The Honest State

What Happens When Everyone Shows Up

#

The food stamp program serves about 82% of eligible Americans. Not because 18% don’t want help. Because 18% can’t survive the process of getting it.

Medicaid participation varies wildly by state. Some hit 90%. Others hover around 70%. Same program, same eligibility rules, different administrative friction.

The Earned Income Tax Credit, one of the most effective anti-poverty programs ever designed, goes unclaimed by roughly 20% of eligible families every year. That’s billions of dollars sitting in federal accounts that were budgeted for people who never received them.

This is not a bug. It’s the budget.

Programs are funded assuming friction will suppress enrollment. The 18% who don’t show up for SNAP are not an accident. They are a fiscal assumption. The billions in unclaimed EITC are not lost. They were never expected to be claimed.

This is the safety net’s dirty secret: it is priced for partial delivery.

The AI Problem

#

Now imagine AI that actually works. AI that navigates benefit applications for people. That identifies eligibility across programs. That fills out forms, gathers documents, meets deadlines. That does for free what navigators and attorneys and social workers do for money or as charity.

Everyone eligible shows up.

The 82% becomes 95%. The 70% becomes 90%. The unclaimed EITC gets claimed.

This breaks the budget. Not because the programs are too generous. Because the programs were never funded for full enrollment. The friction was load-bearing.

Three Honest Futures

#

When AI removes friction, society faces a choice it has been avoiding. Three options, none comfortable.

The Honest Expansion. Government acknowledges the gap, funds programs for full enrollment, accepts the tax implications. AI becomes a tool of genuine expanded safety net. This requires political will to raise revenue to match actual promises.

The Honest Contraction. Government explicitly tightens eligibility to match actual budget capacity. Fewer people qualify, but everyone who qualifies actually receives benefits. AI helps everyone who’s eligible, but eligibility is narrower. At least it’s transparent rationing rather than bureaucratic rationing.

The Uncomfortable Middle. AI deployed selectively. Some populations get navigators, others don’t. Geographic variation. New inequities replacing old ones. Most politically likely. Least honest.

Each is a legitimate policy position. What is not legitimate is the current arrangement: promising benefits you’ve budgeted not to deliver, then blaming the people who can’t survive the process.

The Household Metaphor

#

Families don’t run eligibility determinations for dinner.

A family with four people and a tight budget doesn’t make each person apply for food. Doesn’t require documentation of hunger. Doesn’t deny the third child because the form was submitted late.

A family looks at what it has and distributes across who needs it. Tight month, pasta. Good month, steak. No shame, no theater. Everyone knows the constraints. Everyone sees the math.

Government could work this way.

Not unlimited benefits. Not equal benefits. But honest benefits. Variable allocation based on real resources, real needs, real constraints. Explained clearly. Adjusted transparently.

The Optimization Contract

#

Replace decisions with math. Replace complexity with transparency. Replace hidden rationing with honest variability.

The current system makes binary decisions. Eligible or not. Approved or denied. Above the line or below it.

Binary decisions are expensive. They require determination infrastructure. Appeals processes. Error correction systems. Each decision point costs money that never reaches the person the program was designed to serve.

Optimization is different. Instead of asking “does this person qualify?” you ask “given available resources and everyone who needs them, what is the optimal allocation?”

This is not rationing. Rationing hides behind complexity. This is allocation. Allocation explains itself.

The contract requires four elements:

The math is visible. You can see why your allocation is what it is.

The constraints are real. People can verify the budget is the budget.

Appeals are meaningful. “My situation is different because X” actually gets heard.

Variability is explained, not hidden. Your benefit adjusted because here’s why. Not “denied due to form 27B missing.”

The Dignity Difference

#

Variability isn’t cruelty when it’s explained.

“Your food assistance is $180 this month because your income increased and there are 47,000 households in the county sharing a fixed allocation” is manageable. You may not like it. But you understand it. You can see whether it’s fair. You can argue if it isn’t.

“Denied due to failure to submit recertification documents within the 30-day window” is violence. You didn’t fail. The system failed to be navigable. The denial isn’t a decision. It’s friction pretending to be policy.

Honest variability preserves dignity. Hidden denial destroys it.

The Cost Inversion

#

Here is the counterintuitive possibility: honesty might be cheaper.

The administrative infrastructure of complexity is expensive. Eligibility workers. Appeals judges. IT systems that manage determination logic. Error correction processes. Fraud detection built on suspicion rather than verification.

If you replace binary eligibility with portfolio optimization, you eliminate most of this infrastructure. You don’t need to determine if someone qualifies. You need to determine how much they need relative to everyone else and what’s available.

The administrative savings don’t cover the full enrollment gap. But they narrow it. And they redirect money from process to people.

The Trust Problem

#

None of this works without trust.

Can citizens believe the optimization is fair? Can they verify the math? Will the AI actually advocate for them, or will it become another tool of gatekeeping dressed in friendlier clothes?

The AI has to be your advocate, not the system’s enforcer. It fights for your optimal allocation within honest constraints. It explains in your language. It catches errors that hurt you. It remembers your story so you don’t have to keep proving your existence.

The constraints have to be real and visible. People need to see that the budget is the budget, that the optimization isn’t rigged, that their share reflects genuine math rather than hidden policy choices.

And appeals have to matter. “My situation is different because X” has to actually get heard by something capable of understanding nuance.

Transparency is the price of legitimacy. Without it, optimization becomes just another word for algorithmic denial.

The Forcing Function

#

The administrative state’s complexity was never really inefficiency. It was policy. A way to make promises without fully funding them. A way to ration without rationing.

AI forces us to choose a different policy.

When full enrollment becomes possible, society has to decide what it actually wants to guarantee. Not what it wants to pretend to guarantee while relying on friction to manage the gap.

This is either terrifying or liberating. Terrifying if you think democratic deliberation will produce cruelty when forced to be explicit. Liberating if you think honesty about constraints is better than bureaucratic theater.

Either way, the choice is coming. The technology removes the comfortable fiction. We can keep pretending, or we can optimize.

Households don’t run eligibility determinations. They stretch what they have across who needs it, acknowledge when things are tight, and explain the constraints to everyone at the table.

Government could do the same. The question is whether we’re ready to stop lying about what we can afford to promise.

This is Part 46 of The Approximate Mind, a series examining how AI might serve human flourishing rather than human extraction. Part 44 explored the paperwork burden. Part 45 examined whether rights without capacity are meaningful. This article asks what happens when AI removes the friction that safety net budgets depend on, and whether honest allocation can replace hidden rationing.