In 2024, Ballotpedia covered 76,902 American elections. 53,485 of them had nobody to beat. That is the highest uncontested rate ever recorded, and Ballotpedia says it is an undercount.
476 federal offices. Roughly 6,300 state offices. And a local layer nobody has counted since 1992.
| Federal, 7 Nov 2028 | Offices | Note |
|---|---|---|
| President / Vice President | 1 | one ticket · 538 electors · 56 allocation contests |
| US Senate — Class III | 34 | not 33. Class I 33 / II 33 / III 34 |
| US House | 435 | 2020 apportionment; 2030 Census does not apply |
| Delegates + PR Resident Commissioner | 6 | PR runs on the presidential cycle |
| Federal total | 476 | + ~858–872 scheduled major-party primaries |
| State | Offices | Note |
|---|---|---|
| Governors | 13 | 11 states + American Samoa + Puerto Rico; identical set to 2024 |
| State legislative seats | ~5,800 | 44 states, 85 of 99 chambers · 7,386 seats exist nationwide |
| State executive (incl. governors) | ~160–170 | Ballotpedia currently catalogues 125; expect it to converge up |
| State appellate judicial | ~280–310 | 2024 actual 306 · trial courts have no national count at all |
| State total | ~6,300 | derived, not published anywhere |
The last time the US government counted its own elected officials was 1992: 513,200. Every Census of Governments since publishes unit counts only. The figure circulating as "519,682 Census" is not a Census figure — it is a blog's hybrid whose components do not sum. Use 513,200 and always print its date.
This is the single most important number on this page, and almost nobody uses it.
| Uncontested rate by office, 2025 | Share |
|---|---|
| State judges | 85% |
| Mayors | 69% |
| Local council | 68% |
| School board | 61% |
| State executive | 48% |
| Federal (Nov general) | 3% |
2018–2025 average: 65%. Range 49% (2021) to 70% (2024). Source: Ballotpedia, Analysis of uncontested elections, 2024 & 2025.
Your $22–31B is defensible — on one specific definition. Two corrections before you say it in a room.
Federal + state combined, nominal. 2016/2018 are federal-only (no comparable state series). 2026 has no combined figure — only AdImpact's $11.6B ad projection, which already exceeds the 2024 presidential ad cycle ($11.1B). That is unprecedented and is the strongest single argument for real growth into 2028.
Quoting $31B as the expected figure overshoots. $31B is the high case. Say: "roughly $20.6B in 2024, on track for $26–27B in 2028, plausible range $21–31B."
And never put $22–31B next to an FEC or federal-only number. On a federal-only reading the correct range is $18–23B, and $31B is far too high. The range only works on the broad definition: federal + state + ballot measures + non-disclosing issue ads, de-duplicated.
No published 2028 projection exists. AdImpact's forward product stops at 2026. This is an extrapolation and must be labelled as one.
The FEC's own 2023–24 summary: President $2.0B + Congress $3.8B + Party $2.7B + PAC $15.7B = $24.2B. Add state and you get ~$28B — for 2024, not 2028.
The PAC line aggregates 9,233 committees and counts the same dollar up to four times as it transfers. USAFacts summing FEC data naively reported $25.3B for 2020 against OpenSecrets' de-duplicated $14.4–15.1B — a 70% overstatement from identical underlying data.
Same trap with Brennan's "$1.9B dark money": $1.3B of that is contributions to super PACs, counted again when spent. Only ~$0.5–0.6B is genuinely additive.
If someone in the room knows campaign finance, the $31B number is where they will cut you. Lead with $26–27B and show the method.
~$180M across five ventures. Four of five were killed before a single voter rejected them. That is the finding, and it is good news.
29 states of ballot access. Their own nomination threshold: 1,000 support clicks in each of 10 states — 10,000 nationally. Top finisher Buddy Roemer got 5,979. Ron Paul drew the most raw support and refused the nomination.
An organisation with $35M and ballot access in 29 states could not find ten thousand Americans — 0.003% of the population — to click a button for one human being.
They changed their bylaws mid-2012 so all money raised afterwards went to repaying the millionaires who seeded it.
$104.4M in verified c4 revenue FY2014–24, excluding seven affiliated super PACs. Burned $18.5M against $7.4M of revenue in the year it died. 25 states, 18 certified, 147 electors.
They approached roughly 30 candidates — Manchin, Hogan, Huntsman, Haley, Christie, Phillips, Haslam. Every one refused. Third Way ran a counter-operation; Matt Bennett on the record:
"We had to build the idea in the minds of the political elites that if you get involved with this, you're really risking your entire reputation and your legacy."
They never got to test whether voters wanted it.
The FEC classified them a political committee, capping individual contributions at $5,000. Their own words: "we wanted to break the dependence on big money by getting lots of small contributions from millions of members, but needed some up-front big money to generate the millions of members." They won the lawsuit in March 2010 — 26 months after they were already dead.
Cleared New York's 50,000-vote ballot-retention bar by 672 votes. New York then raised it to 130,000 votes every two years instead of four. Deleted by statute. Peak registered NY membership: 649 people.
The landlords set the rent, and the landlords are the two parties you are trying to displace.
250,232 self-reported supporters. Actual registrants in its four largest states: Utah 3,002 · California 1,126 · Florida 1,043 · Colorado 439. Seven elected officials total, nearly all local or party-switchers. Zero US House seats.
A party with no policy has nothing to trade for a vote. This is the one that died on the demand side.
The pattern: every one of them offered candidates a liability — a third-party line that meant career exile. An accreditation that hands the candidate an asset they want inverts the exact mechanism that killed all four. That is why "if you have a link, you're accredited" is a different business from everything in this graveyard.
These are the arguments a serious opponent will make. Three of the four have a real answer. One does not.
Vote Smart is the controlled experiment. Its Political Courage Test response rate went 72% (1996) → 48% (2008) → 20% (2016), because answers became attack-ad material. FY2025: $941,346 revenue against $1,477,997 of expenses. It sold its Montana ranch in 2016.
Why yours differs — but only if you sequence it right. Vote Smart asked candidates to absorb a pure cost, and silence was free. A signature that comes with distribution the candidate wants flips the sign. But that only holds if the distribution exists before signatures open. Open the list first and you have rebuilt Vote Smart with a nicer logo.
This one is better evidenced than anything in the graveyard. Boudreau, Elmendorf & MacKenzie ran exit-poll experiments in the real 2011 San Francisco mayoral election. Among Democrats, the ideology–vote correlation was 0.28 in control and 0.16 with a party endorsement cue — worse. Independents: 0.45 → 0.25. Neither party nor newspaper cues improved alignment. The same research team's forced-choice survey experiments found positive effects; the real election found null-to-negative.
Kalla & Broockman, meta-analysis of 49 field experiments: the best estimate of the persuasive effect of campaign contact in general elections is zero.
Why yours differs. You are not persuading voters. You are making the candidate carry the message — "that turns every single person campaigning for a seat into also adding to our narrative." That is a supply-side play, and supply is precisely where every dead venture failed. Do not build a scorecard that tells voters who is good. Build a register of who signed.
Chiang & Knight: persuasive power comes from surprise. Predictable endorsements move under 1% of readers; surprising ones move about 3%. Credibility requires unpredictability, and unpredictability is the opposite of a brand. Any "trusted common-sense scorecard" that reliably recommends common-sense people is by construction in the under-1% band.
The way out. A binary, verifiable test — signed / did not sign — is not a judgement, so it does not need to be surprising to be citable. This is the argument for the People's Big Beautiful Bill being the object, not a rating.
From randomised evidence: ballot order is worth +1.95 points in a nonpartisan general and +3.44 in a nonpartisan primary (California's alphabet lottery). Occupational designation — three words next to your name — is worth +3 to +4 points. Incumbency in mayoral races is worth +37 points on running-and-winning. A teachers' union endorsement tests at +6 to +8, and that is survey, not field.
A new cue is not entering an empty room. It is competing against free cues printed on the ballot and structurally privileged. There is no clean answer to this one — which is why the value has to come from candidate supply and ballot completion rather than from persuasion.
And the reality check on money: in 2024, roughly $106M was spent on seven state electoral-reform measures. They went 0-for-7. Oregon's campaign outspent its opposition 758 to 1 and lost by 16 points. Missouri banned ranked-choice voting 68–32 with $0 on the winning side. Spending is not the variable.
In an uncontested race there is no opponent to attack a candidate for signing anything. Signing costs zero. This is the free-entry pool that builds list mass — and mass is what makes the credential worth something to a candidate in a contested race later.
In 2022, 40,808 Pennsylvanians voted for Senate and skipped the state legislative line. The Pennsylvania House flipped on 63 votes in a single district. Roll-off is asymmetric — it hit Democratic candidates in 83% of chambers versus 13% for Republicans. This is a completion problem with a measurable denominator, which makes it categorically better evidenced than changing anyone's mind.
| $10M against the field | Per unit |
|---|---|
| Share of the 2028 cycle ($26.5B) | 0.038% |
| Per federal office (476) | $21,008 |
| Per state office (~6,300) | $1,587 |
| Per contested election (23,417) | $427 |
| Per uncontested election (53,485) | $187 |
To move 1% of the cycle you need 26.5× leverage. That leverage demonstrably exists — Missouri got it with $0 — but in 2024 it accrued to the side with the simpler, more legible proposition, not the better-funded one.
The stated line items do not consume $10M. Every irreversible mistake available here is an ordering mistake, and you get one first cohort.