5 — Using the Actual Vote App

This section is the user-facing guide to the Actual Vote app. Other sections describe the app’s architecture (Section 4.2), the legal basis for recording poll tapes (Section 6), the methods for accessing them (Section 7), and the operational practices for teams and committees (Section 10). This section is the practical walkthrough: how to install the app, how to use it, and what to expect at every step.

A reader who follows this section from start to finish should be able to install the app, create an account, complete a practice recording, understand what makes a good recording, and submit a real poll tape recording on election night — without needing to read any other section first.

5.1 — Getting the App

5.1.1 Where to download

The Actual Vote app is free and published by America Counts (Democracy Counts, Inc.). It is available for both iOS and Android.

The app is listed as Actual Vote in both stores:

iOS (App Store): apps.apple.com/us/app/actual-vote/id1499733603
Android (Google Play): play.google.com/store/apps/details?id=democracycounts.actualvote.android

Or search “Actual Vote” in either store.

The app runs on iPhones, iPads, and Android phones and tablets — any device on iOS 15 or newer, or Android 8.0 (Oreo) / API level 26 or newer, with a working camera (see Section 5.1.3 for details).

5.1.2 The app store availability risk

App store availability is not guaranteed. In November 2024, the Google Play Store classified Actual Vote as “government information” and rejected several builds in the days leading up to the general election. The latest Android version was not approved until November 8 — three days after election day. The iOS version was not affected.

This episode established a recurring operational risk: app store gatekeepers make their own decisions about politically-adjacent apps, and those decisions can be unpredictable and poorly timed. AC cannot guarantee the app’s availability through any app store at any given moment. If you are planning to use the app for an upcoming election, install it well in advance — do not wait until election week.

If you cannot install or update the app, the fallback method (Section 5.1.5 below) works reliably.

The two app stores are the only ways to install Actual Vote; there is no separate direct-download (APK) or beta channel. If a store is unavailable to you, use the native-camera fallback in Section 5.1.5 — it requires no installation at all.

5.1.3 Device requirements

Actual Vote runs on any Apple device on iOS 15 or newer, and any Android device on Android 8.0 (Oreo) / API level 26 or newer. In practice that’s most phones and tablets sold from roughly 2017 onward; there’s no per-model list to check beyond the OS version.

Hardware. Beyond the operating-system version, the only hardware requirement is a working camera. There is no minimum resolution, autofocus, or other special capability: any modern smartphone camera is sufficient for legible poll-tape recordings. The limiting factor is usually technique (holding the phone steady, panning at the right speed) rather than hardware.

Storage for the app itself. The app is small. On iPhone and iPad it is roughly a 25 MB download (Apple tailors the download to your specific device, so the figure varies a little by model) and about 70 MB once installed. On Android it is about a 22 MB download for a new install. Either way, the app itself takes up far less room than a single short video recording.

Storage per recording. Each recording is a short video, so the footprint is small. As a rough guide, a single poll-tape recording is around 20 MB. A handful of recordings is therefore a couple hundred megabytes, and even 100 recordings is only about 3 GB. The exact size depends on how long the tape is — longer tapes mean longer videos — and on recording at a quality that keeps the numbers legible. Record at the quality you need and no longer than necessary; oversized files take longer to upload without improving the evidence. Free up storage before election night if your device is near capacity.

5.1.4 Permissions the app requests and why

When you first open the app, it will request several device permissions. Each serves a specific purpose in the evidentiary chain:

Camera and microphone — required. This is how the recording is made. The app uses your device’s standard camera and microphone hardware; there is no special-purpose video processing on-device. The recording is a normal video file. Audio is captured alongside video, which provides an ambient record of the environment — useful for establishing context if the recording’s authenticity is ever questioned.

Location services — strongly recommended. When granted, the app automatically captures GPS coordinates at the moment of recording. These coordinates are embedded in the recording’s metadata, establishing where the recording was made. This is important for chain of custody: it links the recording to a specific polling place with machine-verifiable data that doesn’t depend on your memory or notes. If you deny location permission, the recording will still work, but it will lack GPS metadata, and AC will need to rely entirely on your manually entered precinct identifier to locate it. Grant this permission.

On iOS, “Add to Photos” — optional. On iOS the app saves a copy of each recording to your device’s Photos library, which requires this permission. The copy stays entirely on your device; America Counts never receives it unless you upload the submission through the app. (On Android, the system camera typically saves a copy to your gallery automatically as part of capture.)

Beyond camera, microphone, location, and — on iOS — Photos, the app requests no other permissions. It does not request notifications. It runs no background processing and no analytics or crash-reporting SDK, so it sends nothing to our servers in the background; the only thing the app ever transmits is the submission itself, over the network, when you choose to upload it. The full inventory of what is and isn’t collected is in the Privacy Policy.

5.1.5 The fallback: recording without the app

If you cannot install or use the app — because of an app store rejection, a device incompatibility, or any other reason — you can record with your phone’s native camera app and send the video to America Counts. The recordings produced this way are functionally identical: a video file with GPS coordinates and timestamp metadata embedded by the device’s hardware. What you lose is the in-app precinct identifier field and the automatic upload queue. What you gain is complete independence from app store availability.

To submit without the app:

Open your device’s native camera app. Switch to video mode. Record the poll tape using the technique described in Section 5.4 (start at the top, pan slowly, capture everything). When finished, note the polling place name and address, the precinct number (visible on the tape header), and the date.

Submit the recording by one of these methods:

Email the video to j.flatley@democracycounts.org with the metadata in the email body. This works for one or a few recordings. For larger files, your email client may compress the video or reject the attachment.

For bulk submissions (multiple recordings from one night), email Jason Flatley in advance to coordinate a Google Drive folder. Upload each recording as a separate file with a descriptive filename (e.g., “Precinct_042_MarionTownship_2024Nov05.mp4”). Notify Jason when uploading is complete. AC processes these through the same pipeline as app submissions.

The native camera fallback works but is slower, more error-prone (metadata must be manually communicated rather than automatically captured), and harder for AC to process at scale. It is a genuine safety net, not a replacement for the app. See Section 10.7 for detailed fallback workflows and the Google Drive protocol.

5.2 — Creating Your Account

5.2.1 Why an account is required

Every recording submitted through the Actual Vote app must be associated with a submitter account. This is not surveillance — it is chain of custody. If a recording later becomes evidence in a dispute or is cited in a published analysis, the chain from “a specific person recorded this at a specific place and time” to “AC analyzed it and found X” must be traceable. Anonymous submissions would break that chain.

Your identity is protected by AC’s Privacy Policy (reproduced in full in B.4 of this manual). Your name is not published alongside your recordings in the public archive.

For an approved recording in the public archive, anyone can see the recording itself, the location information associated with it (polling place, precinct, geocoded address, and GPS coordinates), the date, and the analytical records we’ve attached (transcription, comparison, and any investigation notes). What is not public is the submitter’s identity: the “created by” field is visible only to authenticated administrators inside our own system. Other signed-in volunteers cannot see who submitted a given recording, and members of the public cannot see it at all.

Visibility also depends on a recording’s status. Only approved recordings appear in the public archive — practice recordings and rejected recordings are never shown to other users. The one place you see your own full history, including your practice and rejected recordings, is the My Actual Vote Videos tab in the app, which shows only your own submissions. The Privacy Policy’s “What Becomes Public” section is the authoritative description of this boundary.

5.2.2 Account creation walkthrough

The app offers two authentication methods: Google sign-in and email-and-password. Google sign-in is the faster option if you have a Google account. Email-and-password works for everyone else.

5.2.3 Staying logged in

Once you’ve created your account and logged in, the app should keep you logged in across sessions. However, app updates, device restarts, and cleared caches can sometimes reset your login state. Before heading to a polling place on election night, open the app and confirm you are logged in. Do not discover a login problem when you’re standing in front of a poll tape.

5.2.4 Account recovery

Actual Vote uses your Google account to sign in, so account recovery is handled entirely by Google. If you can’t sign in because you’ve forgotten your password or lost access to your Google account, use Google’s account recovery — America Counts has no access to your Google credentials and cannot reset them. Once you regain access to your Google account, you can sign back in to Actual Vote normally. This is worth knowing in advance, because many people set up the app weeks or months before election day and don’t open it again until election night.

5.3 — The Practice Recording

5.3.1 Why practice matters

Recording a poll tape well is a physical skill. The mechanics are simple — you’re holding a phone and slowly moving it down a piece of paper — but doing it well enough to produce a legible, usable recording takes a little practice. You need to manage focus (thermal paper and phone autofocus don’t always cooperate), speed (too fast and the video is blurry; too slow and the file is unnecessarily large), framing (the full width of the tape should be visible at all times), and steadiness (any shake reduces legibility).

On election night, you’ll be doing this in a real environment: possibly outdoors, possibly in a dimly lit hallway, possibly with other people around, and with the knowledge that this is your one chance to capture the evidence. Practice beforehand so the physical mechanics are automatic and you can focus on the environment.

5.3.2 Getting the practice poll tape

America Counts publishes a sample practice poll tape as a printable PDF. The practice tape looks like a real DS200 results tape — precinct header, contest-by-contest results, ballot counts, signature block — but its values don’t correspond to any real election.

Download it here: Actual Vote Practice Poll Tape (PDF). One generic version is provided; real tapes vary by tabulator and are often longer, but the practice tape is enough to learn the workflow.

You don’t strictly need the practice tape, either. The point of practicing is to confirm the app works on your device and to get a feel for framing and uploading — so a video of a blank wall, or even a grocery receipt, works just as well for a first dry run.

Print the practice tape. Tape it to a wall at roughly the height a real tape would be posted — eye level or slightly above. This simulates the most common recording condition: a tape posted on the wall or window of a polling place after polls close.

5.3.3 How to make a practice recording

There’s no separate “practice” button or mode. You record exactly the way you’d record a real poll tape — the workflow is identical (it’s walked through step by step in the recording section below). The only difference is what you point the camera at: because you’re recording the practice tape (or anything that isn’t a real poll tape), a reviewer will simply mark the submission as a practice recording rather than publishing it. So practicing is just doing a normal recording of the practice tape.

Hold your phone in front of the top of the practice tape. The tape header — with the precinct name, election name, and date — should be visible in the frame. Press record.

Pan the phone slowly downward. The goal is to move slowly enough that a viewer can pause the video at any point and read the text on screen. A good rule of thumb: if you can comfortably read the text on your phone screen as you’re recording, you’re moving at about the right speed. If the text is blurring as you move, slow down.

Keep the full width of the tape in frame at all times. Resist the urge to zoom in — if you need the text to be larger, move the phone physically closer to the tape rather than using digital zoom, which reduces resolution.

When you reach the bottom of the tape — the signature block and any trailing whitespace — stop recording.

5.3.4 What a good recording looks like

A good recording has five qualities:

Complete. It captures the entire tape from the first line of the header to the last line of the signature block. Nothing is cut off at the top or bottom. Every contest is visible.

Legible. A viewer can pause the video at any frame and read the text. Contest names, candidate names, and vote totals are all clearly readable. The font on thermal paper is small, so legibility depends on getting close enough, keeping the camera steady, and maintaining focus.

Steady. The camera doesn’t shake. Minor movement is inevitable when hand-holding a phone, but large jitters or rapid movements make frames unreadable. Tuck your elbows against your body for stability. If you can lean against a wall or rest your arms on something, do so.

Well-lit. The text is evenly illuminated with no deep shadows or bright glare spots. Poll tapes are often posted in hallways, on exterior walls, or behind glass — all environments where lighting can be challenging. If the lighting is poor, use your phone’s flashlight or bring a small handheld flashlight. If the tape is behind glass, angle your phone slightly to avoid capturing the reflection of your own screen.

Properly focused. Thermal paper has low contrast (dark gray text on a lighter gray background), and phone autofocus can struggle with it, especially in low light. Before you start recording, tap the screen to focus on the tape text. If the focus drifts during recording (you’ll see the text go soft), pause your downward motion, tap to refocus, then continue.

5.3.5 Common practice mistakes

Moving too fast. This is the most common mistake. When you watch the recording back and try to pause on a specific line, the text is blurred. Slow down. A full-length results tape for a long ballot might take 60 to 90 seconds to record. That’s fine.

Not capturing the header. The header — precinct name, election name, date, machine ID — is how AC identifies which precinct the tape belongs to. If you start recording mid-tape and miss the header, the recording may be unusable.

Zooming in digitally. Digital zoom reduces resolution. If you need the text larger in the frame, move your phone closer to the tape. Physical proximity beats digital zoom every time.

Orientation. Record in portrait — it’s strongly preferred. Poll tapes are tall and narrow, so portrait is the natural fit, and the large majority of recordings are made this way. Portrait recordings are also faster to transcribe, both by hand and with our AI tools, which helps us turn your submission around quickly. That said, the real goal is capturing the vote-total data legibly: if an awkward posting spot makes landscape (or any other angle) the only way to get a clear, complete recording, do whatever you have to do to get the numbers.

Not watching the recording back. After you submit, watch your practice recording in the My Actual Vote Videos tab. Can you read every line? Can you read the precinct name? Can you read the smallest vote totals? If not, try again. It’s better to discover technique problems now than on election night.

5.3.6 The practice recording in the system

A practice recording is a valid submission — it enters the Actual Vote system and appears in your My Actual Vote Videos tab. But it is classified as a practice submission and is not transcribed, analyzed, or published in the public archive. The practice tape’s values don’t correspond to any real election, so there’s nothing to compare against.

The practice recording does serve a secondary purpose for AC: it demonstrates that you’ve engaged with the system and practiced the technique. Users who have completed practice recordings are more likely to produce usable submissions on election night.

Practice is mostly self-service: if your practice recording is clear and the numbers are readable, you’re ready — there’s no formal score or sign-off. America Counts doesn’t usually give feedback on practice recordings. We may reach out to you at your Actual Vote account email if something warrants it, but in practice the more common reason we contact someone is the opposite situation — a volunteer in the field submitting illegible recordings, where we need to ask them to adjust their technique.

5.4 — Election Day: Recording a Real Poll Tape

5.4.1 Before you go: the preparation checklist

Start preparing at least a day before the election. On election night itself, you want the mechanics to be automatic so you can focus on the environment.

App and account. Open the app and confirm you’re logged in. Check for updates — if a new version is available, install it now, not on election night. Confirm that your practice recording appears in My Actual Vote Videos.

Device. Charge your phone fully. Bring a portable battery charger if you have one — election night recording can drain your battery, especially if you’re recording at multiple locations. Free up storage space; each recording is roughly 20 MB, so a full night of recording at several locations is still only a few hundred megabytes (see Section 5.1.3).

Knowledge. Know your assigned polling place and its address. Know your precinct number (this may be on your voter registration card, your state’s voter lookup website, or your county’s elections website). Know when polls close in your jurisdiction — this is when tapes will be printed and posted. Know the legal basis for your access: does your state require poll tapes to be publicly posted? See Section 6 (Legal Basis) and Section 7 (Access Methods) for the full treatment, and the state-by-state reference in B.3 for jurisdiction-specific details.

Equipment. A small flashlight (or your phone’s built-in flashlight function) for low-light conditions. Warm clothing if it’s November and you’ll be standing outside. Something to write on in case you need to note precinct numbers, machine IDs, or anything unusual. A printed copy of your state’s relevant statute on poll tape posting, if you’re using Method 1, in case anyone questions your right to be there.

5.4.2 Arriving at the polling place

Timing depends on your access method. For Method 1 (publicly posted tapes after polls close), plan to arrive at the polling place 15 to 30 minutes before polls close. This gives you time to find the location, identify where tapes will be posted, and be in position when closing procedures begin.

Poll tapes are typically posted in a publicly visible location: on the front door of the polling place, on a wall or window near the entrance, or on a bulletin board outside. The posting location varies by jurisdiction, by polling place, and sometimes by the preference of the chief judge at that precinct. If you don’t see tapes posted within 30 minutes of poll closing, look around — they may be posted on a side door, an interior hallway window visible from outside, or a different location than you expected.

You do not need to enter the polling place to record a posted tape. If the tape is posted outside (the most common arrangement), you can record from the public sidewalk or parking lot. If the tape is posted inside but visible through a window or door, you can record through the glass (watch for glare). If the tape is posted inside in a location not visible from outside, check whether the jurisdiction allows public access to the interior after polls close. See Section 7 (Access Methods) for the full treatment of each scenario.

Do not enter any area marked as restricted. Do not interfere with poll workers’ closing procedures. Do not touch the poll tape. You are there to observe and record, not to participate in the closing process.

5.4.3 Identifying the correct tape

A polling place may display multiple printed tapes during its closing procedures. You want the results tape — the tape printed after polls close that shows the final vote totals for each contest. This is sometimes called the “closing tape” or “totals tape.”

How to identify it: the results tape will have non-zero vote totals next to candidate names. It will show every contest on the ballot — federal, state, local, ballot measures — with the vote count for each choice. It will typically have a header identifying the precinct, the election, and the date, and a footer with a signature block.

Tapes you do not want to record (or rather, that are not what AV analyzes — record them if convenient, but prioritize the results tape):

The zero tape is printed before polls open and shows all contests at zero. It confirms the machine started with a clean count. You can identify it because every vote total is zero. If you’re not sure whether a tape is a zero tape or a results tape, look at the vote totals: if they’re all zero, it’s a zero tape.

The configuration report is printed during machine startup and shows machine settings: firmware versions, election definition details, serial numbers, battery status. It does not show vote totals organized by contest.

The ballot status accounting report shows ballot counts by category (scanned, rejected, provisional) rather than contest-by-contest vote totals.

For a detailed anatomy of each tape type — what appears on each line, what the fields mean, how to read the formatting — see Section 3 (Understanding Poll Tapes).

5.4.4 Recording the tape: step by step

Open the Actual Vote app. Confirm you are logged in.

Position your phone at the top of the tape. The header — precinct name, election name, date, and machine ID — should fill the frame. The full width of the tape should be visible. Tap the screen to focus on the text.

Press record. Begin panning slowly downward. Maintain a pace that allows every line to be clearly readable if the video is paused. Keep the full width of the tape in frame throughout. If you feel the focus drifting (the text goes soft), pause your downward motion, tap the screen to refocus, and continue.

When you reach the bottom of the tape — past the last contest, past the ballot summary counts, past the signature block — stop recording.

If the tape is long (a full ballot in a large jurisdiction can produce several feet of thermal paper), maintain a steady pace throughout. Do not rush the lower portions because you’ve been recording for a while. The down-ballot contests and ballot measures at the bottom of the tape are just as important as the top-of-ticket races.

If the tape is partially obscured — folded over, taped together at a seam, posted behind glass with a glare spot — do your best. Record what you can see. Note the obstruction.

If a polling place has multiple scanners (common in large precincts or combined early voting sites), there will be multiple results tapes — one per scanner. Record each tape as a separate submission. In the precinct identifier field, include something to distinguish them (e.g., “Precinct 42 - Machine A” or “Precinct 42 - Scanner 001”). If the tape header includes a scanner serial number or machine ID, use that.

5.4.5 Entering the precinct identifier

The precinct identifier links your recording to a specific precinct in AC’s comparison pipeline. Enter whatever precinct information is visible on the poll tape header. If the header shows “PRECINCT 0042” or “MARION TWP 02,” enter that. If the header shows a polling place name rather than a precinct number (e.g., “UNITY CHRISTIAN CHURCH”), enter the polling place name.

If the polling place serves multiple precincts (common with vote centers and combined precincts), the tape header will typically identify which precinct’s results it contains. Enter the precinct identifier from the tape, not the polling place name.

If you’re not sure what to enter, enter your best approximation. The GPS coordinates captured by the app will help AC locate the polling place, and the tape header visible in the video will provide the canonical identification. AC’s transcription team handles the mapping from your entry to the official precinct designation. A reasonable guess is better than leaving the field blank.

5.4.6 Confirming your submission

After you finish recording and enter the precinct identifier, the app processes your submission.

If the upload succeeds, you should see a confirmation. The recording now exists in the Actual Vote system.

If you’re in an area with no cell signal or weak connectivity (common at schools, churches, and government buildings), the recording is saved locally on your device and queued for automatic upload when connectivity returns. You do not need to do anything — the app handles the retry. The recording is safe on your device. When you later connect to Wi-Fi or regain cell signal, the upload will proceed in the background.

Check the My Actual Vote Videos tab to confirm your submissions are showing up. If a recording shows “pending upload” for an extended period and you have connectivity, try force-quitting and restarting the app.

5.4.7 Recording at multiple locations

If you are covering more than one polling place (common for organized teams — see Section 10.2), repeat the process at each location. Each tape is a separate submission. The app handles multiple submissions naturally; there is no limit to the number of recordings you can make in one session.

When moving between locations, confirm each previous submission is either uploaded or queued before proceeding. This is a sanity check — the recordings are saved locally regardless — but it’s good practice to verify that nothing was lost.

For teams covering many locations, the coordination protocol in Section 10.2 applies: plan your route, communicate with teammates about coverage, and report to AC if anything unexpected happens (a polling place that doesn’t post tapes, a tape that looks unusual, a location that has already been covered by another team member).

5.5 — After Election Day

5.5.1 Checking your submissions

After election night, you can review your submissions in the My Actual Vote Videos tab of the app. Approved recordings are also published to the public web archive at av.democracycounts.org, which anyone can browse — the archive is a website, not part of the app.

Each submission moves through a few states, and the My Actual Vote Videos tab shows where each of yours stands:

Pending review — your recording has uploaded and is waiting to be vetted. This is the normal state right after you submit. No action is needed from you.
Approved — your recording passed vetting and has been published to the public archive. This is the outcome you’re aiming for.
Practice — a reviewer determined the recording isn’t a real poll tape (for example, you recorded the practice tape, a blank wall, or a test). Practice recordings are kept on your own tab but are never published or shown to other users.
Rejected — the recording can’t be used (for example, it’s illegible, or it contains personal information). Rejected recordings are also visible only to you, on your own tab.

You’ll always be able to see your own recordings — including practice and rejected ones — in the My Actual Vote Videos tab. Other users and the public only ever see approved recordings.

Vetting times vary with submission volume — expect it to take longer after a major election. If your submission has been pending for longer than expected, you can email j.flatley@democracycounts.org to check on its status.

5.5.2 What happens to your recording

After you submit, your recording enters the Actual Vote analysis pipeline. You do not need to do anything further — the rest is AC’s work. Here is what happens:

Upload. Your video and its metadata (GPS coordinates, timestamp, precinct identifier) reach the Actual Vote Video Archive backend. If you recorded offline, the upload happens automatically when your device reconnects.

Vetting. A human reviewer watches your recording and checks it for legibility (can the text be read?), correct tape type (is this a results tape, not a zero tape or configuration report?), and privacy (are any faces, names, addresses, or license plates visible?). Vetting is a substantive review, not a rubber stamp. See Section 4.3 for the full vetting process.

Approval or feedback. If the recording passes vetting, it is classified as approved and becomes visible in the public Actual Vote Video Archive. If there is an issue — a section that’s too blurry to read, a face visible in the background, an ambiguous tape type — you may be contacted and asked to provide clarification or, if timing permits, to re-record.

Transcription. An AC analyst watches your approved video and manually transcribes the vote totals into a structured spreadsheet. Each contest name, each candidate name, and each vote total is entered by hand. Transcription is currently done manually because automated OCR is not reliable enough for the thermal-printed fonts used on most poll tapes. The manual process is slower but more accurate, and any uncertainties are flagged for review rather than silently resolved by an algorithm.

Comparison. Your transcribed values are matched against the jurisdiction’s officially reported results for the same precinct. AC builds comparison logic specific to each jurisdiction and election, accounting for how the jurisdiction structures its official results, how voting methods are categorized, and how aggregation rules work. See Section 8 (Methodology) for the full comparison pipeline.

Investigation. Any non-match between your tape and the official results is systematically investigated using the seven-check process described in Section 8.9. Most initial non-matches resolve as artifacts of AC’s own process — a transcription error, a matching error, or an aggregation logic issue. Genuine discrepancies are documented, investigated further, and reported.

Publication. Confirmed findings are published in AC’s analysis reports, with the underlying video evidence available in the public archive. See Section 12 (Findings and Reports) for how findings are characterized, presented, and communicated.

5.5.3 If AC contacts you

AC may reach out to you after election day for several reasons:

A vetting issue — a section of your recording is unclear, there’s PII visible that needs to be addressed, or the tape type is ambiguous.

A precinct identification question — AC needs help determining which precinct your recording corresponds to, especially if the precinct identifier you entered doesn’t match the jurisdiction’s official designations.

Context about conditions at the polling place — if your recording reveals something unusual (an anomaly on the tape, an unexpected format, a discrepancy that might be explained by local conditions), AC may ask what you observed on-site.

5.6 — Architectural Decisions and Why They Matter

The Actual Vote app is simple by design. This section explains the reasoning behind each design choice. Understanding why the app works the way it does helps users appreciate what might otherwise seem like limitations, and helps technical or institutional readers evaluate the system’s integrity.

5.6.1 Why video, not photos

A video of a poll tape provides continuous visual evidence. Every line is captured in sequence, and the recording shows the physical context: the tape is posted on a wall, it is one continuous document, nothing has been spliced or rearranged between frames. A photo could capture a portion of a tape, but it wouldn’t show the full document in context, and stitching multiple photos together introduces questions about whether anything was missed or altered between shots.

Video also captures audio, which provides an ambient record of the environment at the time of recording. This is useful for establishing context if the recording’s authenticity is ever questioned — background noise, poll workers’ conversations, ambient sounds all help establish that the recording was made at a real polling place during real closing procedures.

Finally, video handles the variable length of poll tapes naturally. A short ballot might produce a tape two feet long; a long ballot in a large jurisdiction might produce several feet. With video, the user simply pans from top to bottom at whatever pace the tape requires. With photos, the user would need to decide how many photos to take, how to frame each one, and how to ensure overlap between frames — all decisions that introduce error.

5.6.2 Why the app is single-purpose

The Actual Vote app does one thing: it records poll tapes and submits them to the archive. It does not transcribe results, display official vote counts, compare candidates, show predictions or polling data, provide voter information, or communicate with election officials. This is deliberate.

A multi-purpose app that combined recording with analysis, advocacy, or voter information would compromise the system in several ways. Election officials who permit poll tape access are more likely to cooperate with an organization whose tool is obviously limited to evidence collection. App store reviewers are less likely to flag a neutral recording tool than a politically-engaged application — a lesson underscored by the November 2024 Google Play rejection. Legally, the argument that Actual Vote is a transparent, non-partisan evidence-collection system is strongest when the app’s functionality matches that description exactly.

The single-purpose constraint also protects the evidentiary chain. If the app processed, annotated, or analyzed recordings before upload, the defense that “the video shows exactly what was physically present at the polling place” would be weakened. By keeping the app as a pass-through — the device’s camera captures the image, the device’s GPS captures the location, the device’s clock captures the time, and the app transmits all of it unmodified — the recording’s integrity is maximally defensible.

5.6.3 Why metadata is captured automatically

The GPS coordinates and timestamp embedded in each recording are captured by the device hardware at the moment of recording, not entered by the user after the fact. This distinction matters for chain of custody.

Automatic metadata is harder to fabricate than self-reported metadata. A user who claims to have recorded a tape at Precinct 42 at 8:15 PM can be corroborated — or contradicted — by GPS coordinates showing the device was at the Precinct 42 polling place and a timestamp showing 8:15 PM. If a recording’s authenticity were ever challenged, the automatic metadata provides machine-verifiable provenance that doesn’t depend on anyone’s word.

This is also why the app captures a manually entered precinct identifier in addition to the automatic GPS. The two serve different purposes: the manual entry provides a human-readable label (“Precinct 42 - Marion Township”), while the GPS provides machine-verifiable coordinates. Neither alone is sufficient. Together, they are robust: the manual label helps AC’s team sort and categorize submissions efficiently, while the GPS provides the independently verifiable provenance that makes the evidence defensible.

5.6.4 Why offline recording works

Election-night recordings happen at polling places: schools, churches, community centers, government buildings. These are places where cell signal is frequently weak or nonexistent — basements, cinder-block construction, rural areas with spotty coverage. If the app required a network connection to record, it would fail at exactly the moment it matters most.

The app’s architecture separates the two operations that are often bundled together: capture and transmission. Capture happens immediately, on-device, with no network dependency. The recording is saved to local storage the moment you press stop. Transmission happens later, in the background, whenever connectivity is available — whether that’s five minutes later in the parking lot, an hour later at home, or the next morning over Wi-Fi.

This offline-first design means the evidence is captured regardless of network conditions. A user who records a poll tape in a polling place basement with no cell signal has exactly the same quality of evidence as a user who records in a polling place with full LTE coverage. The upload timing is different; the evidence is the same.

5.6.5 Why there is no on-device analysis

The app does not attempt to read, transcribe, or analyze the poll tape on the device. Users sometimes ask why the app can’t tell them what the tape says, or compare it against official results in real time. There are three reasons this would be harmful.

Accuracy. Automated optical character recognition of thermal-printed poll tapes is not reliable enough for evidentiary use. The fonts, contrast, and print quality vary across machine manufacturers, firmware versions, and individual printers. Thermal paper degrades over hours — a tape recorded at 8 PM may have different contrast than the same tape at 10 PM. A false transcription displayed on-device would be worse than no transcription: it would give the user a false sense of what the tape says and could introduce errors into the evidentiary chain that propagate downstream.

Neutrality. If the app displayed a live comparison — “Your tape shows 542 votes for Candidate X, but the official results show 530 — discrepancy detected!” — it would transform the user’s experience from neutral evidence collection into real-time election monitoring. That changes the legal, political, and psychological dynamics of the interaction. A user who believes they’ve just detected a discrepancy behaves differently than a user who is simply recording evidence for later analysis. The former is more likely to confront poll workers, make public accusations, or take actions that could compromise the evidence or the user’s safety. The latter records the tape and moves on. AC wants the latter.

Chain of custody. The evidentiary value of a recording depends on it being an unaltered capture of what was physically present at the polling place. If the app processed the video in any way — even just to extract text — a defense attorney, election official, or skeptic could argue that the software introduced artifacts, altered the image, or otherwise mediated between the physical tape and the digital record. Keeping the app as a pure pass-through eliminates this argument entirely. The video shows what the camera saw. Period.

5.6.6 Why account authentication exists

Every submission is tied to an authenticated account. This is the foundation of the chain of custody: a specific, identifiable person recorded a specific tape at a verified location and time. If AC’s evidence were ever subpoenaed, challenged in court, or questioned by election officials, the ability to trace each recording to a specific submitter is what makes the evidence defensible.

Authentication also protects the system’s integrity against bad-faith submissions. Without accounts, anyone could flood the archive with fabricated recordings, recordings of non-existent tapes, or recordings designed to discredit the system. Authentication doesn’t prevent bad-faith submissions entirely, but it creates accountability: a bad-faith submitter has an identity that can be investigated.

5.6.7 Why the archive is public

[coming soon]

5.6.8 Why the app doesn’t do more

Over time, users and potential partners have suggested many features: displaying official results for the user’s precinct, showing real-time discrepancy alerts, providing coverage maps of which precincts have been recorded, coordinating users to fill gaps in coverage, integrating with social media for sharing, and adding gamification elements to encourage participation.

Each of these features would add utility. Each would also compromise something.

Displaying official results on-device could be characterized as voter influence — showing election results in proximity to a polling place. Real-time discrepancy alerts could trigger panic, premature accusations, or social media storms before AC has had a chance to investigate (and most initial non-matches turn out to be artifacts of comparison logic, not genuine errors). Coverage maps would reveal which precincts are unmonitored, creating a roadmap for anyone who wants to know where independent evidence will not exist. Social media integration would tie the app to platforms that make their own content moderation decisions, adding another layer of dependency beyond app stores. Gamification would undermine the seriousness of the evidence-collection mission.

The app’s simplicity is not a limitation. It is a design principle that protects the system’s credibility, legal defensibility, and operational reliability. The app captures evidence. Everything else happens downstream, under controlled conditions, with methodological rigor and human judgment.

5.7 — Quick Reference

5.7.1 Election night preparation checklist

Phone fully charged; portable charger packed
Actual Vote app installed, updated to latest version, and logged in
At least one practice recording completed
Polling place address confirmed
Precinct number identified (check voter registration card or state voter lookup site)
Poll closing time confirmed for your jurisdiction
Legal basis for access reviewed (see Section 6; print relevant statute if using Method 1)
Flashlight available (phone flashlight or small handheld)
Warm clothing if November and outdoor recording expected
Contact information for AC saved (j.flatley@democracycounts.org)

5.7.2 Recording technique checklist

Open app and confirm login
Position phone at top of tape; full width of tape visible in frame
Tap screen to focus on text before pressing record
Pan slowly downward — slow enough that paused frames are readable
Keep full width of tape in frame throughout
If focus drifts, pause motion, tap to refocus, continue
Capture every contest, including down-ballot races and ballot measures
Capture the signature block at the bottom
Stop recording after the last line
Enter precinct identifier (use whatever is on the tape header)
Confirm submission is uploaded or queued
If multiple tapes at this location, record each as a separate submission

5.7.3 Troubleshooting quick reference

Problem	What to do
Can’t install or update app	Install well before election day; if blocked, use native camera fallback (Section 5.1.5)
App crashes during recording	Restart app; the video may have been saved locally; re-record if possible
Can’t log in	Clear app cache; try alternate sign-in method; contact AC
No cell signal at polling place	Recording is saved locally and uploads automatically when signal returns
Upload stuck on “pending”	Check connectivity; force-quit and restart app; try Wi-Fi; contact AC if persistent
Don’t know precinct number	Enter polling place name; AC can match it using GPS and the tape header
Tape is behind glass	Angle phone slightly to reduce glare; use a flashlight at an angle rather than straight-on
Tape is in a dark area	Use phone flashlight or handheld flashlight; hold light to the side, not directly in front
Tape is folded, taped together, or partially obscured	Record what’s visible; note the obstruction in the comments field or email AC
Poll worker questions your right to record	Know the legal basis (Section 6); remain calm and polite; show printed statute if available; see de-escalation guidance in Section 10.6
Tape is very long (multi-page)	Maintain a steady pace; if the recording becomes too long for your device, submit what you have and start a new recording for the remainder
GPS location appears wrong	Enter precinct identifier manually; note actual location; AC can correct the mapping
Multiple tapes at one location	Record each as a separate submission; differentiate in the precinct identifier field
You’re not sure if this is the results tape	Look for non-zero vote totals next to candidate names; if all zeros, it’s a zero tape; see Section 4.3

5.8 — For Developers and Technical Reviewers

This section provides technical detail for readers who want to evaluate the app’s implementation, security model, or data integrity. General users can skip this section.

5.8.1 Technology stack

[coming soon]

5.8.2 Data flow

The data flow from recording to archive is designed to preserve integrity at each step:

On-device recording. The device’s camera hardware captures video and audio. The device’s GPS hardware captures coordinates. The device’s clock captures the timestamp. The app packages these together with the user-entered precinct identifier and the authenticated user’s account ID.
Local storage. The recording package is saved to the device’s local storage immediately. This is the offline-first guarantee: the evidence exists on-device regardless of network conditions.
Upload. When network connectivity is available, the app transmits the recording package to the archive backend.
Backend storage. The archive backend stores the recording package.
Vetting interface. Vetters review submitted recordings and approve or reject them for the public archive.
Public archive. Approved recordings are published to the Actual Vote Video Archive, accessible on the web at av.democracycounts.org.

5.8.3 Security model

[coming soon]

5.8.4 Open source status

[coming soon]