Even a well-run election can leave a meaningful share of the public unconvinced. That’s not a new phenomenon, and it’s not necessarily evidence that anything went wrong — close, high-stakes contests tend to produce genuine disagreement about outcomes regardless of how carefully the count was handled. The harder question isn’t how to prevent all disputes from happening. It’s what officials, and the rest of us, can reasonably do afterward to rebuild confidence once one has.

Recent survey data gives a sense of how real this challenge has become. According to a PBS News, NPR, and Marist poll, confidence that elections are run fairly has dropped to its lowest point in years, with meaningful shifts across party lines rather than concerns concentrated on one side. That kind of broad, bipartisan erosion in trust is, we think, a more serious problem than any single contested race — it suggests the issue isn’t really about any one outcome, but about the underlying credibility of the process itself.

We don’t think confidence gets rebuilt through messaging alone. Reassurances from officials, however well-intentioned, tend to carry less weight than verifiable evidence people can check for themselves. This is where the tools we’ve spent time discussing elsewhere — paper records, post-election audits, transparent chain-of-custody procedures, open observation — earn their keep. They’re not primarily preventive measures; they’re also the raw material officials can point to after a disputed result, to demonstrate rather than simply assert that the count was accurate.

Speed matters here too, in a way that’s easy to underestimate. A credible explanation offered quickly, before rumors and speculation have time to harden into settled belief, tends to land very differently than the same explanation offered weeks later after a narrative has already taken hold. Election offices that have pre-built plans for public communication after a close or contested result — rather than improvising one under pressure — tend to fare better at containing disputes before they escalate.

We’d also gently note that rebuilding confidence is not the same thing as demanding that everyone agree the result was correct. Reasonable people can review the same audit and still disagree about a political outcome. What verifiable process can realistically offer is a shared set of facts to disagree about, rather than a dispute over the facts themselves. That’s a meaningfully lower bar than universal agreement, and one we think is actually achievable.

None of this happens automatically. It requires investment in the underlying infrastructure — the audits, the paper trails, the trained staff — well before a contested election ever occurs, so that when confidence is tested, there’s something concrete to point to. Waiting until after a disputed outcome to start building that record is, by definition, too late.

Voter rolls have a thankless job. When they’re accurate, nobody notices. When they’re not — when a voter shows up to find their registration missing, or when a list turns out to include people who’ve moved away or passed on — the resulting frustration tends to land squarely on election officials, even when the underlying cause is something as mundane as outdated address data.

Maintaining an accurate voter roll means running two goals that can pull in opposite directions at the same time: removing entries that no longer belong (people who have died, moved out of the jurisdiction, or are otherwise ineligible) while making sure eligible voters aren’t mistakenly purged along the way. Lean too far toward caution and the rolls fill up with outdated entries. Lean too far toward aggressive cleanup and you risk disenfranchising someone who’s still very much eligible to vote.

Most states rely on a mix of tools to manage this balance: matching voter data against death records and change-of-address filings, cross-checking against rolls in other states through information-sharing programs, and sending mailers to inactive voters before removing them, so there’s a chance to confirm eligibility before anyone is struck from the list. None of these methods are perfect on their own, which is part of why officials generally use several in combination rather than relying on a single data source.

We think transparency plays as large a role here as it does in the ballot-counting process itself. When roll-maintenance procedures are published, when the criteria for removal are clear and consistently applied, and when the public — including the kind of election observers we’ve written about before — can review how the process actually works, disputes tend to focus on real questions rather than vague suspicion. Opaque processes, even accurate ones, invite more distrust than transparent processes that occasionally make a correctable mistake.

Timing matters too. Roll maintenance conducted well before an election, with adequate notice and appeal periods, looks very different from last-minute changes close to voting day, even if the underlying justification is identical. We’d argue that predictable, well-publicized maintenance schedules — rather than irregular, reactive purges — do more to build confidence in the accuracy of the rolls than any single technical improvement could.

None of this is a simple problem with an obvious fix. Voter rolls will always be playing catch-up with a moving population, and any maintenance system will occasionally get an individual case wrong in one direction or the other. What we can reasonably ask for is a process that’s consistent, well-documented, and open to correction — one where errors get fixed quickly and the criteria for making changes in the first place are available for anyone to examine.

One of the simplest tools for building confidence in an election doesn’t involve any technology at all: letting people watch. Election observers — whether representing political parties, nonpartisan civic groups, or in some cases international organizations — have been a fixture of American elections for well over a century, and their basic purpose hasn’t changed much in that time.

Observers are there to watch, not to intervene. They can note whether procedures are being followed, raise questions with election officials through proper channels, and, in most jurisdictions, file formal challenges if they believe something was handled incorrectly. What they generally cannot do is touch ballots, interfere with voters, or disrupt the process — the rules governing observer conduct are typically as detailed as the rules governing the election itself. Legal scholars have documented how these roles and their legal boundaries have evolved over time, including their expanded use during the civil rights era to help guard against discriminatory practices at the polls.

We think the underlying value of observation is fairly intuitive: it’s much harder to credibly claim that something improper happened behind closed doors when representatives from multiple interests were standing in the room the whole time. That’s true whether the “room” is a polling place on election day or a counting facility processing ballots afterward. Transparency doesn’t just catch problems — more often, its real function is preventing disputes from ever gaining traction in the first place, because there were witnesses to what actually happened.

That said, observation only builds confidence if it’s genuinely open to a balanced range of participants, not just one side of a political contest. Lopsided access — where only one party’s observers are meaningfully present — can end up doing the opposite of its intended purpose, creating the appearance of oversight without the substance of it. We’d argue that election offices have a real interest in actively recruiting observers from multiple perspectives, rather than simply leaving the door open and hoping for balance to happen on its own.

There’s also a training question here, not unlike the one we’ve raised about poll workers generally: an observer who doesn’t understand the procedures they’re watching can misread a routine step as a problem, or miss an actual irregularity because they didn’t know what they were looking at. The most effective observer programs tend to pair access with a real orientation on what normal procedure looks like, so that observation produces useful, credible feedback rather than just noise.

Ultimately, we see observers as a low-cost, high-value complement to the technical safeguards — audits, paper trails, chain-of-custody logs — that get more attention in security conversations. Letting people see the process for themselves remains one of the most direct ways to answer the question “how do we know this was done right?”

Modern voting equipment sits at an odd intersection: it has to be simple enough for a nervous first-time voter to use without instructions, and secure enough to withstand scrutiny from security researchers actively looking for weaknesses. That’s a genuinely hard design problem, and it’s worth understanding a little about how the systems actually work before drawing conclusions about how safe — or unsafe — they are.

The first thing worth knowing is that the vast majority of voting machines in use today are not connected to the internet while polls are open. That single fact surprises a lot of people, since we’re used to thinking of “hacking” as something that happens remotely, over a network connection. Election officials generally treat air-gapping — physically isolating voting equipment from any outside network — as a baseline requirement precisely because it closes off the most common attack vector used against ordinary computers.

That doesn’t mean the machines are risk-free. Security researchers have identified vulnerabilities in various voting systems over the years, usually through controlled testing environments rather than real-world attacks. Finding a flaw in a lab, however, is a different thing from that flaw being exploitable during an actual election, given the physical security, chain-of-custody procedures, and pre- and post-election testing that surround the equipment in practice. We’d argue this is a point that gets lost in a lot of public conversation: a theoretical vulnerability and a practical, exploitable attack path are not the same thing, and treating them as identical tends to generate more fear than clarity.

This is also where the physical safeguards we’ve discussed elsewhere — sealed transport, logged custody, bipartisan witnesses — do a lot of quiet, unglamorous work. A machine that’s been properly secured and monitored the entire time it’s in use is a much harder target than the same machine sitting unattended. Cybersecurity and physical security aren’t separate conversations; for voting equipment, they’re really the same conversation viewed from two angles.

Independent testing matters too. Most jurisdictions require voting systems to be certified against federal or state security standards before they’re ever deployed, and many states also conduct their own pre-election logic and accuracy testing on every machine that will be used. None of this is a guarantee against every possible failure mode, but it does mean the systems aren’t simply taken on faith — they go through a testing pipeline before a single vote is cast.

We think the most useful posture for voters isn’t blind trust or blanket suspicion, but informed skepticism: understanding that these systems are tested, monitored, and — critically — backed up by paper records that let officials verify results independently of the software itself. That combination of layered defenses is a far more realistic picture of voting machine security than either the “totally safe” or “totally rigged” narratives that tend to dominate the loudest parts of the conversation.