# Post-Election Audit Crosswalk [[How a Ballot Is Counted]] ends at a number. [[How Election Results Are Transmitted and Reported]] moves that number out of the room. [[Election Ballot Chain of Custody Flow]] preserves the paper the number came from. [[How a Midterm Election Is Certified]] turns the number into a certificate that is hard to undo. This map is the layer that sits between the last two: the **audit** — the one procedure in American election administration whose entire purpose is to check the machine count against the paper before anybody has to live with it. Every state and the District of Columbia now has a post-election audit law. That universality is recent and it is the least informative fact about the system, because the laws are not variations on one design — they are **four different procedures that share a name**, and two of them do not count votes at all. The map was built to answer one question: **does the audit happen before or after the certificate?** An audit that finishes after the result is final looked, going in, like a performance review rather than a remedy. That premise turned out to be wrong, and the inversion is the finding. **Ohio** audits *after* certification and the statute directs that the certified results **be amended to reflect the audit**. **Wisconsin** audits *before* certification and its audit "is not binding upon election results and cannot be expanded to a full recount." Timing and teeth are **independent variables**, and the column everybody argues about is the weaker predictor of whether an audit can change anything. Part C sorts the states into the four quadrants that fall out of that. The second finding is about the numbers themselves. Nearly every published count of "how many states require risk-limiting audits" is a **definitional artifact** rather than a fact — the answer is seven, eight or nine depending on whether a statute that *authorizes* an audit counts as requiring one, and whether a phased mandate that arrives mid-cycle counts from the statute's date or the first election it governs. The disagreements are named in the Notes rather than resolved, because the sources genuinely disagree and the vault should not launder that into a single number. > [!note] How to read this crosswalk > `■` marks a part · `▸` marks a design, column or lane · `⚡` marks what switches the audit on · `→` marks what it actually checks · `⏱` marks the timing against the certificate · `⚖` marks whether the finding binds · `⊘` marks a duty stated with no remedy attached · `✱` marks the structural point of a block · `⚑` marks something that **changed for November 2026** · `◈` marks **a count that is a definitional artifact rather than a fact**, explained in the Notes · `[bracketed]` notes are dated caveats. Everything inside the fence is a **statute, a design or a state**, not an officeholder — this map is built to stay true as people rotate, and **no current officeholder is named anywhere on it**. Statute and organization names inside the fence are plain text; the linked versions are in the Notes and See also. ``` POST-ELECTION AUDIT CROSSWALK — the four designs, and the two columns that decide whether any of it binds (as of 4 October 2026) │ Legend: ▸ design / column ⚡ trigger → what it checks ⏱ timing vs. the │ certificate ⚖ does the finding bind ⊘ duty with no remedy │ ⚑ changed for November 2026 ◈ count is a definitional artifact │ │ THE ONE-LINE VERSION: all 51 jurisdictions audit, and in most of them the audit │ cannot change anything. TIMING AND TEETH ARE INDEPENDENT — see Part C. │ ■ PART A — THE FOUR DESIGNS (what is actually counted) │ ├─▸ 1. TRADITIONAL FIXED-PERCENTAGE ............ 36 states + DC [the modal design] │ ⚡ the statute, automatically — no margin condition, no official's discretion │ → hand-count a FIXED share of precincts, machines or ballots drawn at random, │ and compare that hand tally against the machine tally for the same units │ → the share is written in advance and does NOT move with the margin: │ 1% (Arizona, New York) · 3% (West Virginia) · 5% (Alaska) · 10% (Hawaii, │ Connecticut) │ ✱ THE DEFECT IS ARITHMETIC, NOT POLITICAL. A fixed percentage is wasted work │ in a landslide and far too little in a close race — the only case where the │ answer could change an outcome. A 1% sample does not resolve a 0.2% margin │ at any confidence, and no amount of good faith fixes that. │ ⚖ mostly advisory; the escalation thresholds that give some of them teeth are │ in Part D │ ├─▸ 2. RISK-LIMITING AUDIT (RLA) ............... ◈ 7 require · 6–9 permit or pilot │ ⚡ the statute, and then the MARGIN sets the workload rather than a fixed share │ → draw paper ballots at random and keep drawing until the evidence reaches a │ stated confidence that the reported winner really won — the RISK LIMIT, the │ maximum chance a wrong outcome survives the audit uncorrected │ → if that confidence is never reached, the audit ESCALATES TO A FULL HAND COUNT │ by its own logic, with no official deciding to escalate it │ → three methods: BALLOT-POLLING (tally a sample, test the lead) │ BALLOT-COMPARISON (match each sampled paper ballot to the │ system's record of that ballot — needs a cast vote record) │ BATCH-COMPARISON (hand-count whole batches against subtotals) │ ⚑ risk limits are not uniform and are moving: Georgia set SIX PERCENT for 2026, │ dropping to five percent or less thereafter [SB 3EX, 2026 special session] │ ✱ THIS IS THE ONLY DESIGN ON THE MAP WHOSE WORKLOAD SCALES WITH THE STAKES, │ and the only one that can reach a full hand count without anybody's consent. │ ◈ the require/permit counts are contested — see Notes 2 and 3 │ ├─▸ 3. PROCEDURAL AUDIT ........................ 7 states │ Indiana · Michigan · Mississippi · Nebraska · New Hampshire · Oklahoma · │ South Carolina │ ⚡ the statute, on a fixed cycle │ → checks whether the PROCESS was followed: forms completed, seals logged, │ reconciliation performed, chain-of-custody paperwork present │ → DOES NOT RECOUNT VOTES and is not designed to detect a wrong tally │ ✱ A PROCEDURAL AUDIT IS AN AUDIT OF THE PAPERWORK ABOUT THE BALLOTS, NOT OF │ THE BALLOTS. It is a real control — it is how a broken seal or a missed │ reconciliation surfaces — but a state with only this design has no │ statutory procedure that compares paper to machine at all. │ ⚖ advisory by construction; there is no tally for a finding to correct │ └─▸ 4. FULL HAND COUNT ......................... not a sampling design at all ⚡ in most states, a RECOUNT trigger or a candidate's request — not an audit ⚑ → GEORGIA MADE IT AN AUTOMATIC STATEWIDE DUTY FOR NOVEMBER 2026. For eight statewide offices (Governor · Lieutenant Governor · Secretary of State · Attorney General · State School Superintendent · Insurance Commissioner · Agriculture Commissioner · Labor Commissioner), where the office appears FIRST OR SECOND on the ballot and the margin is 0.5% OR LESS, the Secretary of State MUST order a full hand recount of every ballot cast statewide, replacing the machine recount counties have used. Counties finishing within 17 days are reimbursed by the state, and THE HAND-COUNTED RESULT BECOMES THE OFFICIAL ONE. [SB 3EX, signed June 2026] ✱ THE DESIGN THE AUDIT MOVEMENT'S OPPONENTS PREFER, AND THE ONE STATISTICIANS TRUST LEAST. A full hand count is slower and less accurate per ballot than a well-run sample, and it returns the count to rooms full of partisan observers. Georgia's 2026 version is narrow and margin-gated, which is why it reads as a recount rule rather than as the hand-count-everything demand the denial movement has pressed since 2020. │ ■ PART B — COLUMN ONE: TIMING AGAINST THE CERTIFICATE │ │ The question the certification map raises and does not answer. An audit that │ reports after the certificate is issued is reporting about a result that is │ already, in most states, the operative one. │ ├─▸ BEFORE CERTIFICATION .................... 31 states + DC │ Alaska · Arizona · California · Colorado · Connecticut · Georgia · Hawaii · │ Idaho · Illinois · Indiana · Kansas · Kentucky · Maryland · Massachusetts · │ Minnesota · Missouri · Montana · Nevada · New Jersey · New York · │ North Dakota · Oregon · Pennsylvania · Rhode Island · South Carolina · │ Tennessee · Utah · Virginia · Washington · West Virginia · Wyoming │ ⏱ the audit is a PRECONDITION of the certificate, and in the strongest │ versions the canvass can be held open until the audit finishes │ ├─▸ AFTER CERTIFICATION ..................... 9 states │ Arkansas · Delaware · Florida · Iowa · New Hampshire · New Mexico · Ohio · │ Vermont · Wisconsin │ ⏱ the certificate issues first; the audit reports into a settled result │ ⚠ [Wisconsin is listed here by NCSL and BEFORE by Verified Voting — its │ timing is set by COMMISSION POLICY rather than statute. See Note 4.] │ └─▸ NO SPECIFIED DEADLINE ................... 10 states Alabama · Louisiana · Maine · Michigan · Mississippi · Nebraska · North Carolina · Oklahoma · South Dakota · Texas ⊘ the state commands an audit and names no date by which it must be done, which means nothing in the statute prevents it from landing after the certificate — or, in principle, long after ⚑ [Texas's 2026 risk-limiting audit is scheduled for 19–20 NOVEMBER 2026, and the hand-count audit that follows it must begin no later than the first business day after the Secretary of State certifies the RLA complete and finish by the 30th day after the election. The sequence is pinned to the ELECTION DATE, not to the certificate.] ◈ [NCSL's own prose labels these buckets 10-after and 9-no-deadline while its lists name 9 and 10. The lists are used here. See Note 1.] │ ■ PART C — COLUMN TWO: TEETH, AND THE FOUR QUADRANTS │ │ ✱ THE FINDING. Timing does not predict teeth. These are two independent columns, │ and all four combinations exist in the United States right now. The worked │ example in each quadrant is given with its citation; the quadrant populations │ are not published anywhere and are NOT asserted here. │ ├─▸ QUADRANT 1 — BEFORE + BINDING .............. the audit as an actual check │ ▪ COLORADO .............. Colo. Rev. Stat. § 1-7-515; 8 CCR 1505-1:25 │ → statewide BALLOT-COMPARISON RLA; the audit unit is the INDIVIDUAL BALLOT │ → target contests selected by the Secretary of State by the Friday after │ election day; the RANDOM SEED is drawn at a PUBLIC MEETING on day 13 │ ⏱ county audit boards must report by the day before the canvass deadline, │ and the Secretary of State MAY DIRECT COUNTIES TO DELAY THE CANVASS │ until the audit is done │ ⚖ the audit runs until the risk limit is met OR A FULL HAND COUNT RESULTS; │ where it produces a recount, THE RECOUNT RESULT IS BINDING │ ✱ the only design on this map where the certificate waits on the evidence │ ▪ MARYLAND ............. Md. Code, Elec. Law § 11-309 [HB 426 / SB 313, 2025] │ → RLA of at least one statewide contest after every statewide election │ ⏱ must be COMPLETED BEFORE certification by the Board of State Canvassers │ ⚖ BINDING: on a discrepancy the state administrator must direct the │ relevant board of canvassers to investigate immediately "to determine an │ accurate election result" │ ⚑ newest mandate on the map — enacted 2025, so November 2026 is the first │ federal general election it governs │ ├─▸ QUADRANT 2 — BEFORE + ADVISORY ............. good clock, no remedy │ ▪ WISCONSIN ............ Wis. Stat. § 7.08(6) │ → ~10% of REPORTING UNITS drawn at random, at least one audit in each of │ the 72 counties, at least five units per type of voting equipment; │ three contests drawn at random plus the top-of-ballot contest │ → the audit measures VOTING-SYSTEM ERROR RATE, which is what it was │ written to measure │ ⚖ "not binding upon election results and cannot be expanded to a full │ recount." On a discrepancy the commission may ask the EQUIPMENT │ MANUFACTURER to investigate. │ ⊘ the finding has no addressee who can act on the result │ ✱ A STATE CAN DO THE AUDIT AT THE RIGHT TIME AND STILL BUILD NO REMEDY. │ ├─▸ QUADRANT 3 — AFTER + BINDING ............... remedy with no clock │ ▪ OHIO ................. Ohio Rev. Code § 3505.331; Election Official Manual │ ch. 11 § 11.03 │ → choice of design: RLA, a 5%-of-ballots percentage audit, or an │ alternative approved by the Secretary of State; at least three contests │ ⏱ begins NO SOONER than 6 business days and NO LATER than 21 days AFTER │ the official certification of results │ ⚖ escalate if county accuracy is under 99.5% at margins of 1% or more, or │ under 99.8% at margins below 1% │ ⚖ AND THE CERTIFICATE IS NOT FINAL AGAINST IT: where the audit differs │ from the certified result, "the official certified results are to be │ amended to reflect the results of the audit" │ ✱ THIS IS THE BLOCK THAT BREAKS THE PREMISE OF THIS MAP. Ohio audits after │ the certificate and can move it anyway, because the statute wrote an │ AMENDMENT MECHANISM. Timing was never the load-bearing column — the │ existence of a mechanism to change the number is. │ └─▸ QUADRANT 4 — NO DEADLINE + ADVISORY ........ the modal American audit → a fixed-percentage or procedural audit, on no stated deadline, whose finding is reported and filed ⊘ no clock, no amendment mechanism, no escalation threshold, no addressee ✱ THE HONEST SUMMARY OF THE SYSTEM: the universal post-election audit law is real, and in the typical state it produces a document rather than a remedy. │ ■ PART D — THE REMAINING COLUMNS (where the variance actually lives) │ ├─▸ WHO PICKS THE CONTEST, AND WHETHER THE METHOD IS PUBLIC │ → the Secretary of State alone — the most common design │ → county or local boards, drawing their own random selection (Ohio) │ → a PUBLIC RANDOM SEED drawn in a noticed meeting (Colorado, day 13), the │ strongest transparency design on the map because the sample can be │ recomputed by anyone afterward │ ⚑ → GEORGIA 2026: one additional audited contest is chosen by MAJORITY VOTE OF │ FIVE OFFICIALS — the Governor, Lieutenant Governor, Speaker of the House, │ Senate Minority Leader and House Minority Leader — on the Thursday after │ election day, and if they do not agree in time the chair of the State │ Election Board decides alone. [SB 3EX, 2026] │ ✱ A BIPARTISAN FIVE-MEMBER PANEL WITH A SINGLE-OFFICIAL FALLBACK IS A │ DEADLOCK-TO-DISCRETION DESIGN. Failing to agree is how the choice moves to │ one person, and nothing requires anyone to agree. │ ├─▸ WHAT GETS AUDITED │ → the TOP-OF-BALLOT contest, always, in nearly every design │ ⚑ → Georgia 2026 expanded the audited set from the top contest plus one to the │ top contest PLUS President, U.S. Senate and U.S. House where applicable │ PLUS Governor, Lieutenant Governor, Secretary of State and Attorney General │ PLUS one more chosen as above [SB 3EX, 2026] │ → three contests minimum (Ohio, Wisconsin — Wisconsin's three drawn at random) │ → at least one statewide contest (Maryland) │ ⊘ DOWN-BALLOT AND LOCAL CONTESTS ARE THE UNAUDITED MAJORITY of races on an │ American ballot, in every state, under every design on this map │ ├─▸ WHAT THE AUDIT UNIT IS — and it decides how much work the audit is │ → the INDIVIDUAL BALLOT ...... ballot-comparison; needs a cast vote record │ → the BATCH ................. batch-comparison; coarser, no CVR needed │ → the PRECINCT or REPORTING UNIT ... the fixed-percentage designs │ → the MACHINE ............... the device-level audits │ └─▸ THE ESCALATION THRESHOLDS (where a fixed-percentage audit grows teeth) → WEST VIRGINIA ... a discrepancy over 1% triggers a FULL HAND RECOUNT → MINNESOTA ....... a difference over 0.5% EXPANDS the audit's scope → OHIO ............ accuracy under 99.5% (or 99.8% in a close race) requires a SECOND audit sample ✱ THESE ARE THE THREE SHAPES A THRESHOLD CAN TAKE — jump straight to a full count, widen the sample, or draw again — and a state with none of them has written an audit that reports a discrepancy and stops. │ ■ PART E — WHAT NO AUDIT ON THIS MAP REACHES │ ├─ ELIGIBILITY. Every design audits the TALLY OF BALLOTS THAT WERE COUNTED. None │ audits whether the right ballots were counted — the acceptance decisions on │ mail and provisional ballots sit upstream and are reviewed, if at all, by a │ different procedure in a different statute. │ ├─ THE CHAIN OF CUSTODY. An audit presumes the paper in the box is the paper that │ came out of the machine. That presumption is the subject of a different map, and │ where it fails, a clean audit is a clean audit of the wrong objects. │ ├─ BALLOTS NEVER CAST. Nothing in an audit speaks to the voter turned away, the │ mail ballot that did not arrive, or the registration cancelled in August. │ ├─ THE DOWN-BALLOT MAJORITY. See Part D; it is not a gap in one state's design, it │ is the design everywhere. │ └─ FEDERAL LAW. ⊘ THERE IS NO FEDERAL POST-ELECTION AUDIT REQUIREMENT. The Help America Vote Act regulates the voting SYSTEM — Title III standards, the paper record, the error-rate ceiling — and commands no audit of any result. The Election Assistance Commission publishes audit guidance and runs a voluntary standards program. The entire audit layer of American elections is state law, which is why this is a crosswalk and not an org chart. ``` ## Notes 1. **The source of the state counts, and a defect in it.** The three timing buckets and the design counts come from the [[National Conference of State Legislatures]] survey of post-election audit laws, last updated **24 September 2026** — the freshest fifty-state survey available and the one that establishes that all fifty states and DC now have an audit law. Its prose and its own lists disagree: it labels the after-certification bucket **10 states** and the no-deadline bucket **9**, while naming **9** states in the first and **10** in the second. The lists are used on this map, because a named list is checkable and a parenthetical is not. The totals reconcile either way — 31 + 9 + 10 = 50 — so the discrepancy is a labelling error rather than a missing state, but it should be cited as a list, never as a count. 2. **"Seven states require risk-limiting audits" is a definitional artifact.** NCSL's September 2026 survey names seven: Colorado, Georgia, Maine, Maryland, Nevada, Rhode Island and Virginia. [[Verified Voting]]'s state database describes **Maine** differently — its statute, Me. Stat. tit. 21-A § 726, *authorizes* audits and "does not require" them, with a mandatory pilot after the 2024 general election and a February 2025 report recommending audits where the margin is under 4% in races not already subject to recount. Those two descriptions are not reconcilable by reading harder; one counts an authorization as a requirement and the other does not. The [[Movement Advancement Project]]'s mid-2026 count, already cited on [[Risk-Limiting Audit]], resolves it the other way again: it drops Maine and adds **Texas**, for a different seven. 3. **Texas is the other half of that problem, and it cuts the opposite way.** Texas SB 598 (2021) phased RLAs in with statewide application **by 2026**, which is why NCSL still files Texas as a pilot. The Texas Secretary of State's own 2026 election-seminar materials describe a risk-limiting audit "for each election on a statewide race or measure," scheduled for **19–20 November 2026**, conducted by the general custodian of election records, with notice posted in the office and on the website and the proceeding open to poll watchers; the post-election hand-count audit must begin no later than the first business day after the Secretary of State certifies the RLA complete, and finish by the 30th day after the election. By the statute's calendar Texas is a mandatory RLA state for this cycle. By the surveys' snapshot it is a pilot. **Both are accurate statements about different moments**, which is the general hazard in citing any audit count mid-phase-in. 4. **Wisconsin's timing is administrative, not statutory, and that is why the sources split.** NCSL files Wisconsin in the after-certification bucket; Verified Voting describes the § 7.08(6) audit as generally completed **before** certification, with a Wisconsin Elections Commission policy permitting completion afterward where a statewide contest qualifies for a recount. Both are describing the same state. The lesson generalises past Wisconsin: **where the deadline lives in a commission policy rather than in the statute, the timing column is only as stable as the commission**, and the commission can move it without a legislature. Any state's row in this column should be read as "as administered," not "as enacted." 5. **The binding/advisory distinction is not symmetric with the timing distinction, and that is the map's finding.** Four quadrants, all four occupied: Colorado and Maryland audit before the certificate and bind; Wisconsin audits before and does not; Ohio audits after and amends the certificate anyway; and the modal state has neither a deadline nor a mechanism. The queue item that commissioned this map predicted that the before/after column "decides everything, because an audit that runs after the certificate is issued cannot change a result." **Ohio Rev. Code § 3505.331 is the counterexample, and it is not an edge case** — it is a large state with a binding amendment provision and no pre-certification clock. The correct load-bearing column is the presence of an amendment or escalation mechanism; timing determines how disruptive exercising it would be, not whether it exists. 6. **What "binding" means varies more than the word suggests.** Maryland's mechanism is an *investigation duty*: the administrator directs the board of canvassers to determine an accurate result. Colorado's is *procedural*: the audit continues until the risk limit is met or a full hand count results, and the recount that follows is binding. Ohio's is *direct*: the certified numbers are amended. West Virginia's is a *trigger*: a discrepancy over 1% produces a full hand recount. None of these four is the same instrument, and a crosswalk that collapsed them into a "binding: yes" column would be lying by abbreviation. 7. **Georgia's November 2026 rules are new law and should be re-verified before citation.** SB 3EX passed in a **2026 special session** and was signed in **June 2026**. It expands the audited contests from the top-of-ballot race plus one to the top race plus the federal and statewide slate listed in Part D; sets the risk limit at **six percent for 2026**, dropping to five percent or less thereafter; moves selection of the one discretionary contest to a five-official panel with a State Election Board chair fallback; and creates the **0.5% full-hand-recount duty** for eight statewide offices appearing first or second on the ballot, with 17-day county reimbursement and the hand count becoming official. The same bill postponed the state's QR-code tabulation ban to 2028 and created a voting-equipment committee — both already noted on [[Ballot Marking Device]]. Whether Georgia's RLA result is itself *binding* is **[unconfirmed]**: the statute does not say so in terms, and the state's best-known RLA, the 2020 presidential full hand tally, confirmed rather than changed the reported winner, so the question has never been tested. 8. **No officeholder is named on this map, by design.** An audit crosswalk is unusually safe ground for a reference page that has to survive an election: the content is statutes, percentages and designs, and the only people in it are offices — "the Secretary of State," "the chair of the State Election Board." The [[2026 Midterms Hub]] and [[Battleground Election Officials Hub]] carry the names, where they can be dated and sourced individually. The one place a name would have been load-bearing is Georgia's five-official selection panel, and it is written as offices for exactly this reason: four of the five seats are on the ballot on 3 November 2026. 9. **What this map deliberately excludes.** Recount law as such — the candidate-initiated and automatic recount tracks, their margin thresholds and who pays — is [[Recount]], and is a different instrument with a different purpose: a recount is a remedy a *party* invokes, an audit is a check the *state* runs on itself. Equipment certification is [[How a Voting System Is Certified and Decertified]]. The custody of the paper the audit samples is [[Election Ballot Chain of Custody Flow]]. The method inside the RLA design is [[Risk-Limiting Audit]]. The acceptance decisions in Part E belong to [[Provisional Ballot]] and [[Mail-in Voting]]. 10. **A legislature-commissioned "forensic audit" is none of the four designs.** The best-known "audit" of the 2020 cycle, the Arizona Senate's [[2021 Maricopa County Presidential Ballot Audit]] run by [[Cyber Ninjas]], appears nowhere on this map, and that is the point. It was a full hand count, but not design 4. No statute authorized it, the officials who run elections did not conduct it, the contractor was chosen by one party and changed its method as it went, it had no deadline, and its finding could bind nothing. Its own count gave Biden 360 more votes than the canvass, and the result changed nothing in either direction. Read against this map, that is not a paradox. **An audit's authority comes from the rule that runs it, not from the count it produces.** A review with no rule behind it cannot confirm a result any more than it can overturn one. What it can do is impose costs: [[Maricopa County]] had to replace $2.8 million of voting equipment that had left official [[Chain of Custody]]. 11. **Questions left open.** How many states' audits have ever changed a reported outcome — no one appears to publish a count, and the answer may be very close to zero, which would be the most important number on this subject. Whether any state's audit has been completed late enough that a binding finding arrived after the winner was seated. What the escalation rate is nationally under fixed-percentage designs. Whether the Ohio amendment provision has ever been used. Whether a state with a public-random-seed design has ever had its sample independently recomputed by an outside party, which is the only thing that makes the design's transparency more than nominal. How the four quadrants actually populate — the quadrant *populations* are not published and are not asserted here, only the four worked examples. And whether any state audits a contest selected because it was close, rather than because it was at the top of the ballot: the margin-targeted audit is the obvious design and almost nobody has written it. ## See also - [[Ballot Marking Validity Crosswalk]] — the definition of a vote every hand-count audit on this map has to apply, and the ceiling on how precise any audit design here can be - [[How a Ballot Is Counted]] — the count this map checks; the audit has nothing to compare against unless that map's paper record exists - [[How a Midterm Election Is Certified]] — the certificate the timing column is measured against, and the map this one was built to complete - [[Election Ballot Chain of Custody Flow]] — what preserves the paper an audit samples, and the presumption Part E says an audit cannot test - [[How Election Results Are Transmitted and Reported]] — the unofficial number the audit is run against - [[How a Voting System Is Certified and Decertified]] — the equipment whose error rate the procedural and fixed-percentage designs measure - [[Risk-Limiting Audit]] — the method inside design 2, its three variants and its politics - [[Recount Trigger Crosswalk]] — the instrument that CAN replace a certified number, built the day after this map and the direct sequel to its finding; its three-tier structure is the inversion of this map's four quadrants - [[Recount]] — the party-initiated remedy this map is not about · [[Election Canvass]] · [[Election Certification]] - [[Cast Vote Record]] — the artifact a ballot-comparison audit cannot run without - [[Voter-Verified Paper Audit Trail]] — the precondition of every design that compares paper to machine - [[Ballot Reconciliation]] · [[Logic and Accuracy Testing]] · [[Election Management System]] — the adjacent controls, before and after the count - [[Ballot Marking Device]] — the equipment whose barcode tabulation Georgia's 2026 bill also addressed - [[Provisional Ballot]] · [[Mail-in Voting]] — the eligibility decisions in Part E that no audit reaches - [[Help America Vote Act]] — Title III, which regulates the machine and requires no audit - [[Election Assistance Commission]] · [[Voluntary Voting System Guidelines]] — the federal standards body and its voluntary program - [[National Conference of State Legislatures]] — the source of the fifty-state counts, and the subject of Note 1 - [[Verified Voting]] · [[Movement Advancement Project]] · [[Brennan Center for Justice]] — the trackers whose state databases this map reconciles - [[Election Denialism]] — the movement that rejects sampling and demands design 4 - [[2021 Maricopa County Presidential Ballot Audit]] · [[Cyber Ninjas]] — the partisan full hand count that is none of the four designs (Note 10) - [[Brad Raffensperger]] — the office that orders Georgia's audits and hand recounts under the 2026 rules - [[Battleground Election Officials Hub]] — the named officials this map deliberately leaves out - [[State Election Administration Tree]] — the offices that run every procedure above - [[Elections Clause]] — why the audit layer is state law in the first place - [[2026 Midterms Hub]] — the cycle this map is built for - [[Foundations Hub]] — parent hub for the map series