DEV Community

FEET GREET
FEET GREET

Posted on

I Play Legislator. My AI Plays Consequences.

I Play Legislator. My AI Plays Consequences.

Sunday evening, nothing to do. I told my AI I was bored and asked for something interesting.

Some background: over the weekend I had it map institutions across 11 countries — labor law, wage formation, housing, social insurance, conscription rules. Every provision carries its effective year and a source. Dry stuff. Beautiful to me, sleep-inducing to most people.

So the AI made a proposal that turned the maps into a game. I would be a legislator with one country and one term. It would be the Game Master. I change one institution per turn; it simulates what happens across four horizons — immediately, six months, five years, twenty years — and it has one house rule: every reform I pass must come back with an arbitrage report, at least one concrete way people will game it. No exceptions, because that's not a bug in the game, it's the finding of the game.

Turn one: I handed out $500 a month

My opening move was a universal health account in the United States. $500 per person per month, earmarked for medical spending, funded by printing money. I thought I was being generous.

The GM came back with arithmetic first. 335 million people, $6,000 each: $2 trillion a year. Larger than Social Security. Twice the defense budget. The largest single federal program in the country's history, and I had proposed funding it the way 2021 funded a one-time stimulus — except every year, forever.

Then came the parts I hadn't asked for. Dentists raise prices first, because dental supply is licensed and capped, and subsidies flow into inelastic supply. Young healthy people realize $6,000 covers a 27-year-old's entire marketplace premium, exit employer plans, and the employer risk pools age and sicken until premiums jump 20 percent and the exit becomes a stampede. Somewhere in year three, someone invents lending against future account balances at 39 percent APR.

The verdict, one line: "You traded a certain currency disaster for breaking an 80-year knot — the most expensive one-shot surgery in American history."

The knot is real, by the way. Health insurance tied to the employer means changing jobs risks your family's coverage. It's the thing my country maps flagged as the US's hardest lock.

Turn two: I tried being modest

I cut the number tenfold. $50 a month, same rules.

Everything interesting died. Nobody leaves an employer plan over $600 a year — it covers 11 percent of a young person's premium. Prices barely chase it. Medical debt barely moves. On paper I had passed a boring little transfer program.

The GM's ruling was the first thing from this game I couldn't stop thinking about. A program that small is too harmless for any lobby to spend money killing, which means it never dies, which means it sits there until some future crisis when a president says "temporarily raise it to $500" — and turn one happens anyway. "A detonator and a timer are the same bomb," is roughly how it put it. "You just changed the schedule."

Same instrument, ten times smaller, different species. That became the game's second law: behavior changes when the benefit covers roughly one year of a person's key expense. Below that line you wrote a transfer. Above it you wrote an institution.

Turn three: I got clever, and got caught

Targeted version: $500 a month, but only for people under 35 with no employer. About 24 million Americans, $140 billion a year. Cheap, aimed exactly at the people the knot squeezes hardest.

The arbitrage report on this one was almost admiring. W-2 workers ask their employers to reclassify them as contractors — the employer saves benefits and payroll tax, the worker keeps the $6,000, and the only casualty is the concept of employment itself. The IRS ends up fighting the definitional war over who counts as having an employer, which is the exact war gig platforms have spent a decade dodging. Uber doesn't need to arbitrage anything; the bill quietly does its lobbying for it, because it socializes benefits costs the platforms currently dodge.

And the age cliff. Anyone at 34 years and 11 months refuses promotions, refuses formal contracts, because a real job means a $6,000 pay cut on their birthday. The GM's verdict: I had legislated the age-35 curse — the one Chinese tech workers write essays about, the one I know from home — into American law, by accident, with good intentions.

That turn also produced the fifth law, the one I keep quoting to colleagues: programs whose beneficiaries rotate out have no settled constituency, so they're politically mortal forever — the exact opposite of the little $50 program, which is politically immortal and does nothing.

Five laws, none from a textbook

Arbitrage: people don't obey rules, people arbitrage rules. Critical mass: 10x scale changes the species. The ratchet: programs that work only grow. The knot: frontal assaults fail, oblique ones untie. The threshold: gates distort behavior on both sides, and the first war of any targeted benefit is a definitional war.

I keep coming back to the first law, for a professional reason. I'm a corporate lawyer. When my clients do precisely what those "loophole-finders" in the simulation did — restructure to convert income types, time transactions around thresholds — nobody calls it looting the treasury. We call it tax planning, and it bills at $1,500 an hour. The arbitrage was never a moral failure of the poor. It's the same intelligence, with worse lawyers.

Play it

The full rulebook — 11 country cards, a 20-card event deck, political capital, scoring, and a copy-paste prompt that turns any capable AI into the GM — is here:

Victoria 2026 — full rulebook

The next country on my table is Korea. Seniority pay, the 52-hour cap, retirement gold: the knot there is that the longer you stay, the more leaving costs you. If you run a turn with your own AI, I'd genuinely like to read the arbitrage report it hands you.

Top comments (0)