Stakeholder briefs

What this is. A short brief for each of the 62 stakeholders on the design council, as of the reruns of Simulations 1 to 5 on 29 October 2026. Every stakeholder is made up, and Claude played them all. No stakeholder speaks for a real person, organization, region, or faith.

The rules every stakeholder follows

• They watch; they do not act. Stakeholders never vote, write or change bills, take a side on any bill, publish cost estimates or fact-checks, broker deals, or attack the system. Those jobs belong to the simulated members and other simulated roles.
• They watch from their own point of view. Each stakeholder looks at whether the system works for the people, places, or communities they know.
• Anyone in a simulation can complain to them. Members, and simulated roles outside the House, such as voters, states, hospitals, congregations, or the Senate, can bring a complaint about how the system treated them. The stakeholder records it and checks it against the record.
• What they find shapes the system, not the laws. Their findings go to the design council, which decides how the system's rules should change. Technical details are kept in the design log and the coding prompt.

Moved out of the stakeholder list. Four former participants are now simulated roles, because they act in the simulations rather than watch them: the residents of each district, the president and federal agencies, the Senate, and the people affected by a bill.

Builders

Ledger builds a test copy of the system run by several independent computers that must agree on every step, so no single operator controls the record.

• Watches for: whether every copy reaches the same result, and whether anyone can recheck it from scratch.
• What they saw: all five copies agreed on every result in all five simulations.
• Still unresolved: removing a comment, such as a parent's story about a child's illness, would break everyone else's ability to recheck the record. Ledger has not yet found a way around this.

Forge builds a test copy on the same tools programmers use to track changes to software.

• Watches for: whether bills and their revisions can be reviewed with ordinary tools.
• What they saw: every bill, revision, and alternative version could be traced change by change.
• Still unresolved: classrooms and student governments are the most likely first real users, and the system should stay simple enough for them.

Docket builds the simplest, cheapest test copy: one public log that anyone can audit.

• Watches for: whether the system stays easy and cheap to run.
• What they saw: each new feature, from ranked votes to start dates by part to laws that apply only in some states, added work, but its copy kept up.
• Still unresolved: whether a simple copy watched by outside auditors protects the record as well as Ledger's design. Only real tests at full size will tell.

Loom builds a test copy that works offline and catches up later.

• Watches for: people who are not always connected.
• What they saw: its copy correctly marked results as "not yet final everywhere" until it caught up. The 72-hour minimum voting period, added after Simulation 5, gives late-connecting members a fair chance.
• Still unresolved: deadlines are hard to enforce without a shared clock.

Margin builds a read-only copy that rechecks everything the others do.

• Watches for: whether every result can be confirmed by someone who trusts no one.
• What they saw: it confirmed every count in every simulation.
• Complaints received: none.

Video creators

Quick Take makes short videos for a young, fast-scrolling audience.

• Watches for: freedom to clip and share public proceedings without asking anyone's permission.
• What they saw: in Simulation 1, a simulated creator posted "a majority backed the ceasefire bill." That was true of votes cast but not of all members, and the system then showed only one percentage. Every result now shows both.
• Complaints received: from simulated creators, about a clip that was accurate but missing context. That is still an open question.

Deep Dive makes long explainer videos, paid for by ads and sponsors.

• Watches for: full source material and a clear history of every change.
• What they saw: a simulated creator built an explainer from a bill's history, with every source traceable.

Independent is a creator funded by subscribers.

• Watches for: links that keep working, and records that do not depend on any one platform.
• What they saw: links to an old version of a bill still showed that version, with a pointer to the latest one.

Fact-checkers

Verdict watches checks of public claims about bills and votes.

• Watches for: whether a claim can be checked against the exact text.
• What they saw: the plain summary made checking misleading descriptions fast. Translations tied to exact quotes let fact-checkers catch a serious mistranslation in Simulation 2. But claims made on television could not be checked until members gained a way to put outside statements on the record.
• Complaints received: from simulated fact-checkers, that a correct fact-check was buried for a day under 2,000 copies of a false claim (Simulation 2).

Lint watches checks of bills for contradictions and undefined terms.

• Watches for: whether bills can be tested for contradictions before the vote.
• What they saw: the contradiction test caught a loophole in Simulation 1 and a clash between two sections in Simulation 4. In Simulation 5, the simulated legal reviewer found an undefined term.
• Still unresolved: some rules, like what a fast-track budget bill may contain, cannot yet be written into the system.

Record watches comparisons of what members say with how they vote.

• Watches for: a complete, public record of votes and statements.
• What they saw: a member's false claim about their own vote was caught in one step (Simulation 1).
• Still unresolved: secret ballots would protect members from threats but make this kind of check impossible.

Markets and rankings

Odds watches betting markets that take bets on outcomes.

• Watches for: clear, timely, and identical answers to questions like "Will this bill pass?"
• What they saw: every question settled the same way on every copy, and questions about simulated sessions had to say "simulated."

Tape watches forecasters who publish odds.

• Watches for: forecasts that can be checked against what actually happened.
• What they saw: every forecast was settled as right or wrong, and the system never added up forecasts into official odds.
• Still unresolved: whether published forecasts change how members vote.

Leaderboard watches websites that rank bills and members.

• Watches for: rankings anyone can reproduce.
• What they saw: in Simulation 1, a simulated ranking site used the yes share of votes cast, which made a failing bill look like a winner. Rankings now show both percentages.

Parties and influence

Lobby watches industry groups that suggest wording for bills.

• Watches for: the same treatment as everyone else, with no special rules for interest groups.
• What they saw: when a simulated trade group's suggested wording appeared in a bill, the record showed it, along with who funds the group (Simulation 1).

Cause watches advocacy groups funded by members and grants.

• Watches for: the same access as well-funded groups, and public view of who shaped a law.
• What they saw: in Simulation 1, one test copy blocked all of a simulated coalition's accounts together. Limits now apply to each account separately.
• Complaints received: from that coalition.

Whip watches large parties that coordinate their members.

• Watches for: whether a party can state its position and organize without controlling anyone's vote.
• What they saw: party positions were shown, but only members could vote, and no party could vote for them.
• Complaints received: from simulated party leaders, that the 72-hour wait could not be shortened during the tanker strike (Simulation 1). An emergency path now exists.

Third watches small parties that often hold the balance.

• Watches for: protection from a majority that could change the rules to shut others out.
• What they saw: the rules for changing the House's own rules could not be weakened during a session. In Simulation 4, members could not decide one paragraph of a package without breaking the package up, until alternative versions were added.

Ombud is an independent complaints officer.

• Watches for: fair process, and evidence it can verify.
• What they saw: signed receipts and a public list of refusals let it prove a test copy's error in Simulation 1.
• Complaints received: from a blocked coalition (Simulation 1); from members whose stolen logins cost them their votes (Simulation 2); from voters and agencies who feared a shutdown fight (Simulation 3); from 212 members whose passable bill was sunk by a ranked vote (Simulation 4); and from 268 Sabbath-observing members shut out of a vote (Simulation 5).

Scholars and lawyers

Spatial is a political scientist who studies voting records.

• Watches for: complete vote data that anyone can reanalyze.
• What they saw: two questions divided each simulated chamber. Full turnout in Simulation 1 made passing look too easy, so every simulation now models absences and retests each vote with every group 10% larger and 10% smaller.

Agenda is a political scientist who studies how voting rules shape results.

• Watches for: rules whose effects are understood before they are used.
• What they saw: ranked votes could not create a majority that did not exist (Simulations 1 and 2). In Simulation 4 a ranked vote sank a bill that had a majority. The runner-up rule fixed that, but in Simulation 5 it twice passed a bill fewer members preferred, without telling them.
• Complaints received: from Humanitarian members in Simulation 5, asking why the version more of them preferred lost.

Bench is an appeals judge.

• Watches for: knowing exactly which text is the law, and from when.
• What they saw: bargaining between the House, Senate, and president left no record in Simulation 3. The final compromise appeared as a new bill with no link to the offers behind it.
• Complaints received: from Restraint members, who could not see which offer gave up the deadline.

Chambers is a trial judge.

• Watches for: records that can be proven genuine in court without an expert.
• What they saw: a copy of any record can be checked offline, with no network connection.

Counsel is a lawyer who advises organizations on what the law requires.

• Watches for: the law in force on any date, and stable references to it.
• What they saw: the record showed when an approval of force would expire (Simulation 1). It would have shown health price rules in force from day one though contracts changed in 2028, until each part of a bill could have its own start date (Simulation 4).

Advocate is a public-interest lawyer.

• Watches for: proof that procedure was followed, and laws not rushed through.
• What they saw: the record proved the 72-hour wait had passed (Simulation 1). Advocate warned that an emergency path could become routine; it now requires two-thirds of all members and expires after 30 days.

Security, law, and education

Command is a retired military officer and military lawyer.

• Watches for: whether end dates and limits in a law can be carried out safely.
• What they saw: end dates applied automatically, and a 30-day withdrawal period kept a funding cutoff from stranding troops (Simulation 2).

Badge is a law-enforcement official focused on corruption and threats against officials.

• Watches for: evidence that stays intact, and members kept safe.
• What they saw: the permanent record preserves evidence, and required disclosure of any money riding on an outcome gives investigators a starting point in corruption cases. No authority, including police, gets special access.
• Complaints received: from members who received 14 threats after their votes were shared (Simulation 2).

Clearance is a national-security professional.

• Watches for: whether the rule against secret law holds when classified matters come up.
• What they saw: it held in Simulations 2 and 3. Classified details went to the intelligence committees, and nothing secret entered any law.
• Still unresolved: real defense laws sometimes rely on classified attachments.

Classroom is a civics educator.

• Watches for: students' safety and understanding.
• What they saw: a 30-student class ran a full session privately and could erase it afterward (Simulation 2).
• Complaints received: from the simulated teacher, that students could not follow how the winner of a ranking could still fail.

Score is a nonpartisan budget expert.

• Watches for: cost estimates that can be compared and checked.
• What they saw: competing estimates appeared side by side, and none was official. But dollar amounts buried in sentences could not be added up (Simulation 3), and estimates did not show what a temporary program costs if it is extended (Simulation 4). Both have been fixed.
• Complaints received: from the simulated budget office (Simulation 3) and from Fiscal members (Simulation 4).

Diplomacy, security testing, access, accounts, and privacy

Envoy is a diplomat and international-law expert.

• Watches for: whether the record can keep up with negotiations.
• What they saw: negotiations moved daily, and the only way to reflect them was an ordinary comment.

Red is a security tester.

• Watches for: attacks the system cannot stop.
• What they saw: simulated attackers flooded comments with 2,000 new accounts (Simulation 2). New accounts must now wait before posting. Attackers then used accounts registered weeks ahead (Simulation 4), and accounts falsely claiming to speak for churches (Simulation 5).
• Complaints received: from simulated fact-checkers whose work was buried by the flood.

Access is an advocate for accessibility and other languages.

• Watches for: bills everyone can read.
• What they saw: translations tied word for word to the bill caught a serious error (Simulation 2), and plain-language versions helped students, nurses, and farmers.
• Complaints received: from a simulated reader in Isfahan about a Persian translation that made a non-binding statement sound binding.

Registrar is an election administrator.

• Watches for: member logins that are issued, protected, and replaceable.
• What they saw: three stolen logins cost real members their votes (Simulation 2). Now a member and the account office together can replace a stolen login. The account office cannot do it alone.
• Complaints received: from those three members.

Privy is a privacy and civil-liberties expert.

• Watches for: what the public record reveals about people.
• What they saw: members' votes and statements together made easy target lists (Simulation 2). A parent's story about a child's diagnosis became permanent (Simulation 4). No one without legal status could safely take part (Simulation 5).
• Complaints received: from threatened members, and from the parent.

Health care

Clinic is a family doctor in a mixed rural and suburban practice.

• Watches for: whether a doctor can tell from the record what a bill means for a practice.
• What they saw: the record shows what a bill changes, but nothing points a doctor to the parts that matter for a practice.
• Complaints received: from simulated family doctors, that nothing addressed primary care.

Ward is the finance chief of a hospital system with three rural hospitals.

• Watches for: whether effects on particular kinds of hospitals show up in the record.
• What they saw: warnings about rural hospitals appeared only as ordinary comments.
• Complaints received: from simulated rural hospital finance officers: "Our warnings are buried in thousands of comments."

Payer is an insurance pricing expert.

• Watches for: whether the record shows when each part of a bill could realistically start.
• What they saw: insurers had set 2027 prices months before any vote. Each part of a bill can now have its own start date, and the record warns when a date comes before anyone says it is possible.
• Complaints received: from simulated insurers whose start dates went unchecked.

Formulary is an economist at a drug-industry association, with funding disclosed.

• Watches for: fair, side-by-side treatment of competing estimates.
• What they saw: its simulated counterpart's estimate and the budget office's appeared side by side, each with its funding, and neither labeled official.

Benefits is a benefits manager at a 900-person manufacturer.

• Watches for: whether employers can find the parts of a bill that affect them.
• What they saw: the plain summary led straight to the contract rules that drive employer costs.

Statehouse is a state Medicaid director.

• Watches for: what states must do, and by when.
• What they saw: most health provisions work differently by state, but every bill applied the same everywhere. After Simulation 5, a part of a bill can now be limited to some states or apply only where states opt in.
• Complaints received: from simulated state Medicaid directors whose start dates went unchecked.

Bedside is a nurses' union representative.

• Watches for: whether working people can follow bills and comment on them.
• What they saw: the plain-language versions and translations let nurses follow the bills without a lawyer.

Patient is an advocate for people with long-term illnesses.

• Watches for: whether people can share health stories safely.
• What they saw: every comment is permanent. "A design that makes those stories permanent will hear fewer of them."
• Complaints received: from the parent who shared a child's diagnosis.

Cities, suburbs, and rural areas

Metro is a deputy mayor for immigrant affairs in a large city.

• Watches for: cost estimates that show costs landing on cities and counties.
• What they saw: the estimates showed only federal dollars. Estimates can now show local costs and who bears them.
• Complaints received: from simulated city officials about shelter costs.

Ring is the executive of a fast-growing suburban county.

• Watches for: whether results can be read for suburbs.
• What they saw: the suburbs split exactly in half on legal status, but the labels behind that finding had no definition or limit. Labels are now limited to region and type of place.

County is a commissioner of a rural county with a meatpacking plant.

• Watches for: start dates small governments can understand.
• What they saw: E-Verify's three start dates were clear in the record.

Acre is a dairy and vegetable farmer.

• Watches for: a record a non-lawyer can use on a phone.
• What they saw: the plain-language versions worked on a phone, but start dates were buried deep in the detailed comparison.

Regions

Harbor works in a Northeast state office for new Americans.

• Watches for: whether effects on regions can be compared across estimates.
• What they saw: regional effects sat in written notes. Estimates can now show them in a form anyone can compare.

Piedmont is a Southeast state economic-development official.

• Watches for: whether rules phased in by employer size can be written part by part.
• What they saw: they could.

Prairie works on a Midwest development commission for rural counties.

• Watches for: people with slow internet.
• What they saw: a rural member could download and check the record in time. The voting period, not the connection, was the limit.

Frontera is a judge in a Southwest border county.

• Watches for: whether the record can show a law that applies only in some places.
• What they saw: it could not, until after Simulation 5.
• Complaints received: from simulated state officials who could not tell what the law would be in their state.

Range works for a group of Mountain West resort and construction employers.

• Watches for: whether each simulation sets its own regions instead of one fixed map.
• What they saw: regions are now set by each simulation. The territories and the District of Columbia had no place in the map.
• Complaints received: from simulated territory residents.

Pacific is a West Coast state farm and labor official.

• Watches for: whether states' opposite choices can be recorded without favoring either.
• What they saw: after Simulation 5, a state's choice to opt in or out can be recorded, and neither choice is the default.

Religious traditions, old and new

Cedar is a cultural official of a tribal nation whose land crosses the border.

• Watches for: whether a tribal government can take part as a government.
• What they saw: it could only comment as an ordinary adviser.
• Complaints received: from simulated tribal government officials.

Parish is a Catholic diocesan director of services for migrants and refugees.

• Watches for: whether claims to speak for a church can be checked.
• What they saw: 300 fake accounts claimed to speak for churches and synagogues. Organizations can now vouch for the accounts that speak in their name.
• Complaints received: from simulated dioceses whose names were used.

Pulpit is a Baptist pastor in a fast-growing county on a city's edge.

• Watches for: whether voting periods fit weekly worship.
• What they saw: a weekend voting period would have shut out Sunday worshippers as surely as Sabbath observers.
• Complaints received: from Sunday-worshipping members.

Covenant is the director of a Jewish refugee-resettlement agency.

• Watches for: whether voting periods shut out people observing the Sabbath.
• What they saw: 268 members could not vote in Simulation 5. Every voting period is now at least 72 hours.
• Complaints received: from those members, and from simulated federations whose names were used.

Mission is a Latter-day Saint lay leader in the Mountain West.

• Watches for: respect for religious observance without naming any faith.
• What they saw: the 72-hour minimum does this. It favors no calendar.

Iglesia is the pastor of a Spanish-speaking Pentecostal church whose families include people without legal status.

• Watches for: whether people at risk can take part safely.
• What they saw: they could not.
• Complaints received: from congregants who wanted to speak.

Minbar is an imam and board member of a Muslim civic group.

• Watches for: anything that could become a religious test.
• What they saw: a request to add religion to the district labels. It was refused, and religion can never be a label.
• Complaints received: that request, from members.

Sangat is the secretary of a Sikh temple in a farming valley.

• Watches for: translations in the languages its members read.
• What they saw: a Punjabi version was published, tied word for word to the bill.

Humanist is the director of a secular humanist association.

• Watches for: equal standing for people with no religion.
• What they saw: vouching by organizations is never required, so people with no organization are not marked down for it. The rules on voting periods name no faith.
• Complaints received: the same request to add religion to district labels, which Humanist also opposed.