Confidence, Not Faith, in the Ballot Box

An election system should provide something that transcends disagreements: a process that can be examined.

By James E. “Trey” Trainor, III

I have spent more than two decades working in and around American elections, as a lawyer for candidates and campaigns, as general counsel to the Texas Secretary of State, as counsel to presidential campaigns, and eventually as a commissioner and two-time chairman of the Federal Election Commission. That has given me the opportunity to see elections from a lot of different angles. I have worked with candidates and campaigns trying to win, lawyers challenging election procedures, election officials trying to administer them, and regulators responsible for enforcing federal election law.

One thing I have learned is that election law works best when the rules are clear and the process can be independently checked. In short: Election integrity is ultimately an accounting problem.

While that may not sound exciting, it is an insight with deep roots, and it gets to the heart of the matter. Plato, in The Republic, argued that justice requires each part of a system to fulfill its proper function within an ordered whole. The same principle applies to elections: every component of the process must do its part, and the results must reconcile.

Elections are the process by which citizens bind themselves to the authority of their chosen leaders, a process that can succeed only if all can agree their will is reflected in the result. Believing this requires confidence that only those eligible to vote did so, and that votes cast were accurately tallied and reported. Thus, every election requires officials to account for who is eligible to vote, who actually votes, how many ballots are cast, where those ballots go, how they are tabulated, and whether the final result is supported by the underlying records.

At each stage, the records should reconcile. When they do not, the discrepancy should be identified, documented, and resolved rather than simply passed along to the next stage.

I learned the importance of that principle firsthand during the 2016 presidential recount in Michigan. I spent an entire eight-hour day observing the recount process in Wayne County, which includes Detroit. There were roughly 30 ballot boxes in the area I was observing. As I remember it, only one actually reached the point where the presidential ballots could be recounted.

The procedure in effect at the time began with a basic question: Did the number of people recorded as having voted in the precinct match the number of ballots in the box?

If the numbers matched, the recount could proceed. If they did not, the discrepancy went through an additional verification process. If the precinct still could not be reconciled, the ballots could not be recounted, and the election-night count remained the official count for that precinct.

The experience made a lasting impression on me. A recount is supposed to provide an opportunity to verify an election result. But that verification is only as useful as the records supporting it. If the number of ballots and the number of recorded voters cannot be reconciled, the problem exists before anyone starts examining individual votes.

An imbalance does not necessarily mean fraud. There can be perfectly innocent explanations: a spoiled ballot, a replacement ballot, a data-entry mistake, or some other administrative error. But if the underlying records do not reconcile, the ability to independently verify the reported result can be compromised before a recount even begins. The discrepancy still needs to be identified and explained.

That lesson applies to the entire election system. When the number of votes for each candidate sums to the number of ballots cast, which matches the number of voters who voted, each of whom appears on the voter rolls, which list only people eligible to vote, then the result is a true one. When people we trust have the ability to ascertain that this is the case, we have a good system.

Start with the Voter Rolls

The ballots themselves are only part of the election record. The first entry in the accounting system is the voter registration list.

Before an election can accurately count votes, its operators must know who is eligible to cast them. Federal law requires states to maintain accurate and current voter registration lists while also establishing safeguards against the erroneous removal of eligible voters.

That requires records. Officials need to know who was added to the rolls, who was removed, why a registration changed status, and what information supported the change.

The SAVE America Act represents an attempt to address this challenge. The legislation would establish additional federal requirements concerning documentation of citizenship for registration in federal elections and identification requirements for voting. The Senate version, S. 3752, was introduced in January 2026, while the House considered H.R. 7296 and passed the legislation in February. The Senate measure remained pending. In accounting terms, the SAVE Act addresses the front end of the ledger: if the eligibility determination is not documented at the point of registration, the gap carries forward into every later stage of the process.

Whatever one’s view of those particular requirements, the administrative question is straightforward: What establishes eligibility? Who verifies it? What records are created? And can the process be examined afterward?

Accurate voter rolls are not achieved by simply adding as many names as possible or removing as many names as possible. They require maintaining records that accurately reflect eligible voters and can withstand legitimate scrutiny. Both preventing eligible voters from casting ballots and allowing participation by ineligible voters can undermine an election’s legitimacy. List maintenance therefore needs to be both effective and documented.

Voter rolls also present a challenge for outside verification. I saw that from another angle when I represented True the Vote in Mississippi in 2014. Following a Republican primary runoff, True the Vote sought records under the National Voter Registration Act’s public disclosure provision. Mississippi law prohibits a voter who cast a ballot in one party’s primary from voting in another party’s runoff. The organization wanted to examine poll books and voter registration records to find out whether voters who had cast ballots in the Democratic primary had then voted in the Republican runoff. It alleged that state and county officials had denied access to those records, delayed it, or allowed it only with birthdates and other information redacted, in violation of the NVRA.

The federal district court rejected the plaintiffs’ NVRA claims. Among other things, the court held that the NVRA did not require Mississippi to disclose completely unredacted voter registration records containing information such as birthdates under the circumstances presented. At the same time, the court’s opinion in True the Vote v. Hosemann, 43 F. Supp. 3d 693 (S.D. Miss. 2014), recognized the importance of the public disclosure provision and the legitimate interest in examining records relating to the accuracy and currency of voter registration lists.

That case illustrates a tension that election administrators have to deal with: Records need to be available for legitimate oversight, but they can also contain personal information that deserves protection. The answer is not that every election record should be public without restriction. Nor should privacy concerns become a blanket reason to put election administration beyond public examination. The rules should be clear about what must be retained, what can be inspected, and what information stays protected. When records are withheld or redacted, there should be a legal basis for doing so.

We should not only keep the books. We should keep them in a way that allows for legitimate examination to validate them.

Account for Every Ballot

Once a voter is registered and checked in, the system has to account for the ballot itself.

How many ballots were issued? How many were cast? How many were spoiled or replaced? How many required duplication or adjudication? Where were the ballots stored? Who had custody of them? When did custody change? How were they transported and secured?

These are not unusual questions. Businesses keep inventory records. Hospitals document chain of custody for specimens. Financial institutions record transactions and transfers. Elections should be no different.

A gap in a chain-of-custody record does not, standing alone, establish that someone tampered with ballots. It establishes that the record is incomplete. The proper response is to document the gap, determine what happened, and determine whether the security of the ballots was compromised. A strong election system is not one in which mistakes never happen. It is one in which mistakes can be found and explained.

Routine reconciliation should be part of ordinary election administration, not something reserved for elections that become controversial. The closer the connection between voter records, ballot records, tabulation records, and final results, the easier it becomes to determine whether an apparent problem is a clerical error, a procedural failure, or something more serious.

Clear rules established before Election Day are also critical for the delivery of timely election results. Voters, candidates, campaigns, and election officials should know in advance which ballots may be counted, when those ballots must be received, what verification is required, and when the counting process ends. States can make different choices about particular deadlines and categories of ballots. What matters is that those choices be established in advance, clearly communicated, and consistently applied, and that they generate an easily documented and reconciled process.

The Supreme Court’s June 2026 decision in Watson v. Republican National Committee is a useful illustration. The court held that the federal Election Day statutes at issue did not prevent Mississippi from counting certain absentee ballots that were postmarked by Election Day but received within the following five business days. The question had divided the lower courts.

Federal law sets a single national Election Day for federal offices. Some have argued that an election is not complete until every ballot is in the hands of election officials, so a ballot that arrives after Election Day is effectively cast after it. In 2024, the Fifth Circuit agreed and held that federal law preempted Mississippi’s post-Election Day receipt window. That ruling conflicted with other courts that had upheld similar grace periods, and it put the rules in roughly 18 states that accept late-arriving mail ballots in doubt. The Supreme Court took the case to settle that conflict. The court’s analysis turned in significant part on the distinction between casting an absentee ballot and the later receipt of that ballot, and on what the federal statutes actually required.

If a state’s law permits a ballot cast by the statutory deadline to arrive later, that should be clear before voters cast their ballots. If the law establishes an Election Day receipt deadline, that should be clear too. And in making that choice, a state should bear in mind its obligation to reconcile its results. For my part, Election Day receipt deadlines are the better policy. That is not because I relish the idea of a voter’s ballot going uncounted because it arrived late. It is because every day a ballot window stays open after Election Day is another day the books cannot be closed.

Late-arriving ballots must be logged, matched against voter records, checked for postmarks that may be missing or illegible, and folded into totals that have already been reported in part. Each step is another place where the ledger can fall out of balance. Some states may manage a post-Election Day window well. In my experience, though, a longer window has meant a longer period of uncertainty, more categories of ballots to reconcile, and more opportunities for disputes about which ballots belong in the count. A clear Election Day deadline, communicated well in advance, gives voters a rule they can plan around and gives officials a fixed point from which to reconcile.

Conversely, even with an Election Day receipt deadline, everyone should be on the same page that election night does not yield the final count. Some ballots legitimately take longer to process because the law requires officials to verify eligibility, signatures, identification, or other information before those ballots can be counted. Military and overseas ballots, provisional ballots, and other categories may be governed by different deadlines and procedures depending on the jurisdiction.

History offers a useful reminder. The presidential election of 1860 was arguably the most divisive in American history. Within weeks of the result, states had begun to secede. Yet on election night, Abraham Lincoln waited at the telegraph office in Springfield, Illinois, and went home after midnight knowing he had won. He did not have to wait weeks to find out. The formal canvass took longer, of course, and the voting process of 1860 was far simpler than ours. But the lesson holds. Southern leaders rejected Lincoln’s election; they did not claim he had lost the count. When the tally is prompt and credible, disputes are about the result, not the arithmetic. If a nation whose fastest communication was the telegraph could tell its citizens who won by morning, a modern election system with modern tools should be able to complete and reconcile its count far faster than many states do today.

None of this is an argument for rushing. It is an argument for designing the system so that speed and accuracy reinforce each other. Florida is a useful example. Its election officials may begin processing and tabulating mail ballots before Election Day, and ballots generally must arrive by the close of polls. As a result, Florida typically reports nearly complete results on election night, and the state has largely moved past its 2000 reputation.

California, by contrast, accepts ballots that arrive days after Election Day and routinely takes weeks to finish counting. Large states may have their reasons for those choices. But a count that stays open for weeks leaves a long gap between the first reported numbers and the final ones, and that gap fills quickly with speculation. Finishing the count and reconciling the result as quickly as the law and accuracy allow is not just administrative convenience. It is vital to producing a credible record and preserving public confidence.

Recount When Necessary, But Always Audit

A recount and an audit perform distinct functions. A recount is generally intended to determine the number of votes reflected in the ballots for a particular race or question. An audit can serve a broader quality-control function by examining procedures or comparing a sample of ballots with the reported tabulation. An audit should not simply produce another number. It should provide evidence that the reported number is supported by the underlying records.

States use different approaches, including traditional tabulation audits, procedural audits, and risk-limiting audits. Michigan, for example, conducts several types of post-election audits, including risk-limiting audits that use randomly selected ballots for hand counting. The particular method matters less to the larger principle than whether the method is established in advance, the underlying records are secure, and discrepancies are documented and investigated.

Audits also keep procedures sharp. Any given state, county, or precinct will rarely be the place where an election is decided by a few hundred votes. But every jurisdiction should conduct routine, randomly selected audits as though it might be. Random audits show that wherever one looks, the process is working as designed. When they turn up a weakness, such as a sloppy chain-of-custody log, a poorly trained poll worker, or a reconciliation step that keeps getting skipped, officials can fix it before it matters. A business doesn’t wait for an embezzlement scandal to start auditing its books. It audits every year precisely so that a scandal is less likely to happen.

A steady stream of clean audits builds confidence. Confidence is not something an election official can create when a result turns out to be razor-thin and the accusations start flying. By then, it is too late. Confidence is built every election, regardless of the margin, so that when a close one arrives, the public already knows that procedures are in place, that they have been tested, and that they have held up. A track record of audits gives officials something better to point to than their own assurances.

Public confidence is difficult to manufacture after the fact. It is much easier to build when people can see how the system works. That does not mean every voter needs to understand the technical details of every voting machine or election management system. It does mean that the basic process should be observable and understandable. People should be able to know how voters are registered, how ballots are secured, how votes are tabulated, what audits are conducted, and what happens when something does not initially reconcile.

There will always be people who distrust election officials. Others will distrust political parties, candidates, campaigns, lawyers, voting machine companies, government agencies, or the media. Designing an election system around the assumption that everyone will simply trust the same institution is unrealistic. Especially in an era of such low trust in institutions, and such frequent failure, distrust of a system that is both highly opaque and highly consequential is entirely natural. Mocking or dismissing the distrust only deepens it.

In Federalist No. 51, James Madison wrote: “If men were angels, no government would be necessary.” Likewise, when it comes to elections, no audit would be necessary. In our imperfect world, though, the better approach is to make trust less dependent on personalities and institutions, and more dependent on structure and records.

Rather than “trust us,” the answer should be: Here are the rules. Here are the records. Here is how the ballots were accounted for. Here is how the result was checked. Here is what happened when something did not initially reconcile. That kind of transparency does not require unlimited public access to sensitive information. It requires enough access to allow legitimate observers to understand and verify the process.

One concrete step would be state laws requiring that election officials publish reconciliation reports after each election, documenting the number of registered voters, the number of ballots cast, the number of ballots adjudicated or duplicated, and any discrepancies identified during canvassing. Several states already require some version of this reporting, but making it a uniform, publicly available standard would give voters, candidates, and the courts a common set of books to examine.

In all these tasks, technology can be a double-edged sword. On one hand, technology can make election administration faster and more accurate, providing tools for maintaining and updating records, counting ballots and summing totals, and reconciling numbers. On the other hand, in many cases, election officials have allowed technology to push toward processes that look “easier” for the voter or the administrator but weaken the institutional integrity of the system.

Most importantly, technology does not eliminate the need for records. It should create an audit trail, not eliminate one, and always provide a paper record against which an electronic tabulation can be checked. Of course, the paper record itself must be preserved, secured, connected to the appropriate election records, and available for whatever audit or recount process the law provides.

If the only answer to a question about an election result is that a computer produced the number, the system has not provided enough information for independent verification. The public does not need to understand every line of code. But it does need to know that there is an underlying record against which the reported result will reconcile.

The good news is that paper records remain common. The U.S. Election Assistance Commission’s 2024 Election Administration and Voting Survey reported that more than 98% of election jurisdictions used equipment that requires the voter to mark a paper ballot or that produces an auditable paper record.

Confidence, Not Faith

An inaccurate registration record is not automatically an illegal vote. A discrepancy between ballots and voter records is not automatically evidence of tampering. A gap in a chain-of-custody log does not, by itself, prove that anyone manipulated ballots.

Rather than begin with an assumption of either innocence or misconduct, the better approach is to ask what happened, what records document it, what caused the discrepancy, whether it can be explained, whether the ballot or election record remained secure, and whether the problem affected the result.

That is how a good accounting system works. It does not begin by assuming that every employee is dishonest. But it also does not assume that every transaction is correct simply because someone entered it into a ledger. It creates controls that make mistakes and misconduct discoverable.

Americans do not all have to agree about election policy. They do not have to trust the same institutions. They do not even have to agree about how much of the election process should be handled by federal versus state versus local governments. But they can and should all agree that the process is clearly stated, that it can be examined, and that its outcomes are thus legitimate.

We should build an election system in which the public does not have to take anyone’s word for the result. The records should tell the story, and the books should balance.

Trey Trainor is a partner at Dhillon Law Group, Inc. and a former commissioner on the Federal Election Commission, nominated by President Donald Trump and confirmed by the U.S. Senate in 2020.