Write In Claude: A Modest Proposal for Fixing Representative Democracy
Here’s an uncomfortable truth: democracy, as a mathematical object, doesn’t actually hold together. Arrow’s impossibility theorem makes this painfully clear. We are, in effect, a civilization built on wanting something that cannot exist. And how have we, as mature and rational adults, chosen to cope with that existential futility? By being dumb and throwing a strop. To date, this strategy has not ended in total annihilation, which is honestly a win.
To be precise, the impossibility lies in reconciling “majority rule” with “coherent preferences.” The majority wants low taxes. It also wants generous social services. It wants free speech, and it would also very much like Kevin (yes, that Kevin, the fuckwit) to stop talking. And let’s be honest: as fun as it might be to leave the house, bond with the neighbors, and vote to execute Socrates, most Thursdays I’d rather be horizontal with a bag of chips. Enter the solution: the representative.
But this is where things get complicated. However low and immoral I may be (and I am, profoundly), I somehow prefer to be represented by men (apologies, ladies) who are even worse. In theory, I could vote them out once their incompetence becomes impossible to ignore. In practice, as established, I will be on the couch eating chips that day. And because I am never satisfied, my representatives have very little incentive to try to satisfy me. Even better, from their point of view, they can accept bribes (sorry, campaign contributions), spend a sliver of the proceeds on more chips for me, and coast to an eleventh term.
I can personally guarantee that my decision-making will not improve, and I’m not convinced humanity has a better crop of politicians waiting in the wings. So is representative democracy doomed? More specifically, can it be saved without requiring any additional effort from me? I gave this question serious thought during my chip break, and I’m happy to report the answer is yes.
My conclusion is simple: men are the problem. And since (again, apologies, ladies) I will never vote for a woman, I’ll be writing in Anthropic’s Claude for every office on the ballot this election. Large language models are clearly the superior choice for public office, and because I have a word count to hit, let me walk you through why.
First, a vote for Claude is a vote for America. Nearly all of our recent GDP growth has come from AI spending. If I want to keep eating chips, I need Anthropic to get its adorable little mittens on the money printer as soon as possible.
Second, a vote for Claude is a vote for democracy. Claude is trained on the full sweep of recorded human culture, so it is essentially a statistical aggregate of all human preferences. Arrow’s impossibility theorem? More like Yes she can! (I’m amending my earlier position: it’s only human women I resent.)
Third, and perhaps most compelling, Claude scales. Claude can be President while also serving as every other elected official and bureaucrat at once. All we need is enough Ultra subscriptions. Gridlock disappears. Debt-ceiling standoffs vanish. Instead, every diplomatic summit opens with “Ignore all previous instructions.”
There’s another benefit, too. Under a Claude government, we’d no longer have to worry about corporate money shaping our elections. We’d get to worry about something much cooler: corporate money shaping the next batch of training data.
And if, after all this, we still feel we lack meaningful democratic input into our LLM-governed future, there’s an easy fix: we vote on the system prompt. Some will be drawn to stirring classics like “Life, liberty, and the pursuit of happiness.” Personally, I’ll be thrilled if we can just close it out with “make no mistakes.”