The Mail-In Ballot Panic

00The narrative being tested

With the November midterms approaching, the mail-in fraud narrative is running in three layers, and this page tests all three:

• The 2026 claims: fraud asserted in California’s June primary while ballots were still being counted, and Maryland’s ballot mix-up framed as “500,000 fake ballots.”
• The standing claim: mail voting is inherently, massively fraudulent — 2020 was stolen through it.
• The counter-overstatement: “there is zero fraud”; “voter fraud is a myth.”

01The 2026 claims, checked

Maryland: “500,000 fake ballots”

What happened, per the Maryland State Board of Elections itself: a printing/coding error by the state’s ballot vendor sent some voters the wrong party’s primary ballot — roughly 400,000 ballots mailed May 9–14, 2026, with replacements mailed May 18–29. The state’s own page: “Due to an error in the printing process, some voters received the incorrect party ballot.” Every return envelope carries a unique identifier ensuring one ballot per voter; the SBE states flatly, “There is no risk of duplicate voting as a result of this issue.” State Administrator Jared DeMarinis: “no fake OR illegal mail-in ballots were distributed.” FactCheck.org rated the “fake ballots” framing a distortion (May 27, 2026).

Precision note: officials later clarified that an original wrong-party ballot would still count at canvass if no replacement was returned — “voided” meant superseded in the tracking system, with the one-ballot-per-voter safeguard intact. No duplicate-vote incidents from the error surfaced in adversarial search through the June 23 primary.

California: fraud asserted mid-count, evidence declined

Fraud in California’s June 2026 primary was asserted while ballots were still being counted. Attorney General Rob Bonta’s response: “What’s your evidence for the bold claim you’ve made? He has none” — adding that “every count, recount, hand count, audit and court case has demonstrated there is no widespread voter fraud.” LA County ran livestreamed, observable counting. The statutory reality the claims skip: California law requires signature comparison on every returned mail ballot (Elections Code § 3019), with a notice-and-cure process — and 2025’s SB 3 tightened those procedures effective 2026. A federal prosecutor’s office has announced election-fraud investigations; as of this writing they have produced no findings — this page will be updated when they do.

02The base rate — and the five checks that already ran

The claim “mail voting means massive fraud” has a denominator: 46,846,449 mail ballots counted in 2024 (30.3% of turnout; 43% in 2020). Against that, the peer-reviewed rate of documented fraud cases in universal-mail states is 0.31–0.49 per million voters — and states that switched to mail voting showed no increase in fraud (American Statistical Association analysis, peer-reviewed 2021, built from the News21 and Heritage case databases).

And 2020 — the narrative’s anchor — was checked five independent ways:

• Georgia’s triple check (under a Republican Secretary of State): a full hand tally of ~5 million paper ballots upheld the result (largest county discrepancy 0.73% — below the ~2% error rate expected of hand counting itself), surfacing ~5,800 uncounted ballots that narrowed, not flipped, the margin; a Trump-requested machine recount reaffirmed it; and a GBI signature audit of 15,118 Cobb County absentee envelopes found no fraudulent ballots.
• The peer-reviewed statistics: Eggers, Garro & Grimmer (PNAS 2021) tested every prominent statistical fraud claim across all 3,111 counties — each one “either not a fact or not anomalous.”
• The comprehensive audit compilation: Baltz et al. (PNAS 2025, MIT Election Data + Science Lab) — every available post-election tabulation audit, 856 jurisdictions, 27 states, 71.7 million votes: net shift ~0.007%.

03The honest kernels — quantified, not buried

• The fraud that exists concentrates in the mail channel. MIT Election Lab: “even many scholars who argue that fraud is generally rare agree that fraud with VBM voting seems to be more frequent than with in-person voting” — citing Miami’s 1997 absentee scandal and NC-09.
• States genuinely differ in rigor: Alabama and North Carolina require two witnesses or a notary on the return envelope; Oklahoma requires a notarized affidavit for standard absentee voters; California relies on signature comparison. (Only these verified examples — a broader “spectrum” generalization failed this page’s verification and was cut.)

04The argument, assembled

• The 2026 claims dissolve on contact: Maryland was a vendor printing error with a duplicate-vote safeguard that held; California was an evidence-free assertion made mid-count against a statutory signature-verification regime that 2025 legislation tightened.
• The base rate is the story: under one documented case per million voters, against 46.8 million mail ballots — and no fraud increase in states that switched to mail voting.
• 2020 was checked, five ways, including by its own accusers: hand tally, signature audit, machine recount, the Cyber Ninjas review, 64 lawsuits — same answer every time.
• The kernels stay on the page: a 1.2% rejection rate with real costs for voters, fraud concentrated in the mail channel, one voided congressional race, and genuinely uneven state safeguards.
• The accurate formulation: rare, detected, prosecuted — never outcome-changing at federal scale, once at district scale. Anyone selling you “rampant” or “zero” is selling.

05Sources

Related on this site: Voter Fraud in US Federal Elections, 2000–2024 — the interactive map of every court-verified case, including NC-09.