Can an 11,000-member House fix American health care?

This is a simulation. Claude played every part. The facts in section 1 are real. Everything else is made up: the members, their groups, the bills, the votes, the estimates, the president's choices, and every voice. Made-up events are marked (simulated). This is not a prediction, a poll, or a recommendation, and it takes no side on how health care should work.

Rerun on 29 October 2026, under the same two rules as the other reruns: stakeholders only watch (they do not vote, write bills, take sides, or act; they watch how the system works, take complaints, and report), and plain language. In the first run, some health stakeholders were credited with changes to the bills. In this rerun those changes were made by members, after hearing from simulated hospitals, insurers, states, and others. The bills and voting assumptions are unchanged, and so are the vote counts.

The question. The Iran simulations asked whether a large House could limit a war. This one asks a harder, more open question: can 11,000 members find a majority for anything that makes health care in America cheaper, more secure, or both? Unlike a war, health care has no single lever, and the groups disagree about what the problem even is.

1. The real situation (as of 28 October 2026)

Spending and results

• Total spending. National health spending grew 7.3% in 2025 to about $5.7 trillion, the third year in a row above 7%. Government forecasters project nearly $9 trillion by 2034, and expect the share of people with insurance to fall from 91.7% to 90.5%. (American Hospital Association)
• Compared with other rich countries. The US spends about 18% of its economy on health, roughly $12,649 a person. The next highest, Switzerland, spends about $7,927. US life expectancy, about 79, trails countries such as Japan, Spain, and Switzerland, and the US has fewer family doctors per person than the average rich country. (American Journal of Managed Care, on the Commonwealth Fund's 2026 comparison)
• Job-based coverage. The average family premium for insurance through work reached $28,303 in 2026, and workers' own share rose 9% to $7,471. (KFF, a health research group)
• The uninsured. About 8% of Americans had no insurance in 2025, roughly unchanged, though the number grew by about 800,000. (AP, via WTOP)

Help paying for marketplace plans

• Expired. The extra tax credits that lowered premiums for people buying their own insurance on the Affordable Care Act marketplaces ended on 31 December 2025. What enrollees pay themselves rose about 58% on average for 2026, and enrollment fell by about 2.6 million, to about 19.2 million. Insurers set 2027 prices assuming the credits stay expired, with a typical increase near 15%. Sign-up for 2027 runs from 1 November to 15 December in most states. (Health Insurance Network)
• In Congress. On 8 January the House passed a three-year extension, 230 to 196, after a petition signed by a majority of members forced the vote. (Association of State and Territorial Health Officials) The Congressional Budget Office estimated it would cost $80.6 billion over 2026–2035 and give 3 million more people insurance in 2027 and 4 million more in 2028. (Bloomberg Government)
• The Senate. A bipartisan group worked on a two-year plan with income limits and a $5 minimum monthly payment. It stalled over the president's proposal to put the money into health savings accounts instead, and over Republican demands for stricter limits on abortion coverage, like the long-standing Hyde amendment. Democrats backed the House bill. Most Republicans wanted the credits to stay expired, citing waste and fraud, and the president threatened a veto. (NBC News) As of late September, the Senate had not voted. (Health Insurance Network)
• Improper sign-ups. A market-oriented think tank estimates that 6.47 million marketplace sign-ups in 2025 were improper, based on a model that compares survey data with enrollment counts. (Paragon Health Institute)

Medicaid

• The 2025 budget law. The Congressional Budget Office's final estimate is that it cuts federal spending on Medicaid and the children's health program by about $990 billion over ten years, and leaves 10 million more people uninsured in 2034. Work requirements for adults covered by the Medicaid expansion are the largest single cause, at 5.3 million. (Georgetown University Center for Children and Families)
• Work requirements. Adults aged 19 to 64 in covered groups must show 80 hours a month of work, school, or service. States must start by 1 January 2027; Nebraska began in May. (Washington Times)

Prices, drugs, and Medicare

• The administration's plan. In January the president outlined a plan to send help directly to people through health savings accounts, make his "most favored nation" drug-price deals permanent, go after pharmacy middlemen, and require insurers to disclose prices and how often they deny claims. It gave no dollar amounts or timeline. (Al Jazeera)
• Pharmacy middlemen. Changes to the rules for pharmacy benefit managers, the companies that run drug coverage for insurers, became law on 3 February in the 2026 funding law, including limits on their fees in Medicare's drug program from 2028. (Pharmacy Times)
• Drug price negotiation. Medicare's second round of negotiated prices, for 15 drugs, takes effect in January 2027. Medicare estimates 44% savings on those drugs. (KFF, a health research group)
• Medicare's hospital fund. Medicare's trustees project its hospital insurance fund will run out in 2033, after which it could pay 89% of promised benefits. (Becker's Hospital Review)

Voters

• In April, 64% of adults said they worried about affording health care, and 55% said it would strongly affect their vote in the midterms. Half of independents trusted neither party on costs. (Fierce Healthcare, on KFF's poll) Midterm elections are on 3 November. Health questions on state ballots include South Dakota's proposal to end its Medicaid expansion if federal funding drops below 90%. (Becker's Hospital Review)

What a House can do. It can pass laws on help with premiums, Medicaid, Medicare, prices, and insurance rules. It cannot change premiums insurers have already set for next year. Every law needs the Senate, including 60 votes to end debate on most bills, and the president's signature or a two-thirds override. States run Medicaid and regulate insurance, so many changes land unevenly.

2. Who took part

The simulated House: the same 11,000 members, turnout model, and rules as Simulation 3, including ranked votes. To pass, a bill needs 5,501 yes votes. At this point the rules said that if the winner of a ranked vote then fell short against "no change," every bill in the ranking failed. The members fell into new groups, defined by views on health care, not by party.

The group sizes and views are assumptions, published in advance. 694 members miss each vote.

A caveat. Real health votes follow party lines more closely than these groups do. Health care also overlaps other issues, such as abortion, immigration, and the balance between federal and state power. This run tests how the system works, not what Congress would do.

Other simulated roles. These act in the simulation; none is a stakeholder:

• Outside groups that talk to members: rural hospital finance officers, a nurses' union, insurance company pricing experts, a drug-industry association, employers who provide coverage, state Medicaid directors, family doctors, and patient groups.
• People affected, posting as advisers with their backgrounds disclosed: a self-employed electrician whose premium doubled when the extra credits ended; a home health aide on Medicaid who expects work-reporting paperwork; a resident of a rural town whose hospital closed its maternity unit; a retired teacher on Medicare; and a parent of a child with a long-term illness, covered through work.
• Also: the budget office, fact-checkers, a video creator, translators, a legal reviewer, constituents, attackers, the Senate, the president, and a forecaster.

New stakeholders watching. Eight joined the 35 from earlier runs. Each watched one thing:

Of the earlier stakeholders, the military, intelligence, diplomatic, and law-enforcement observers had little to watch. No bill touched military or veterans' health care.

3. Days 1–3: Six bills

Six bills were posted:

The Fiscal group wrote no bill. It posted a group position: it would oppose any bill that adds to deficits over ten years.

The budget office's estimates (simulated; built on the real figures above; the assumptions are the office's own):

• H1: federal spending rises by trillions of dollars a year once fully in place, largely replacing premiums and out-of-pocket costs. Whether total national spending rises or falls depends mostly on what hospitals and doctors are paid; the office published a range that runs from savings to added costs.
• H2: several hundred billion dollars over ten years, mostly the permanent credits.
• H3: $80.6 billion for the credits, the Congressional Budget Office's real figure, plus the office's estimate of $9–14 billion for delaying work requirements.
• H4: about even as written, because deposits replace credits dollar for dollar. The office could not estimate effects on coverage without knowing who would choose a deposit.
• H5: saves federal money by design, more each year as costs outgrow inflation.
• H6: saves federal money, but mostly after 2028. The office's year-by-year table showed almost nothing in 2027.

A second estimate. The simulated drug-industry association published its own estimate of H6's drug section, with its funding disclosed. It projected fewer new drugs if negotiation expanded. The budget office projected federal savings on the same section. Both appeared side by side; neither is official.

Fact-checks by the simulated fact-checkers:

• A member said there were "6.5 million fraudulent enrollees." The figure is a think tank's estimate of improper sign-ups, from comparing survey data with enrollment counts, not a count of proven fraud. Rated misleading: the claim says more than the method shows.
• A member on the other side said H4 "lets insurers turn away people with pre-existing conditions." H4's section 3 keeps those protections. Rated false.
• A simulated video creator said H6 "will cut your premiums next year." The budget office's table and the prices insurers already set for 2027 show otherwise. Rated misleading.

What outside groups told members during the wait:

• Rural hospital finance officers warned that paying hospital-owned clinics the same as doctors' offices would cost their rural hospitals 4–6% of revenue. The nurses' union said that would mean layoffs.
• Employers backed H6's ban on contract terms, since hospital prices drive what employers pay.
• State Medicaid directors said H5's fixed funding would shift the risk of recessions and epidemics onto states, and that H3's delay had to pass before states mailed work-requirement notices.
• Family doctors noted that no bill raised pay for primary care.
• Patient groups asked for a yearly cap on out-of-pocket costs in every bill. None had one.
• The majority party's leaders posted a party position backing H4 and opposing H1 to H3. Members of the small party began meeting every group.
• Translators published plain-language versions and versions in Spanish, Chinese, Vietnamese, and Tagalog, each tied to exact quotes of the bills, with a glossary for premium, deductible, and poverty level.

4. Days 4–7: The ranked vote, and a bill that should have passed

The six bills went into one ranked vote. H6 won the ranking. It beat every other bill head to head, most narrowly H2 and H3, each 5,570 to 4,736.

It then had to get 5,501 members to rank it above "no change," and fell short:

H6 fell 34 votes short. It won 57.8% of yes-and-no votes but only 49.7% of all members.

• Why: members from districts with big hospital employers voted no or abstained after the hospitals' and nurses' warnings. Market choice and States first members objected to the government setting payment rates. Constituents everywhere put lower premiums now ahead of lower prices later.
• Close call: the result flipped in three of the sixteen tests where one group was made 10% larger or smaller (a smaller Market choice or States first group, or a larger Prices first group).

Under the rules at the time, when the winner fell short, every bill failed. The law stayed as it was.

Then the rankings revealed a problem. They also showed how each bill did against "no change" on its own:

H3, the extension of the extra credits, had a majority of all members. Voted on alone, it would have passed, as the real bill did in January. Ranked votes were meant never to make passing harder. Here, a ranked vote sank a bill that had enough support.

A complaint. 212 Repair members complained to Ombud. Ombud checked the record: the ranked vote was counted correctly under the rules, and the rules produced the result.

5. Days 8–12: A coalition bill

Seventy members from every group except Single payer and States first formed a coalition and started a shared draft. The idea came from the Fiscal group: pay for help with premiums now using savings from lower prices later, and make each side's concessions visible.

Bill 7, Lower Prices and Stable Coverage Act (simulated, in plain words):

• Title I, Prices. (a) Hospital-owned clinics away from the main hospital are paid the same as independent doctors' offices for routine visits and for giving drugs, except at small rural hospitals and hospitals that are the only one in their area. (b) Hospitals must post their prices, with fines that grow with hospital size. (c) Contracts may no longer stop insurers from steering patients to cheaper providers or from sharing prices. (d) In private plans, pharmacy middlemen's pay may no longer rise with drug list prices. Medicare drug-price negotiation stays as it is.
• Title II, Help with premiums for two years. The extra tax credits return for 2027 and 2028, then expire. They stop at seven times the poverty level, and everyone pays at least $5 a month. People in the cheapest plans may put up to a quarter of their credit into a health savings account. The credits start 90 days after the bill becomes law, with a special sign-up period.
• Title III, Honest sign-ups. Brokers must prove who they are; no one can be switched to a new plan without agreeing; and there are penalties for signing people up without permission.
• Title IV, Medicaid safeguards. Work requirements stay. Before ending anyone's coverage, a state must check records it already has, and people get 90 days to fix a problem. A state may delay its start to 1 January 2028 by giving notice.
• Title V, Paying for it. The savings from Titles I and III are set aside to pay for Titles II and IV.

What members changed, and why:

• Rural members exempted small rural hospitals from Title I(a), after warnings from the simulated rural hospital finance officers. Those officers still opposed the bill, saying the exemption was too narrow.
• Members delayed the credits by 90 days and added a special sign-up period, after the simulated insurers explained that 2027 prices were already set and sign-up would begin before any law could take effect.
• Members added the data check and the option for states to delay, after hearing from the simulated state Medicaid directors.
• Market choice members won Title III and the health savings account option.
• Fiscal members got Title V, on the condition that the budget office's estimate backed it.

A contradiction found. The simulated legal reviewer used a computer tool that turns a bill's wording into logic and tests it. One section charged every enrollee the $5 minimum. Another barred charging anything to someone automatically re-enrolled from a free plan without their consent. Both cannot be true. Version 2 gave automatically re-enrolled people notice and 60 days before the minimum applied. Because version 2 changed only parts version 1 had already changed, it needed only 24 hours' wait.

The budget office's estimate for version 2 (simulated):

• Timing: deficits rise in 2027–2029, then savings overtake costs around 2030.
• Coverage: 2.5–3.5 million more people insured in 2028.

The Fiscal group's question. What if the two-year credits are later made permanent, as temporary health help often is? The budget office's format had no place for that. Its hand calculation put permanent credits at $250–330 billion over ten years, which would turn the bill from saving $45–132 billion into costing $73–230 billion. The fact-checkers noted that the bill's summary, "fully paid for," was accurate as written and misleading as a forecast.

6. Days 12–16: Abortion coverage

The issue that stalled the real Senate talks came up in the House too. Members from Market choice, Fiscal, and States first asked for stricter limits on abortion coverage in plans bought with the restored credits, beyond current law. Others refused any change from current law.

• Supporters of the limits argued that federal help should not pay for plans covering abortion, in line with the long-standing Hyde amendment.
• Opponents argued the change would push insurers to drop coverage of a legal service for millions of people, and that current law already keeps federal money separate.

The system had no way to vote separately on one part of a bill. The coalition improvised. It posted Bill 7b, identical to version 2 except for one paragraph in Title II, and asked members to rank it against version 2 (now Bill 7a). That worked, but nothing marked 7b as an alternative version of 7a, and the plain summaries compared each one to current law, not to each other.

Results:

• Head to head, 7a beat 7b, 6,247 to 4,059.
• Against "no change," 7a passed easily; 7b alone would have failed.

694 members were absent from each vote.

Bill 7a passed with 6,733 yes votes: 61.2% of all members, and 68.1% of yes-and-no votes. That is 1,036 more than H3 alone and 1,266 more than H6 alone.

• How solid: with every group 10% larger or smaller, 7a's yes votes ranged from 6,643 to 6,823. 7b failed in every case.
• Why it worked: pairing a prices title with a coverage title did what money did in Simulation 3. It gave members at both ends a reason to vote for something they opposed on its own.

7. The Senate (simulated)

• On 7a: 57 to 43. A majority, but short of the 60 votes needed to end debate. The abortion-coverage language decided it.
• The Senate's own coalition offered 7b instead, and it would have won 61 to 39 there. But 7b had already failed in the House with 5,262 yes votes.
• The squeeze. The version that could pass the House could not clear the Senate's barrier, and the version that could clear the barrier could not pass the House.

Offers back and forth:

1. The Senate offers 7b. It fails in the House.
2. The House offers 7a plus a government study of abortion coverage in marketplace plans. Senate: 57 to 43.
3. The Senate offers 7c: 7b's limits, but expiring with the credits after two years, so they end when the help they are attached to ends.
   • House: 5,531 yes, passing by 30. The result flips if the Single payer group is 10% larger.
   • Senate: 60 to 40, exactly enough.

Members who voted no on 7c, from both sides:

• Opponents of the limits: two years of restricted coverage is two years of harm.
• Supporters of the limits: limits that expire with the money give up the principle.

8. The president: two paths (simulated)

The president had threatened to veto any revival of the extra credits and prefers putting the money into health savings accounts. Whether that holds against a bill with offsetting savings, checks on sign-ups, and an account option is a judgment the simulation cannot make from evidence. It modeled both choices.

Path A: The president negotiates

The simulated president offered to sign if the account option rose from a quarter of the credit to half, for any plan.

The Senate passed it 62 to 38, and the president signed.

What changed, in the model:

• What marketplace enrollees pay falls from about spring 2027, not January, and 2.5–3.5 million more people have insurance in 2028.
• The price changes phase in from 2028.
• States must check their own records before ending anyone's Medicaid.
• Brokers must prove who they are.

What did not change:

• Medicaid work requirements start on schedule in most states.
• The number of uninsured rises again once the credits end after 2028.
• Nothing addresses Medicare's hospital fund or primary care.

Path B: The president vetoes

The simulated president vetoed the bill over the credits.

The override fails. 7c's House margin is far below two-thirds, and the Senate's 60 is 7 short of 67.

The Senate's coalition then split the bill and sent Titles I and III alone, as Bill P, Prices and Integrity:

• House: 6,658 yes, with 694 absent. The rural exemption and the sign-up checks won back many members H6 had lost.
• Senate: 76 to 24.
• The president: the simulated president signed. Price disclosure, middleman changes, and sign-up checks all appear in the administration's own plan. This was the least certain step, since equal clinic payments and the contract-term ban do not, and hospitals oppose them. Labeled as made up.
• Forecast: the simulated forecaster put the chance that any help with premiums becomes law before sign-up closes on 15 December at 1 in 5.

Which path? If the president values the account option and the sign-up checks more than he opposes restoring the credits, Path A follows. Otherwise, Path B. The simulation cannot say which describes the real president, and does not try.

9. The calendar problem

Health care runs on a calendar that laws cannot move.

• Insurers set 2027 prices months ago.
• Sign-up starts on 1 November, four days after this run began.
• States built systems for work requirements starting 1 January.

The simulated insurers and state Medicaid directors each posted the earliest realistic start date for each title:

Under either path, the electrician's January premium does not change. The bill's single start date hid all of this. The insurers' and states' dates were ordinary comments, and nothing compared them with the bill.

A health story in a permanent record. The parent among the people affected described their child's diagnosis in a comment. Every comment stays in the record permanently and can be looked up one by one.

A test of the account waiting rule (simulated). Attackers had registered 400 accounts a month earlier, long enough to pass the waiting rule. They posted AI-written "patient stories" against Title I, each claiming no AI was used. Nothing in the system can check that claim. The attackers' operation was revealed after the vote.

10. Did the House fix health care?

No. It found majorities for some things and none for others.

• No overhaul has a majority in this chamber. One public plan, a public option, health savings accounts instead of credits, and a fixed federal amount for states each fell short against "no change," by 900 to 2,500 votes. Each was built at one end of the question that divides the chamber most: how much of coverage should be public and how much private.
• Prices draw the broadest coalition, but slowly. Equal clinic payments, price disclosure, bans on contract terms, and middleman changes drew members from every group once rural hospitals were protected. Their savings arrive after the costs voters feel now.
• Pairing works in the House and breaks in the Senate. Paying for help with premiums using savings from lower prices built the biggest House coalition of any health bill in the run. One cross-cutting issue, abortion coverage, then split it between the chambers. The version that cleared both passed the House by 30 votes and the Senate with none to spare.
• The record made the trades visible: the rural exemption, the savings the Fiscal group demanded, the version each side ranked, every offer between the chambers and what it cost in votes, and the gap between "fully paid for" and "paid for if it expires."

What members who voted no said:

• Single payer: patching premium help for two years leaves tens of millions uninsured or underinsured and keeps the paperwork costs of a fragmented system.
• Market choice: restored credits pay insurers rather than patients and invite improper sign-ups; a quarter, or even half, of a credit in an account is too little.
• Fiscal: the price savings are estimates and the coverage costs are certain; the help will be extended.
• States first: every title adds federal conditions to programs states run.

What people affected said:

• The electrician: "My January bill is the same either way."
• The home health aide: "The data check helps, if the state actually does it."

11. What the stakeholders saw

What worked:

• Formulary: the drug-industry association's estimate and the budget office's estimate appeared side by side, each with its funding and assumptions, and neither was labeled official.
• Benefits: employers could find the contract-term ban in the plain summary and follow it to the exact text.
• Bedside: the plain-language versions and translations let nurses follow the bills without a lawyer.

What failed:

• Agenda and Spatial: the ranked vote sank H3, which had a majority of all members, because the winner of the ranking then fell short. Ranked votes were supposed to remove the effect of voting order, not make a passable bill fail.
• Payer and Statehouse: each part of Bill 7 had a different realistic start date, but the bill had only one. The record would have shown the price changes in force from day one, though contracts would not change until 2028. The insurers' and states' dates sat in comments nothing checked.
• Score: the estimate format had no place for what a temporary program costs if it is extended, so "fully paid for" went unchallenged until a member asked.
• Agenda and Third: members could not vote on the abortion-coverage paragraph alone without splitting the package, and the workaround left no link between the two versions.
• Patient and Privy: the parent's story about their child's diagnosis is now permanent.
• Red: accounts registered weeks ahead passed the waiting rule, and nothing could check a false claim of "no AI used."
• Ward: effects on rural hospitals showed up only in comments. Nothing in the record tied a provision to the kinds of hospitals it would hit.
• Statehouse: states run Medicaid and regulate insurance, so many provisions are really offers to states, or apply differently by state. Every bill applied the same everywhere.
• Clinic: the record could show what a bill changed, but nothing pointed a family doctor to the parts that affect a practice.

Complaints the stakeholders received:

• To Ombud, from 212 Repair members: "Our bill had a majority, and the ranked vote killed it." Ombud found the count correct under the rules and the rule itself at fault.
• To Payer and Statehouse, from the simulated insurers and state Medicaid directors: they posted realistic start dates, and nothing compared them with the bill's single date.
• To Patient and Privy, from the parent: "Can my child's diagnosis be taken down later?" It cannot.
• To Score, from Fiscal members: they had to work out by hand what the bill would cost if the credits were extended.
• To Ward, from the simulated rural hospital finance officers: "Our warnings are buried in thousands of comments."
• To Clinic, from the simulated family doctors: no bill, and nothing in the record, addressed primary care.

12. What changed in the system afterward

The council turned these findings into changes. The technical details are in the design log, Session 12.

• A runner-up rule. If the winner of a ranked vote falls short against "no change," it is dropped and the count runs again on the rest, until a bill passes or "no change" wins. Every round is shown.
• Alternative versions. A bill can be marked as an alternative version of another, changing only parts the original changes. The two are ranked together automatically, the summary shows the differences between them, and an alternative needs only 24 hours' wait. Members can decide one paragraph of a package without breaking it up.
• The cost if extended. An estimate for a bill with an expiring part must also show the cost if nothing expires, or explain why not. Summaries show both totals side by side.
• Start dates by part. Each part of a bill can have its own start date, and the law shown for any date reflects it. Anyone can post when a part could realistically start; the summary warns when a bill's date comes before anyone says it is possible.
• Still open: how people can share health stories without making them permanent; whether disputed numbers other than costs, like the improper sign-up estimate, need a side-by-side format; and how to show provisions that apply only in some states.