The Electoral Science Office¶

A Primer Series Article of the Congressional Elections Modernization Act (CEMA)¶

Prepared by Albert Ramos for The American Policy Architecture Institute

Before a new medicine reaches a pharmacy shelf, it moves through years of trials, and the results are published for any doctor or researcher to examine. Before a new aircraft carries passengers, it logs thousands of hours of testing. Before a bridge opens to traffic, engineers load it past its rated weight and document what happens. That is how a serious country treats decisions that carry weight: test first, publish the evidence, then deploy. Yet the most consequential decision the United States makes — who governs — runs on machinery no institution has ever been assigned to test. How ballots are designed, how votes are counted, how those votes become representation: no federal body exists whose job is to study whether any of it works as intended.

The Congressional Elections Modernization Act fills that gap by establishing the Electoral Science Office — the nation's first agency dedicated to studying how congressional elections work and proving that they work. The Office tests voting methods before they are used, measures how they perform afterward, and publishes the data, the models, and the computer code that let anyone verify the results independently.

The Office is not an add-on to the Act's other changes. It is the part the rest rely on to stay honest. CEMA changes how House districts are drawn, how votes are counted, and how candidates reach the ballot — and every one of those changes will be questioned and second-guessed for as long as it operates. Without an institution able to answer those questions with public evidence, the changes cannot be tested before adoption, trusted after it, or kept honest over time.

The Gap Nobody Filled¶

The United States has federal institutions that touch elections. One regulates campaign money. One certifies voting equipment and sets administration standards — whether the machines work, not whether the methods running on them serve voters well. One defends election systems against attack. States and counties run the actual contests, staffed by dedicated professionals working under tight budgets and fifty different sets of state laws. And university researchers produce serious work on voting and representation, but they do it without reliable access to election data and without any channel connecting their findings to the officials who design and run elections.

Each of these institutions does a real job, and none is the wrong body for the work it was built to do. But none of them — and no combination of them — does the job that is missing: testing voting methods before they are deployed, measuring them after, and publishing the evidence in a form the public can check. The country has bodies that regulate its elections, certify them, defend them, and administer them. It has never built one whose mission is to understand them.

The gap is not an accident of neglect. It reflects an old assumption — that elections are administrative events to be managed, not systems to be studied. When elections were simpler and the range of possible methods narrower, that assumption was workable. It has not kept pace with what we now ask elections to do. A system that must translate hundreds of millions of individual preferences into representative outcomes, resist manipulation, stay transparent enough to verify, and remain clear enough for every voter to use is a system that benefits from rigorous, ongoing, independent evaluation. The gap is real, it is structural, and filling it has never been anyone's job.

What the Gap Costs¶

Two consequences flow from that vacuum, and they are worth understanding together, because they reinforce each other.

The first is a trust gap. When no independent, technically credible body can show the public how its elections work, the space where that evidence would go gets filled by assertion instead. A claim that a voting method produced fair outcomes — or failed to — cannot be checked against a published record that holds regardless of who is making the claim. Credibility becomes a question of which voice is louder, not which case is better supported. Verification, as things stand, is scattered: a state official here, a court ruling there, a federal statement somewhere else, none of it carrying singular authority. An institution that publishes its work, keeps its data open, and invites independent checking is believed because its record can be checked. Without one, trust in the machinery is a political argument rather than an empirical one — and when trust is contested rather than earned, every election risks becoming a referendum on the machinery itself rather than on the candidates it exists to weigh.

The second is a reform gap. The current landscape creates a bind that reliably stops electoral improvement before it starts. Suppose a state, or Congress, wants to consider a better way of running elections. Critics will reasonably demand real-world evidence that the new method works. But there is no way to gather real-world evidence about a voting method without first adopting it somewhere — and adopting it without evidence is exactly what the critics object to. The demand is fair. The system simply makes it impossible to satisfy. The only options on the table are computer models, which skeptics discount, and full deployment on a live electorate, which is reckless. Nothing exists in between.

So the methods already in use carry on by default — not because anyone evaluated them and found them best, but because the machinery for evaluating anything at all was never built. The first gap costs the country confidence in the elections it has. The second costs it the ability to improve them. The Electoral Science Office is designed to close both.

Trust Through Transparency¶

The Office is an independent federal agency with a single mission: make the mechanics of congressional elections a matter of public evidence rather than official assurance. Its design rests on one organizing insight — in a democracy, an institution earns trust not by claiming authority but by opening its work to scrutiny. What the Office does matters less than how it does it: in the open, with every input, method, and result published for anyone to check.

Its first job is testing before deployment, and the Act lays out a ladder of evidence, each rung stronger than the last.

It begins with computer simulations. The Office runs the Act's voting methods through thousands of synthetic electorates — different mixes of voters, candidates, and political conditions — to show that the methods behave as designed before any human participant is involved. All the code, the inputs, and the results are published, so anyone can run the same model and confirm they get the same answer.

Analysis of real ballots comes next. Where jurisdictions already keep detailed records of voter preferences from past elections, the Office can re-count those ballots under the Act's methods and publish what would have happened — genuine voter preferences, no new election required.

Then come full-scale mock elections: fictional candidates with published platforms, compensated participants with real preferences to express, certified voting equipment, and actual election offices running the whole process from ballot printing through results reporting. The Act requires these in no fewer than five geographically and demographically varied jurisdictions before the first congressional election held under the new rules — a floor, not a ceiling.

Parallel ballots provide the strongest evidence of all. During the transition years, voters in cooperating jurisdictions cast their official ballot as usual and may then, voluntarily, complete a second ballot in the Act's format for the same races. The parallel results carry no legal effect, but they are counted and published — real voters, real candidates, real stakes, before a single binding outcome depends on the new system.

This ladder is what breaks the bind described above. The objection "there is no real-world data" stops being a conversation-ender and becomes a research request — something an institution now exists to answer, step by step, with each answer published as it arrives.

The second job is the one that makes the first believable: the Office publishes everything. The source code. The input data. The models. The counting procedures. Every finding arrives with the materials needed to reproduce it, so that any researcher, journalist, election official, or curious citizen who runs the same inputs can confirm they get the same outputs. This is the load-bearing idea of the whole design. The Office does not ask the public to trust its conclusions because they are official. It hands the public the means to check them.

That is the difference between an institution the public believes and one it merely tolerates. NASA is a useful illustration. When NASA reports something about space, the default public response is belief — not because NASA is infallible, but because decades of its work have been openly documented and independently checkable by outside scientists, foreign agencies, and anyone willing to examine the data. Its credibility rests on transparency, not on rank. The Electoral Science Office is built to occupy that same niche for elections: an institution whose word carries weight because its work is open, and because being open leaves it nowhere to hide.

After deployment, the Office's work continues. It measures how the methods are performing — whether multi-seat outcomes come out proportional, how often voters complete their ballots cleanly, how well people understand what they are marking — and publishes its findings every year. Where the evidence shows a technical detail should be adjusted, it can propose the change through a public process with Congressional review, while the core democratic guarantees written into the Act stay beyond its reach to alter on its own.

Addressing Concerns¶

A proposal like this earns two immediate concerns, and both deserve answers at full strength.

"Isn't this just another agency that will get captured by whoever's in power?"¶

The worry is reasonable. An agency that studies elections sits close to the most contested terrain in American politics, and the temptation to bend it will be constant. Institutions meant to be neutral have been turned partisan before; good intentions on paper do not survive determined pressure. The Act's answer is to treat capture as the expected attack, not the unlikely one, and to build the defenses in layers.

Commissioners serve fixed, staggered terms, so no single president or Congress can install a majority at once. The power to appoint them is shared rather than held in one set of hands, technical qualifications are required, and no party may hold more than a bare majority of seats. Commissioners can be removed only for cause — misconduct or incapacity — never for producing findings someone in power dislikes, and that protection is strongest during active redistricting cycles, when the pressure to interfere runs highest. Executive branch officials, the White House and its budget office included, are prohibited by statute — not by custom — from directing or pressuring the Office's technical determinations, and any attempt must be documented and reported. The Office's core funding is a floor written into the law itself, paid out directly and shielded from the budget maneuvers used to starve inconvenient agencies into silence. And the Office cannot be quietly folded into another agency or reorganized out of existence.

No single one of these protections guarantees virtue. Together, they are built for the worst case — a determined administration trying to bend or silence the agency — because protections that hold only in friendly political weather are no protection at all.

"Who watches the watchers? How do I know it isn't cooking its own results?"¶

The honest answer is that you are not asked to know it. You are equipped to check it. Because the Office publishes its code, its data, and its methods, no finding can stand on the Office's say-so alone. If it shaded a simulation or miscounted a mock election, the proof would sit in the very materials it is required by law to release — available to every rival researcher, every skeptical journalist, and every state election office in the country. An outside advisory panel, drawn from universities, election administration, and international election bodies, also reviews the Office's research agenda and publishes its own independent yearly assessment. But the deeper safeguard is simpler: when everything is published, the watchers are everyone.

The verification is not promised. It is handed over. That is the load-bearing difference.

What the Office Does Not Do¶

Just as important as the Office's mission is its boundary, and the Act draws it deliberately.

The Office does not run elections. States and their local offices administer every congressional election, as they do now; CEMA does not federalize election administration. The Office does not count the official vote, certify results, or declare winners — no ballot with legal effect ever passes through its hands. It does not regulate campaign money; that authority stays where it already is. And it does not tell anyone how to vote or which candidates deserve support. Everything it publishes explains how the system works, never how to use it to any side's advantage.

What it does is bounded and concrete: it studies, it designs technical specifications, it publishes, and it verifies. Naming those limits is not modesty for its own sake. An agency that designed, ran, counted, and policed elections all at once would be a single point of failure and a single irresistible target. An agency that studies, designs, and publishes — while states administer and other federal bodies certify equipment and audit compliance — is far harder to capture and far easier to hold to account. The narrow scope is part of what makes the institution trustworthy, not a concession away from it.

Conclusion¶

The trust gap and the reform gap look like two separate ailments, but they trace to one missing institution: the country never built a body for understanding its own elections. The Electoral Science Office is that body — the testing ground that lets a reform be proven before it is trusted, and the open record that lets trust rest on verification rather than reassurance.

The Office cannot do the work of the changes it serves. House Expansion, Hands-Off Mapping, STAR Voting, and Full-Field Elections each change something the Office does not touch — how many seats there are, how the lines get drawn, how votes are counted, who reaches the ballot. What the Office does is make sure those changes are grounded in evidence before they reach voters, measured honestly after, and open to improvement as the evidence accumulates. A republic that builds the instrument for measuring its own machinery is making a stronger claim than one that asks to be believed. It is saying: examine us. Run the numbers yourself. The evidence does not belong to whoever holds power — it belongs to anyone who checks.

The price of building that instrument is small by federal standards: roughly $280 million over three years to establish the Office, and roughly $75 million a year to run it. The United States spends about $24 billion a year exploring space, and over the last five years it has spent an average of more than $30 billion a year recovering from natural disasters. Against numbers like those, the cost of finally instrumenting the decision that governs all the others is a rounding error. The real question is the cost of going without it — elections defended by scattered voices against coordinated doubt, and reforms stalled forever at a question no one is equipped to answer. A free people choosing who governs them deserves better than flying blind.

Other articles in the CEMA Primer Series:

Works Cited¶

U.S. Congressional Research Service. "NASA Appropriations and Authorizations: At a Glance." R43419. 2025. https://www.congress.gov/crs-product/R43419

USAFacts. "How Much Does FEMA Spend on Disaster Response?" 2026. https://usafacts.org/answers/how-much-does-fema-spend-on-disaster-response/country/united-states/

The Congressional Elections Modernization Act (Rev 6.2) is an internal APAI document available on the same website and is referenced in prose by name. The Office's establishment and operating cost figures are the Act's own appropriations figures (Section 409(k)).

Revision history available in the raw file.

📥 Download this document (opens on GitHub -- click the ⬇ download button)