13 — Philosophy
This section is the closest the manual gets to America Counts’s normative voice. Most of the manual is descriptive: this is what the system is, this is how it works, this is what AC has done. This section steps back and asks why — why this approach rather than another, what principles are at work, what AC believes about the relationship between citizens and elections, and how AC thinks about the broader landscape of election integrity work.
The voice here is more reflective than the rest of the manual but remains measured. AC’s philosophy is expressed through four named principles — Transparency, Non-Partisanship, Independence, and Humility — each of which has operational consequences that appear throughout other sections of the manual. The fifth subsection is new to this edition: it positions Actual Vote honestly within the broader election integrity landscape, mapping AC’s methodology against other tools and organizations, identifying where AV is uniquely valuable and where its limitations leave work for others.
A reader who wants to understand the philosophical underpinnings of AC’s work should read all five subsections. A reader who is skeptical of AC’s claims can look directly at Section 13.3 (the Independence Principle, which develops the structural case for why AC’s evidence can be trusted) and Section 13.4 (the Humility Principle, which articulates what AC’s methodology cannot do). A reader who wants to understand how AV fits alongside other election integrity efforts should read Section 13.5.
13.1 — The Transparency Principle
Making evidence publicly available
The first principle underlying America Counts’s work is transparency: the commitment to making election verification evidence publicly available, in forms that anyone can inspect, and through processes that are themselves open to scrutiny.
Transparency in the AC context has three distinct dimensions. The first is transparency of evidence: the recordings, transcriptions, comparison datasets, and investigation records that constitute an AV analysis should be accessible to anyone who wants to examine them. This is not the same as dumping raw data on the internet — the evidence must be organized, documented, and presented in ways that make it possible for a reader to follow the chain from a poll tape video to a comparison finding. But the organizing principle is openness. The evidence exists to be examined, not to be held in confidence.
The second dimension is transparency of methodology. AC’s analytical process — the nine-step comparison analysis pipeline documented in Section 8, the seven-check investigation process documented in Section 8.9, the submission and vetting standards documented in Section 8.5 — is publicly documented and replicable. A reader who disagrees with an AC finding can follow the same steps with the same data and check whether they arrive at the same conclusion. This replicability is not an accident of documentation; it is a design requirement. The comparison pipeline was built to be auditable because AC believes that an unauditable audit is a contradiction in terms.
The third dimension is transparency about the organization itself. AC’s structure (a project of Democracy Counts, Inc., a 501(c)(3) non-profit), its team, its funding model (non-partisan donations; no payments from candidates, campaigns, or parties; no payments that would bias findings), and its institutional relationships are disclosed. When AC makes mistakes — a published finding that turns out to need correction, an operational issue that affected an analysis — the record is corrected openly rather than buried.
The archive as a permanent public resource
One consequence of the transparency principle is that AC’s evidence accumulates into a permanent public archive. Every AV analysis produces a body of evidence — video recordings, transcribed data, official results downloads, comparison outputs, investigation notes — that is preserved and accessible. Over time, this archive becomes a historical resource: a growing record of what vote reporting looked like in specific jurisdictions across specific elections, documented by independent observers using a consistent methodology.
The archive serves multiple functions. In the near term, it supports the credibility of individual analyses: a reader can go back to the underlying evidence and verify any finding. In the medium term, it supports pattern recognition across elections: trends in posting compliance, recurring procedural issues, or systematic discrepancies become visible only when the data accumulates. In the long term, it serves the public interest: future researchers, journalists, advocates, and election officials will have access to a body of evidence about vote reporting that did not previously exist.
The archive also has a deterrent function, discussed in more detail in Section 13.5. When election officials know that independent observers are recording and preserving evidence of what poll tapes say, the incentive structure shifts. Errors that might have gone unnoticed become detectable. Procedural shortcuts that might have been taken become riskier. The very existence of a permanent, publicly accessible evidence archive changes the calculation for anyone involved in the vote reporting process.
Transparency about methodology, not just findings
A subtle but important aspect of the transparency principle is that it extends to methodology, not just to findings. Many organizations publish findings — “we investigated this election and here’s what we found” — without fully disclosing how they got there. The audience is asked to trust the organization’s competence and good faith based on the organization’s reputation.
AC takes a different approach. The methodology is documented in enough detail that a competent reader can evaluate not just whether AC’s findings seem reasonable, but whether AC’s process for arriving at those findings was sound. This is the difference between “trust us, we’re careful” and “here is exactly how we did it; judge for yourself whether it was careful.”
The distinction matters because election integrity is a domain where trust is in short supply. Asking an audience to take an organization’s findings on faith is asking for a kind of trust that many readers, reasonably, do not extend. Showing the work — every step of it, from how recordings were collected to how apparent discrepancies were investigated to how findings were communicated — gives readers a basis for making their own judgment. Some readers will examine the methodology closely and find it rigorous. Others will take the fact that the methodology is disclosed as evidence that AC has nothing to hide. Either way, the transparency serves the credibility of the work.
This methodological transparency also invites improvement. When the methodology is documented in detail, readers who see a weakness can identify it specifically and suggest improvements. A closed methodology improves only when its practitioners notice problems. An open methodology improves whenever anyone notices a problem. AC’s commitment to transparency is, among other things, a commitment to getting better over time.
13.2 — The Non-Partisanship Principle
What non-partisanship means at AC
America Counts is non-partisan. This is a precise commitment with specific implications, not a vague positioning statement.
Non-partisanship at AC means that the organization does not endorse candidates in any election at any level. It does not endorse or affiliate with any political party, formally or informally. It does not advocate for specific policy positions beyond the broad position that vote reporting should be accurate and that independent verification of vote reporting is valuable — positions that are supported by people across the political spectrum and are not aligned with any particular partisan agenda. It does not coordinate its findings or its operations with any campaign, party, or politically-aligned organization in ways that would make AC’s findings appear motivated by partisan interest. It does not accept donations from candidates, campaigns, parties, or partisan political action committees. Its findings are not aligned with any political narrative. AC reports what it finds, regardless of whether the findings support or undermine any particular political position.
AC’s team members may hold personal political views — they are citizens who vote and have opinions — but they do not let those views influence their work for AC. The institutional culture reinforces this expectation through methodological consistency (the same pipeline runs the same way regardless of which candidates are involved), tone discipline (the same measured language regardless of the political stakes), and communications discipline (findings and methodology are discussed, not political implications).
Why non-partisanship is structurally enforced
AC’s non-partisanship is not merely a stated policy that requires trust. It is structurally enforced by the design of the system itself.
The first structural enforcement is that anyone can audit. AV is not a tool restricted to AC-approved operators. Anyone can download the app, record a poll tape, and submit evidence. This means that any attempt by AC to suppress inconvenient findings would be detectable: other users with recordings from the same precincts could identify the gap. The openness of participation makes selective suppression impractical.
The second structural enforcement is that the evidence is public. AC’s recordings, transcriptions, and comparison datasets are preserved and accessible. A finding of “no discrepancies” for a given precinct can be verified by anyone with access to the same poll tape video and the same official results. A finding cannot be fabricated or suppressed without contradicting publicly available evidence.
The third structural enforcement is that the methodology is transparent. The comparison pipeline is documented. The investigation protocol is documented. The criteria governing how initial non-matches are triaged and how discrepancies move through comparison outcomes are documented. A motivated critic who believed AC was applying its methodology inconsistently — more aggressively when one party’s candidates were involved, less aggressively when another’s were — could examine the documented methodology and the actual analysis records to test the accusation.
These structural features do not make partisanship impossible, but they make it detectable. An organization that relies on secrecy to maintain the appearance of non-partisanship is vulnerable to exposure. An organization whose evidence is public, whose methodology is documented, and whose tools are open to all participants is harder to corrupt — and if it were corrupted, the corruption would be easier to identify.
How AC maintains non-partisan positioning under political pressure
In a polarized political environment, the question of whether election results can be trusted is itself politically charged. Different political camps have different default assumptions about the reliability of elections. Some lean toward the view that elections are generally accurate and that questioning results is irresponsible. Others lean toward the view that elections are suspect and that skepticism is warranted. Each camp has its preferred sources of information and tends to dismiss sources associated with the other camp.
AC’s work exists in this environment and cannot escape it. The strategy for navigating it has several components.
First, methodological consistency. AC applies the same methodology to every analysis, in every jurisdiction, regardless of which candidates or parties are involved. The pipeline does not run differently when the political stakes change. Consistency over time is the primary defense against accusations of bias: a track record of identical treatment across elections involving candidates of all parties is the most persuasive evidence of non-partisanship.
Second, tone discipline. AC’s reports, communications, and public-facing materials use measured language that avoids coded political signaling. Words and phrases that have become associated with one political camp are avoided where possible. The goal is to be a source that readers from all political perspectives can take seriously, even if they disagree with each other about everything else.
Third, the “both sides accuse us” test. A useful sign that AC is maintaining its non-partisan stance is that it sometimes gets accused of bias from both directions simultaneously. Critics on one side perceive AC’s clean findings as covering for fraud they assume must exist. Critics on the other side perceive AC’s existence as legitimating fraud claims by suggesting that elections need extra scrutiny. Neither critique is accurate, but the fact that both exist is evidence that AC is not captured by either camp. When AC starts getting accused of bias from only one direction, that is a warning sign that the organization may have drifted.
Fourth, the non-partisan commitment is documented in AC’s organizational materials — the About page, the Privacy Policy, the User Agreement — and is enforced through donor screening, volunteer guidance, and partner selection. Donors are screened for partisan affiliations. Volunteers are reminded that their work for AC is non-partisan regardless of their personal views. Partner organizations are selected for their own non-partisan positioning or for a clear separation between their partisan activities and their election integrity work.
What non-partisanship is not
Non-partisanship is sometimes confused with neutrality on questions where AC does have a position. The distinction is important.
AC is not neutral on whether election integrity matters. AC believes it matters profoundly — it is the foundational premise of the work. AC is not neutral on whether vote reporting should be accurate. AC believes it should be, and the work is dedicated to verifying that it is. AC is not neutral on methodology. AC has specific views about what constitutes rigorous verification and what does not. AC is not neutral on the value of independent verification. AC believes it is valuable and has built an organization around providing it.
What AC is neutral on is the political identity of the people whose vote reporting is being verified. A discrepancy is a discrepancy regardless of who benefits from it. A clean finding is a clean finding regardless of which candidate won. The methodology runs the same way and the findings are reported the same way regardless of the partisan context. AC cares about the accuracy of the process; AC does not care about the political identity of the beneficiary.
This distinction — non-partisan about actors but not neutral about principles — is the philosophical core of AC’s positioning. It is what allows AC to be mission-driven (election integrity matters) without being politically captured (by any party or candidate).
13.3 — The Independence Principle
Operating independently from elections offices
The third principle underlying AC’s work is independence: AC operates independently from the elections offices whose vote reporting it verifies. AC is not a contractor hired by elections offices. AC is not a government agency with oversight authority. AC is not a vendor supplying election services. AC is an independent non-profit that chose this work because it believes the work needs to be done, and that does the work whether elections offices want it done or not.
The independence is essential because of a structural problem in how US elections are administered. Elections in the United States are run by partisan officials — county elections supervisors, secretaries of state, and other administrators who are either elected as party candidates or appointed by elected partisan officials. This does not mean that elections officials are corrupt. Most are conscientious professionals doing difficult work with limited resources and under enormous political pressure. But it does mean that every elections office in the country operates within a partisan system, subject to partisan incentives, with no structural separation between the people who run elections and the people who benefit from their outcomes.
In nearly every other domain where the stakes are comparably high, the absence of independent verification would be considered unacceptable. Publicly traded companies do not audit their own financial statements — independent auditing firms do, and the requirement for independent auditing is enforced by law and by market expectation. Pharmaceutical companies do not certify their own drugs — independent regulatory agencies do, with independent researchers providing peer review. Airlines do not certify their own aircraft — independent agencies do, with independent investigators analyzing failures. The pattern is consistent: when the stakes are high and the entity producing the results has an interest in those results, independent verification is considered necessary.
The SEC audit analogy
The closest structural analogy to what AC does is the independent financial audit. Publicly traded companies are required by the Securities and Exchange Commission to have their financial statements audited annually by independent accounting firms. The auditors examine the company’s books, test a sample of transactions, verify that the reported numbers agree with the underlying records, and issue an opinion on whether the financial statements present a fair picture of the company’s financial position.
The analogy to AC’s work is direct. The company’s financial statements are like the official election results: a set of reported numbers produced by the entity that has primary responsibility for the activity. The underlying records are like the poll tapes: the most direct available evidence of what actually happened at the transactional level. The independent audit is like the AV analysis: an independent party comparing the reported numbers to the underlying evidence to check whether they agree.
The analogy is instructive in several ways.
First, nobody considers financial self-auditing adequate. The reason publicly traded companies are required to have independent audits is that the companies have an incentive to present their financials in the most favorable light. Even without fraudulent intent, institutional biases, systemic blind spots, and honest mistakes make self-auditing unreliable. The same reasoning applies to elections: elections offices have institutional biases and blind spots, even when (especially when) they are staffed by honest professionals doing their best.
Second, independent financial audits do not replace the company’s own accounting. The company still keeps its own books, produces its own financial statements, and does its own internal controls. The independent audit is a supplement, not a substitute. Similarly, AC does not replace elections offices’ own verification processes. Elections offices still run their own logic and accuracy testing, their own post-election audits (where required), and their own internal quality controls. AC supplements that work with an independent external check.
Third, the existence of independent financial auditing raises the quality of the underlying accounting. Companies that know they will be independently audited maintain better records, follow better procedures, and catch more of their own errors before the auditors arrive. The deterrent effect is as valuable as the detective effect. The same dynamic applies to vote reporting: the knowledge that independent observers are recording poll tapes and comparing them to official results creates an incentive for elections offices to maintain accurate reporting, even apart from whatever errors AC’s analyses might find.
Fourth, independent auditors sometimes find nothing wrong. A clean audit opinion does not mean the audit was pointless — it means the financial statements are confirmed to be reliable by an independent party, which is information that investors and regulators need. Similarly, when an AV analysis finds no discrepancies, the analysis is not pointless — it confirms that vote reporting was accurate for the analyzed precincts, which is information that voters and the public need.
Not substituting for elections offices — supplementing them
The independence principle does not position AC as adversarial to elections offices. The relationship is, in the ideal case, complementary. Elections offices do the vast majority of the work of running elections — voter registration, ballot design, equipment testing, staffing polling places, counting votes, reporting results, conducting internal audits, certifying outcomes. This is difficult, resource-constrained work performed under intense scrutiny and political pressure. AC acknowledges the difficulty and respects the work.
What AC adds is a specific external check on one part of that work: the reporting layer. AC takes the elections office’s reported results and checks them against the most direct independent evidence available — the poll tapes printed by the voting machines. When the check comes back clean, both the elections office and the public benefit: the office’s work is independently confirmed, and the public has additional evidence that the reporting was accurate. When the check reveals a discrepancy, both parties still benefit: the office learns about a problem it may not have caught internally, and the public learns that the verification system is working.
The complementary framing is important because it sets the tone for AC’s interactions with elections offices. AC approaches offices with respect, communicates findings through proper channels (documented in Section 10.8), and treats every contact with the elections office as a cooperative ask — never calling it escalation; that word is reserved for handing an unresolved matter to a court-capable party (documented in Section 6.5). AC is not trying to embarrass elections offices or to replace them. AC is trying to provide a service that elections offices and the public both benefit from — the same service that independent financial auditors provide to publicly traded companies and their shareholders.
Some elections offices welcome the independent check. Others resist it. AC builds relationships over time with offices that are open to the work, and exercises its legal rights (documented in Section 6) to do the work even when offices are not cooperative. The independence principle means that AC’s ability to do its work does not depend on the cooperation of the entity being verified — just as the SEC does not require a company’s permission to mandate an independent audit.
The structural case for trust
The independence principle connects directly to the question every reader of an AC report implicitly asks: why should I trust these findings?
The answer is structural, not personal. It does not depend on the character of any individual analyst or the reputation of any particular leader. The structure of the system produces trustworthy results because of several interlocking features.
AC uses the most direct available evidence. The poll tape is the closest public artifact to the voting machine’s actual tabulation. Using it as the comparison baseline minimizes the distance between the evidence and the underlying reality.
AC’s methodology is replicable. The comparison pipeline is documented, the data is preserved, and the findings can be reproduced by anyone who follows the same steps with the same data. A finding that cannot be replicated is not a finding; AC’s findings can be replicated.
AC operates in the open. The methodology, the evidence, the organizational structure, the funding model, and the findings are all disclosed. An organization that operates in the open is harder to corrupt than one that operates in confidence.
AC has no financial incentive that would bias its findings. No one at AC is paid based on the outcome of an analysis. No donor gets to specify what AC must find. No contractor has a bonus tied to discovering fraud or to confirming clean elections. The incentives are structured to reward accuracy, not any particular result.
AC’s findings are falsifiable. They make specific claims — that a given number of comparisons were performed, that a given number matched, that a given number of likely attributable and apparent discrepancies were found and how each resolved — that can in principle be shown to be wrong. An organization that makes falsifiable claims and invites scrutiny of those claims is one that has confidence in its own work.
AC’s track record demonstrates the methodology in action. The body of completed analyses — from the early Broward County work through the NC 2024 primary and the 2024 general election multi-state operation — provides empirical evidence that the methodology works in real conditions, across multiple jurisdictions and election types. The track record is the proof of concept.
These structural features work together. An organization that uses good evidence but operates secretly would be less trustworthy. An organization that operates openly but uses derivative evidence would be less informative. An organization that has the right incentives but no track record would be unproven. The combination of all these features — direct evidence, replicable methodology, open operation, aligned incentives, falsifiable claims, and demonstrated track record — is what makes the case for trusting AC’s findings.
13.4 — The Humility Principle
Not claiming to resolve every question
The fourth principle underlying AC’s work is humility: the commitment to being explicit about what the methodology can and cannot prove, to flagging uncertainty when it exists, and to respecting the limits of what AV’s evidence supports.
Humility in this context is not modesty for its own sake. It is the recognition that overclaiming — presenting partial verification as complete validation, or treating the absence of detected problems as proof that no problems exist — is the fastest way to destroy the credibility of election integrity work. The election integrity space is littered with organizations that made claims their evidence could not support, and whose credibility was destroyed when the overclaiming was exposed. AC’s humility principle is a deliberate effort to avoid that fate.
The humility principle operates at several levels. At the level of individual findings, it means reporting precisely what was verified and what was not. At the level of methodology, it means being explicit about the categories of error and fraud that AV can detect and the categories it cannot. At the level of the organization’s public voice, it means using measured language that does not overpromise.
The matched poll tape caveat
The most important expression of the humility principle is the matched poll tape caveat: the standing acknowledgment that an Actual Vote finding of agreement between poll tapes and official results does not, by itself, prove that the underlying ballots were correctly counted by the voting machines.
The caveat is important enough that it is restated formally here, though it has appeared throughout this manual wherever the scope of AV’s methodology is discussed.
AV verifies vote reporting — the process by which the numbers printed on poll tapes at polling places are transmitted, aggregated, and published as official results. AV does not verify vote counting — the process by which voting machines translate ballot marks into the numbers that appear on poll tapes. These are different processes with different vulnerabilities, and verifying one does not verify the other.
In concrete terms: if a voting machine miscounts ballots — recording fewer votes for one candidate and more for another — the miscounted totals will be printed on the poll tape. Those miscounted totals will then flow through the reporting process and appear in the official results. An AV analysis comparing the poll tape to the official results would find an exact match, because both reflect the same (incorrect) upstream source. AV would report a clean comparison. The counting error would be undetected.
The same logic applies — and more forcefully — to deliberate fraud at the counting layer. Consider what might be called the matched conspiracy scenario: a malicious actor compromises both a voting machine and the elections office’s reporting system. The compromised machine produces fraudulent counts favoring one candidate; the compromised reporting system records those counts as if they were legitimate. The conspiracy is internally consistent — the poll tape and the official results agree because both reflect the same fraudulent intent. An AV analysis would find an exact match. The fraud would be completely invisible to AV’s methodology because it occurred upstream of the evidence AV uses.
The matched conspiracy scenario is a stronger threat than the simple error scenario because it involves deliberate coordination rather than accidental consistency. It is also less likely, because coordinated fraud at this scale would require multiple people in different roles and organizations to cooperate and to maintain secrecy. But it is conceptually possible, and AC’s methodology cannot rule it out. AC acknowledges this explicitly: there exists a class of attack — one that infects both the machine’s counting and the machine’s tape printing identically — that AV cannot detect by design.
What the caveat does not mean
The matched poll tape caveat is sometimes overinterpreted in ways that do not follow from what AC actually claims.
The caveat does not mean AV’s methodology is useless. AV catches a wide range of errors and frauds — anything that affects the reporting layer independently of the counting layer. This includes transcription errors at any step of the aggregation process, errors in the official results database, errors in the publishing infrastructure, errors in the human handling of poll tapes, and any kind of fraud that involves changing reported totals after the machines produce them. These are real categories of risk. The Case Studies section documents multiple instances where exactly these kinds of reporting-layer problems occurred or would have been caught by AV.
The caveat does not mean elections are insecure. The fact that AV cannot detect counting-layer problems does not mean counting-layer problems are common or undetectable by other means. Counting verification is a different methodology — risk-limiting audits, hand counts of paper ballots, and other techniques exist for verifying counting independently. AV is one tool among several; each addresses different vulnerabilities.
The caveat does not mean clean AV findings are meaningless. A clean AV finding — no surviving discrepancies after thorough investigation — is meaningful evidence about the reporting layer. It tells the reader that vote reporting was accurate for the analyzed precincts and contests, within the scope of the comparisons performed. That is a real, substantive contribution to election confidence even though it does not extend to the counting layer.
The caveat does not mean AC should be less transparent about findings. On the contrary: the caveat is itself an act of transparency. By being explicit about what its methodology does and does not cover, AC gives readers the information they need to calibrate their interpretation of any finding.
The caveat as a standing commitment to intellectual honesty
The matched poll tape caveat is, at its philosophical core, a commitment to intellectual honesty. It says: here is what we can prove, and here is what we cannot prove, and we will never conflate the two.
This commitment is unusual in the election integrity space, where overclaiming is common and intellectual honesty is rare. Organizations that want to convince an audience that elections are fraudulent tend to present partial evidence as if it were conclusive. Organizations that want to convince an audience that elections are secure tend to dismiss any concern about accuracy as if the very question were irresponsible. AC is trying to be a third thing: an organization that presents what it has found, precisely, with appropriate hedging, and trusts its audience to draw appropriate conclusions.
The recommended treatment in any AC report follows a four-part pattern. First, state the finding precisely: how many comparisons were performed, how many matched, how many likely attributable and apparent discrepancies were identified, and how each resolved. Second, state what the finding extends to: the agreement between poll tapes and official results for the specific precincts, contests, and voting methods that were analyzed. Third, state what the finding does not extend to: the accuracy of the underlying ballot counting, the security of the voting equipment, the eligibility of voters, or any aspect of the election system other than the agreement between tapes and results. Fourth, direct readers interested in those other questions to the tools and methodologies that address them.
This four-part treatment is not optional. It is the minimum standard for honest presentation of AV findings. Omitting the third and fourth parts — what the finding does not cover and where to look for coverage — would be a violation of the humility principle.
The paradox is that the matched poll tape caveat, which might seem like a methodological weakness, is actually a strength of AC’s framing. Organizations that overclaim their methodology undermine their own credibility when the overclaiming is exposed. AC’s explicit acknowledgment of what its methodology cannot do protects AC from that fate. A reader who knows what AC does and does not claim can engage with AC’s findings with appropriate calibration. A reader who had been told that AC verifies “the entire election” would feel deceived when they discovered the actual scope. Honesty about limits builds trust that survives close inspection.
The caveat is also an invitation to complementary work. By being explicit about what AV does not do, AC implicitly invites other organizations and methodologies to fill the gaps. The election integrity space is bigger than any single methodology. AC contributes one piece of the verification puzzle and welcomes others who contribute other pieces.
13.5 — AV in the Broader Election Integrity Landscape
This subsection positions Actual Vote within the broader landscape of election integrity tools, organizations, and methodologies. The goal is an honest mapping: where AV is uniquely valuable, where its limitations leave work for others, and how it relates to the other approaches that a reader may already know about.
Honest mapping of AV against other tools and organizations
The election integrity landscape in the United States includes many different approaches, each addressing different aspects of the election process with different methods and different tradeoffs. AV occupies a specific position in this landscape. Understanding that position requires understanding the alternatives.
Risk-Limiting Audits (RLAs). Risk-limiting audits are the gold standard statistical methodology for verifying vote counting. An RLA takes a random sample of paper ballots, hand-counts them, and compares the hand count to the machine tabulation. If the sample confirms the machine result with sufficient statistical confidence, the audit is complete. If not, the sample is expanded until either the machine result is confirmed or a full hand count is triggered.
RLAs are statistically rigorous, well-studied, and increasingly adopted by states and counties. Colorado was an early adopter; many other jurisdictions have followed. They address exactly the question that AV cannot: whether the voting machines correctly translated ballot marks into tabulated numbers.
The key limitation of RLAs from AV’s perspective is that they require election authority cooperation. RLAs are conducted by elections offices, sometimes with academic partners. They require access to the physical ballots, which are controlled by the elections office. An independent citizen cannot conduct an RLA because an independent citizen cannot access the ballots.
The relationship between AV and RLAs is complementary, not competitive. RLAs verify vote counting; AV verifies vote reporting. These are different processes with different vulnerabilities. A jurisdiction that conducts both an RLA and an AV analysis has a more complete verification picture than one that conducts only one. The RLA confirms that the machines counted correctly; the AV analysis confirms that the counted results were reported correctly. Together, they cover both layers of the process.
Ballot image audits. Some voting systems produce digital images of each ballot as it is scanned. Ballot image audits examine these images, either by hand review or by independent software analysis, to verify that the machine correctly interpreted each ballot. Ballot image audits can catch counting errors that even poll tape verification would miss, because they examine the raw evidence that precedes the tabulation step.
Ballot image audits are powerful but face a significant access barrier: most jurisdictions do not make ballot images publicly available. Some jurisdictions do not even retain them. Where ballot images are available, they are typically obtained through public records requests or litigation, which is time-consuming and uncertain. AV, by contrast, uses poll tapes — a public artifact that is either physically posted at the polling place or obtainable through the access methods documented in Section 7. The access barrier is much lower.
The relationship is again complementary. Ballot image audits can catch counting errors; AV can catch reporting errors. Neither substitutes for the other. In jurisdictions where ballot images are available, the combination of ballot image analysis and AV analysis would provide an unusually comprehensive verification.
Verified Voting. Verified Voting is a national advocacy and research organization focused on voting equipment standards, election administration practices, and election security policy. It maintains the Verifier database of voting equipment used in each jurisdiction, publishes research on election administration, and advocates for specific reforms (paper ballots, post-election audits, improved security standards).
Verified Voting’s focus is broader and more policy-oriented than AV’s. It works at the level of systems and standards rather than at the level of individual election verification. AC and Verified Voting are complementary: Verified Voting provides the research and advocacy that shapes the environment in which elections are conducted, while AV provides operational verification of specific election outcomes within that environment.
Scrutineers and poll observer networks. There is an existing infrastructure of election observation in the United States, ranging from partisan poll watchers appointed by parties and candidates to non-partisan civic observers recruited by organizations like the League of Women Voters, Common Cause, and Scrutineers (scrutineers.org). Scrutineers is the partner organization most directly aligned with AC’s mission: it focuses on election protection work that includes voter education, observer recruitment, and verification activities that overlap with AC’s.
AV provides a specific technical methodology for what observers can do with the evidence they collect. A poll observer who is present at closing can watch the poll tapes being printed and posted, but without AV, their observation is ephemeral — it exists only in the observer’s memory and notes. With AV, the observer produces a permanent video record with metadata, chain-of-custody documentation, and integration into a comparison pipeline. AV transforms observation from a passive activity into an evidence-producing one.
Recounts. A recount is an administrative re-tabulation of the ballots in a specific election, triggered by statute (when the margin is close enough) or by petition (when a candidate or party requests one). Recounts vary widely in their rigor: some are full hand counts of all ballots, while others simply re-run the same ballots through the same machines.
AV is not a recount and does not substitute for one. A recount re-examines the counting layer; AV examines the reporting layer. But AV can provide evidence that triggers or informs a recount. If an AV analysis reveals a reporting discrepancy large enough to affect the outcome of a race, that finding may provide the factual basis for a recount petition or a court challenge (as discussed in Section 6.5). AV is upstream of the recount process: it produces the evidence that makes the case for further investigation.
FOIA campaigns. Freedom of Information Act requests and their state-level equivalents are used by election integrity researchers, journalists, and civic organizations to obtain election records — cast vote records, ballot images, equipment logs, internal communications, and other materials that are not routinely published. FOIA campaigns are an important tool for election transparency and have produced significant findings in multiple jurisdictions.
From AV’s perspective, FOIA is one access method among several. Section 7 documents the various methods by which AV evidence can be obtained, and FOIA-equivalent records requests are one of them (Section 7.5 addresses arrangements with elections offices that may include records requests). FOIA campaigns are valuable but slow, uncertain, and dependent on the jurisdiction’s willingness to comply. AV’s primary evidence — poll tape videos — is typically obtained through direct observation at polling places, which does not require a records request. The two approaches are complementary: AV provides real-time evidence collection, while FOIA provides access to records that are not otherwise publicly available.
Academic election security research. Universities and research institutions produce theoretical and empirical work on voting system vulnerabilities, audit methodologies, election administration, and related topics. This work is published in academic journals, informs policy discussions, and provides the intellectual foundations for tools like RLAs. Prominent researchers include Philip Stark (who developed the RLA methodology), J. Alex Halderman (who has demonstrated voting machine vulnerabilities), and many others.
AV is an applied operational tool, not a research program. AC’s work draws on and contributes to the academic body of work, but AC’s primary output is evidence about specific elections, not academic papers about methodology in general. AC’s technical papers (discussed below) are one exception: they apply academic-style reasoning to questions that arise from AC’s operational work. But the core of AC’s contribution is operational, not theoretical.
Where AV is uniquely valuable
The mapping above reveals a specific niche that AV occupies, and that no other existing tool or organization fills in the same way.
AV is independent. It does not require the cooperation or approval of elections offices to operate. Unlike RLAs, which are conducted by the same offices whose work is being verified, AV operates from outside the system. This independence is what makes AV a genuinely external check, analogous to an independent financial audit rather than an internal review.
AV is citizen-powered. Ordinary citizens — not trained election officials, not computer scientists, not academic researchers — can participate in the evidence collection process. The methodology is designed to be accessible to volunteers with minimal training. This means that AV’s capacity scales with public engagement rather than with government resources or academic funding.
AV is evidence-producing. AV does not merely observe elections; it produces a structured body of evidence — video recordings, transcriptions, comparison datasets, investigation records — that is preserved, accessible, and legally usable. This distinguishes AV from observation activities that produce only memories and notes.
AV is publicly accessible. The evidence AV produces and the findings AV publishes are available to anyone. They are not locked behind paywalls, restricted to government officials, or available only to parties in litigation. The public archive is a public resource.
AV is legally usable. The evidence AV produces is designed from the ground up to meet evidentiary standards (as documented in Section 6.4). The chain of custody, the metadata, the vetting process, and the comparison methodology are all designed to produce evidence that can survive scrutiny in legal proceedings.
AV requires no authority cooperation. This is perhaps the most distinctive feature. Every other major election verification tool — RLAs, ballot image audits, recounts — depends on the cooperation of election authorities to provide access to the materials being verified. AV’s primary evidence (poll tapes) is either publicly posted (Method 1, Section 7.1) or accessible through poll observer rights (Method 2, Section 7.2) or through arrangements with elections offices (Method 3, Section 7.3). AC can do its work even when the elections office would prefer that it did not.
The combination of these features — independence, citizen participation, evidence production, public accessibility, legal usability, and no authority requirement — is what makes AV unique in the election integrity landscape. Other tools have some of these features; no other tool has all of them.
Where AV has limitations
The same honesty that requires mapping AV’s unique strengths requires mapping its limitations.
AV only covers the vote reporting layer. As discussed at length in Section 13.4, AV cannot detect errors or fraud at the counting layer. This is its most important structural limitation and the one that defines its complementary relationship with counting-verification tools like RLAs.
AV requires physical presence at polling places (for Method 1 and Method 2 evidence collection). Someone must be at the polling place when the tapes are printed and posted. This creates a logistical challenge that scales with the number of polling places in a jurisdiction. In a large county with hundreds of polling places, achieving comprehensive coverage requires hundreds of volunteers.
AV’s coverage is constrained by volunteer capacity. Because AV is citizen-powered, its coverage is limited by the number of volunteers who participate. In elections where AV has operated, coverage has ranged from a handful of precincts (Broward 2018) to hundreds (NC 2024), but no operation has yet achieved full coverage of all precincts in a large jurisdiction. The rising-tide theory of change (discussed below) addresses this limitation, but it is a real constraint in the near term.
AV does not verify vote counting. This is a restatement of the matched poll tape caveat for the context of landscape positioning. When a reader asks “does AV verify that my vote was counted correctly?”, the answer is no. AV verifies that whatever the machines tabulated was correctly reported. That is valuable, but it is not the same thing as verifying the count itself.
The “rising tide” theory of change
AV’s theory of change does not depend on achieving perfect coverage. The theory is that as AV adoption grows, the deterrence effect increases. When bad actors — whether they are election officials cutting procedural corners or more deliberate manipulators — know that independent observers are recording poll tapes and comparing them to official results, the incentive structure shifts. Even partial coverage changes the calculation, because a bad actor cannot know in advance which precincts will be covered.
The metaphor is a rising tide. In the early stages, AV coverage is spotty — a handful of precincts here, a few dozen there. This is already useful (the evidence from those precincts is real evidence), but it is easy for a bad actor to avoid the covered precincts. As coverage grows, the gaps narrow. A bad actor would have to take increasing risks to exploit the uncovered areas. At sufficiently high coverage, the risk of detection becomes so high that the rational calculation shifts from “can I get away with it?” to “it’s not worth trying.”
The rising tide has additional effects beyond deterrence. Increasing AV activity creates pressure on elections offices to maintain proper procedures — posting poll tapes on time, ensuring accurate transcription, following their own rules. AC’s Broward County field experience documents this dynamic: posting compliance improved over time as AV’s presence became known. The tide also creates pressure on legislatures to strengthen transparency requirements: when citizens are visibly exercising their rights to observe poll tapes, the political case for expanding those rights becomes stronger.
The theory of change is patient. AV does not expect to transform election integrity overnight. The expectation is that the slow, steady growth of a citizen-powered verification infrastructure will, over time, shift incentives, improve practices, and build the body of evidence that supports institutional reform. Each election cycle with AV coverage adds to the archive, trains new volunteers, builds new relationships with elections offices, and extends the reach of the verification network.
Pointers to AC’s technical papers
AC’s philosophical framework extends beyond this manual into a set of technical papers that develop specific theoretical arguments relevant to election integrity. These papers are working drafts being prepared for academic publication. They are not required reading for understanding AV’s operational methodology, but they provide the intellectual foundations for several claims made throughout this manual.
Stochastic Harm: Measuring Indirect Causation Across Scales. This paper develops a framework for understanding how election integrity violations propagate indirect harm. The core argument is that election errors — even ones that do not change the outcome of a specific race — create stochastic harm: they increase the probability of bad outcomes across a population in ways that are real but difficult to attribute to any single event. The paper formalizes this intuition and provides a mathematical framework for quantifying the expected harm from undetected election errors. The stochastic harm framework undergirds AC’s position that even errors that do not change outcomes are worth detecting and correcting, because they degrade the reliability of the system in ways that have real consequences.
Social Well-Being: A Working Construct for Quantifying Collective Flourishing. This paper develops a composite welfare function — Social Well-Being (SWB) — that extends the prioritarian welfare-economics tradition with collective-goods terms, inequality penalties, and floor-violation constraints. The SWB construct provides a framework for quantifying the collective benefit of transparent, accurate elections in dollar-equivalent units. The paper is foundational for the election insurability argument (below) and for AC’s broader claim that election integrity is not merely a procedural concern but a substantive contributor to collective welfare. The SWB framework provides the theoretical basis for the argument that investing in election verification — including tools like AV — has a positive return on investment measured in terms of collective flourishing.
Would You Insure a US Election? Election Integrity as an Insurance Pricing Problem. This paper applies the SWB framework to a specific thought experiment: would a rational insurance underwriter write a policy guaranteeing the accuracy of a US presidential election? The answer is no — the variance of plausible ground truths about election outcomes, combined with the stakes established by the SWB framework, produces an expected loss so large that no rational underwriter would accept the risk at any affordable premium. The argument goes through on the transparency channel alone (meaning it does not require proving that elections are actually inaccurate, only that the absence of independent verification makes the variance unacceptably high). The election insurability paper provides one of AC’s most compelling arguments for the necessity of independent verification: not because we know elections are wrong, but because we cannot prove they are right, and the stakes are too high to accept that uncertainty.
These three papers form a sequence. The Social Well-Being paper establishes the framework for quantifying what is at stake. The Stochastic Harm paper explains how election integrity violations create real costs even when they do not change specific outcomes. The Election Insurability paper applies both frameworks to demonstrate that the current level of election verification is inadequate relative to the stakes. Together, they provide the theoretical backbone for AC’s operational work.
The papers are referenced here rather than reproduced or summarized at length because they are works of academic reasoning that reward careful reading in their own right. Readers interested in the theoretical foundations of AC’s work should consult the papers directly.
13.6 — The Benefits of Actual Vote, Consolidated
The four principles developed in this chapter — transparency, non-partisanship, independence, and humility — are not abstractions. Each implies a concrete benefit that AV delivers in practice, and those benefits are scattered across this manual and the broader site wherever the relevant principle or mechanism is discussed. This final subsection gathers them in one place: first by pointing to the benefits already developed elsewhere, then by elaborating the ones that have been implicit, under-developed, or unnamed. A plain-language version of this material, written for a general audience rather than for the analytical reader of this manual, lives on the Start Here “Benefits of Actual Vote” page.
Four benefits are developed in detail elsewhere and need only a pointer here. Transparency — the public availability of evidence, methodology, and the organization itself — is the subject of Section 13.1. A new mode of civic engagement — the way AV converts ordinary citizens into evidence producers rather than passive observers — runs through the citizen-powered discussion in Section 13.5 and the operational chapters on volunteer participation. Improved accuracy — the detective and deterrent effects that raise the quality of vote reporting itself — is developed through the SEC audit analogy in Section 13.3. Social well-being — the collective-welfare case for accurate, verifiable elections — is the subject of the Social Well-Being and Stochastic Harm papers referenced in Section 13.5. The remainder of this subsection concentrates on the benefits that the manual has so far left implicit or named only in passing.
The deterrent effect
A known, credible, public corpus of precinct tapes changes the incentive structure for everyone who touches the reporting layer, and it does so independently of whether any specific problem is ever caught. The deterrent operates on two distinct audiences. The first is the would-be manipulator — anyone in a position to alter, lose, or misreport a number between the poll tape and the published result. When such a person cannot know in advance which precincts a citizen happened to photograph, every precinct must be treated as potentially observed. The rational calculation shifts from “can I get away with this here?” to “can I afford the risk that this precinct is one of the recorded ones?” The second audience is the bad-faith accuser — anyone tempted to assert, without evidence, that a jurisdiction’s numbers were fabricated or swapped. A public archive of tapes that match the official results is a standing rebuttal: it is much harder to manufacture suspicion about a precinct whose tape anyone can pull up and compare for themselves.
What makes the deterrent valuable is that it does not depend on detection. The independent financial audit raises the quality of corporate accounting not primarily because auditors catch fraud, but because the standing possibility of an audit makes carelessness and manipulation riskier in advance. AV works the same way. Even an analysis that finds nothing wrong has already done deterrent work simply by existing and being known to exist. The deterrent is a benefit that accrues to every precinct in a jurisdiction with active AV coverage, not only to the precincts that were actually recorded — and it accrues whether or not a single discrepancy is ever found.
Proof already realized, not just hypothetical
Much of the case for independent verification is necessarily framed in the conditional: a tool would have caught a given problem, or would help if something went wrong. AV’s published clean analyses are different in kind. Each one is not a promise about what verification might do someday; it is a record that independent verification actually happened in a specific place and time, and a statement of what it found. The value lies in the structure of the evidence. An AV analysis compares two records that were produced independently of each other — the citizen’s poll-tape capture on election night, and the elections office’s later official result — through entirely separate channels. When those two genuinely independent records agree, and there is no hidden path by which one was derived from the other, the agreement is positive evidence: two separate accounts of the same underlying reality converged. That is a stronger and more useful thing than a hypothetical “would have helped.”
This is the point at which the manual’s posture must be precise, because the raw comparison is easy to misread. A first-pass comparison between a tape capture and an official results file surfaces many values that do not match on their face. The crucial fact is that the great majority of those non-matches are not candidate errors or fraud at all — they are artifacts of our own process (transcription choices, formatting, how a multi-page tape was read) together with benign, undocumented reporting conventions used by the jurisdiction, all of which are identified and resolved during analysis. The headline of a clean analysis is therefore not “look how many discrepancies we explained away.” Explaining away one’s own artifacts is housekeeping, not proof. The headline is the verified match: the body of comparisons that, after that housekeeping, show two independent sources in genuine agreement, plus the honest reporting of any discrepancy that remains genuinely unexplained. Framed correctly, “proof already realized” rests on the power of independent agreement absent any tautology — not on the volume of apparent problems that turned out to be nothing. The published analyses that demonstrate this are collected on the our analyses page.
Evidence held outside government custody
AV evidence has a property that distinguishes it from nearly every other record relevant to an election dispute: it is never in the custody of the offices whose work it checks. Volunteers capture the tapes independently, on election night, on their own devices, and the recordings enter AC’s archive directly. The official results, the internal logs, the chain-of-custody paperwork, and in many cases the ballots themselves are all held by the elections office. AV’s record is not.
The consequence is that AV evidence survives events that compromise official records. If documents are destroyed on a retention schedule, lost in a move or a system migration, corrupted, or tampered with — whether through negligence or bad faith — the independently held tape capture remains. An external record that does not depend on the continued integrity or good faith of the audited party is exactly the kind of evidence that verification is supposed to provide. It is the difference between asking an office to vouch for its own paperwork and holding a copy that the office never controlled and cannot quietly alter.
Speed of independent verification
The window during which a wrong or missing number can do damage is the window during which it stands unchallenged. Litigation, formal contests, and academic post-mortems can eventually surface a reporting problem, but they tend to do so months or years later, long after results have been certified, offices seated, and policy set in motion. By that point the number has already done its work, and correcting the record is an exercise in history rather than remedy.
AV collapses that window. Because the evidence is captured on election night and the comparison runs on public data with documented code, an analysis can be produced quickly — in the most valuable case, before certification, while results are still administratively malleable. A discrepancy surfaced in that window can inform a correction, a recount petition, or a contest while there is still something to be done about it, rather than becoming a grievance to be relitigated after the fact. Speed is not a convenience here; it is part of what makes the verification consequential.
Reproducibility as a confidence mechanism
A claim that cannot be independently checked is not evidence, however confidently it is asserted. AV is built so that its central claim — that a given set of tapes agrees with a given set of official results — can be re-derived by anyone. The tape captures are public, the official results are public, the comparison methodology is documented, and the code is available. A reader who doubts a finding does not have to weigh AC’s reputation against their own skepticism; they can re-run the comparison on the same inputs and see whether they reach the same result.
This reproducibility is itself a benefit, distinct from any particular finding. It converts trust from something an audience must extend on faith into something an audience can verify on demand. It also disciplines AC: a methodology that anyone can re-run is a methodology that cannot quietly cut corners, because the corner-cutting would be reproducible too. Reproducibility is the operational form of the transparency principle, and it is the reason AC’s findings are designed to be falsifiable rather than merely authoritative.
Structural resistance to capture
Non-partisanship at AC is not a promise to behave well; it is a property enforced by how the system is built, as developed in Section 13.2. Three design features do the enforcing. Participation is open: anyone can record a tape and submit it, so a covered precinct cannot be quietly dropped without the gap being visible to others who hold the same evidence. The evidence is public: a finding cannot be fabricated or suppressed without contradicting recordings and results that anyone can examine. And the methodology is disclosed: a critic who suspects the comparison was applied more aggressively in one race than another can test that suspicion against the documented process and the actual records.
The benefit of building non-partisanship into the structure, rather than asserting it as policy, is that it survives bad faith and turnover. A policy of even-handedness is only as good as the people currently administering it. A system in which selective suppression would be detectable, fabrication would be contradicted by public evidence, and inconsistent application would be visible in the record does not require trusting the current team’s character. It resists capture by design — which is precisely the kind of assurance that an audience with little reason to extend trust can actually accept.
A cumulative evidentiary baseline
The value of an AV analysis does not stop when the analysis is published; it compounds as the archive grows across elections. A single clean county in a single cycle is a useful fact on its own. But it is also a baseline — a documented record of what that jurisdiction’s vote reporting looked like, captured independently, under a consistent methodology. Against that baseline, a future anomaly becomes legible in a way it never could be in isolation. A reporting pattern that would look unremarkable in a single election reveals itself as a break from precedent when there is precedent to break from.
This is why the archive is treated throughout the manual as a permanent public resource rather than a byproduct. Each cycle of coverage adds comparison points, extends the historical record for the jurisdictions involved, and sharpens the contrast against which later irregularities stand out. The benefit is structurally different from the others in this subsection: it is not realized in any single analysis but accumulates across all of them. The more elections AV covers, and the longer it covers them, the more valuable every individual analysis in the archive becomes — because each one is both evidence about its own election and a reference point for every election that follows.