," I say. "The profitability of a deal to each party—uh, no pun intended—has to be calculated relative to the opportunity cost of not making the deal; my counterparty in a proposed trade should be thought of as buying, not a California candidate-of-their-choice vote, but the difference between a California Johnson vote in the no-deal possible world and a California candidate-of-their-choice vote in the possible world with the deal."
"I agree that agents need to consider counterfactual worlds in order to make decisions, but the counterfactuals are properties of the agents' decision algorithms; you can't treat them like they already exist. Think of it this way: if you wish you could have been a Clinton supporter in the absence of vote-trading, in order so that you could take advantage of vote-trading given that vote-trading exists, you can just ... make the corresponding decisions. All you have to offer is your vote; your swing-state counterparty isn't trying to reward or punish people based on what decision theory they use internally."
"What, and leave the thousand dollars in the second box?" I joke.
Then the missle lands, and we die in a flash of light.
(a fictional 2027, as imagined in November 2020)
My insufficiently-requited love coughs nervously to break the awkward silence as we wait in a crowded holding cell in the Department of Diversity, Inclusion, and Equity. I continue to glare at her, contempt gleaming in my eyes.
She's about to speak, but gets interrupted by the cell's Alexa. "Shu!" it shrills. "2231 Shu J! Room 101!"
Shu appears to be a young Asian wom—no, I can't see her—their—pronoun badge at this angle. That kind of cisnormative perception, detectable through facial-expression microanalysis, is why I'm here.
Well, that and the blog. On reflection, probably mostly the blog.
"No!" screams Shu. "I know I've benefitted from white privilege, but you can't—" A cold-faced young officer enters, a black man. A decade ago, I wouldn't have made a call on his race—possibly white with a slight tan—but in the current year, I can tell by the black trim on his blue HE/HIM badge. Shu puts up a struggle, but is hopelessly outmatched and easily subdued; men are much stronger than wo—the officer is much stronger than Shu. As they leave, I catch sight of Shu's green THEY/THEM badge.
"So," says my insufficiently-requited love. "What do you suppose is in Room 101?"
I stare at her breasts for a moment before I catch myself and avert my eyes. "I read an effortpost about this on themotte.win.onion," I say, eyes closed, head tilted upwards. "They have a transcranial magnetic stimulation machine. Big electromagnet tweaks your brain to eliminate your implicit bias. Really schway technology, actually: they trained a machine-learning model on MRI scans of people who got perfect implicit association and anti-misgendering scores, so they knew exactly what pulses to send to fix all your biased perceptions with no side effects."
"Oh, that sounds wonderful," she says, as I look back towards her to sneak a peek at her breasts again. "I want to be cured of my implicit bias! But," she moves her head to indicate towards the door where the officer had taken Shu, "why—why do you suppose they were so scared?"
"According to the effortpost ... they knew exactly what pulses to send. Until the interpretability team inspected what the model was really doing. Turns out, the algorithm, the perfect algorithm that achieved the desired effects with no downsides—had learned to give different treatments to a.f.a.b. and a.m.a.b. people. That couldn't be allowed, obviously, so they fixed that, but they didn't manage to replicate the side-effect-freeness of the original model. These days, people go in to Room 101, and they come out with a limp, and slurred speech. Some of them start having nightmares. Some of them forget how to read."
"I ... see."
"Do you remember," I say, "the last time I asked you out on a date? I mean, the last time."
"Strangely, yes," she says. "It was seven years ago. I said—I said that if you really cared for me, you'd do more to prevent Donald Trump from being re-elected ..."
"Saotome-Westlake!" shrills the Alexa. "3578 Saotome-Westlake M! Room 101!"
"I blame you!" I yell at my insufficiently-requited love, as the officer drags me away. "I voted for Jo Jorgensen, and I blame you!"