Empowered Vote · Volunteer Briefing · September 29, 2026
Every number on this page comes from the live database, verified git history, or our product analytics, not from estimates. We build tools for shared facts; our progress report should be one. Each briefing is written fresh and kept at its own dated address, so you can read what we said before and check it against what we say now.
North Dakota, Kentucky, Kansas and South Dakota were seated end to end in four days — 635 offices, every one filled, every one on a dated term, and 470 new legislative district boundaries under them. Four legislatures (141, 138, 165 and 105 members), four city halls (Grand Forks, Lexington, Wichita, Aberdeen) and four county governments. Each state broke a rule the last one taught us. North Dakota’s House elects two members per district, so 48 polygons carry 94 seats and every check assuming one office per district is wrong there. Its own website publishes three member lists, and the one that looks official is cumulative — 148 names for a 141-seat body. We used the September special session’s roster instead, and swept the bad list as a control: it found all seven departures, each dated to the day.
Civic Trivia had its whole question bank read, one collection at a time. It began because Chris played the Cambridge collection and was served two questions that were practically the same. Measured on September 26, the bank held 3,855 live questions across 43 collections; 42 of the 43 have now been read question by question, and the bank stands at 3,264. What came out was not mostly duplicates. It was eight questions in one collection asking the same thing with the same answer, thirteen of another’s eighty-two about the municipal airport, a question whose correct answer was not among the four offered — twice, in two different collections — two live questions that contradicted each other, a true option scored as wrong, and, in Phoenix, a plainly wrong fact live in production. Fifty-seven cited sources had gone dead; repairing them uncovered two more wrong facts behind them.
One of those wrong facts could have cost somebody their vote. A Bloomington question told players that Monroe County, Indiana uses vote centers, so a voter may cast a ballot anywhere in the county. It does not, and they may not — and in Indiana, voting at the wrong precinct means a provisional ballot that will not count for most races. That collection carried four more wrong facts, one of them living only in an explanation while the answer above it was right, which no structural rule can see. Milwaukee, audited the same day, carried none. The difference is upstream: Milwaukee’s generator is handed a block of verified facts and Bloomington’s was not, which is the clearest lever this work has found for preventing defects rather than repairing them.
And this page’s own measurement of itself broke, by being fixed. The discovery pool — names lifted from campaign-finance filings, awaiting triage — used to sit on 76,330 placeholder offices, invented so each record had somewhere to hang. On September 24 all 76,330 were archived and retired, taking the office table from 85,134 rows to 8,804. Our published split between “curated” and “pool” was computed by asking who held a placeholder, so with the placeholders gone it answered 89,903 curated, nobody in the pool. Nothing was wrong with the data: the test had stopped discriminating because the thing it detected had been removed, and that failure reads exactly like a clean bill of health. The replacement asks what a record rests on — a seat, a candidacy, or research. The two figures above are on that new basis and are not comparable with last week’s.
Each update is a dated snapshot. We do not edit one after publishing it.
This page is the current briefing, and it always lives at empowered.vote/briefing. Every edition also keeps a permanent address of its own, so a number you read here can be checked against the date it was published on. Each update is written from scratch: work that runs for a month stays the headline, but the sentences around it are rewritten every time. We never edit a published edition. When an earlier one got something wrong, the next one says what it corrects.
One correction to the edition of September 22. It promised Tennessee a row of its own in the coverage table this cycle. It has one below — 119 politicians, 231 stances — and adding it shrank the remainder row, which is the only honest way to add a row to a partition.
And this edition was corrected on the day it was published, before it was archived. It first said 37 of 43 trivia collections were audited, six had never been read, and five were under the expiring-content floor. All three were already out of date: the figures came from a project document read out of a checkout 32 commits behind, and the work it did not know about had closed every one of those six. The true figures are 42 of 43, none unread, two under the floor. It is recorded here rather than fixed quietly, because a page that claims its numbers can be checked cannot change underneath a reader in silence. The archived editions of September 11 and 22 are untouched and stay that way; this one had not been frozen yet.
Open to everyone, no login. Evidence-based, engaging, anti-partisan.
| Feature | Status | Journey | Where it stands | What's next |
|---|---|---|---|---|
| Civic Trivia Championships | Live | alpha |
Where it stands42 of 43 collections audited in four days, and almost every one carried several defects. A rules engine now checks every question the nightly pipeline writes before it is stored, and it has learned this month to catch answer options that are true at the same time, past-tense questions offering a future year, and claims resting on a rule someone else can change. Drafts bank-wide: zero. | What’s nextOne collection left, and it needs writing rather than auditing: a news bank cannot be fixed by reading it. Then the backlog nobody had counted — eleven collections audited before the 10% expiring-content floor existed were still under it, because the rules have moved three times and were never swept back over finished work. “Audited” means “met the rules as they stood that day.” Nine are cleared; two remain. The “According to…” prefix is also gone bank-wide, 1,978 to 0 — and that sweep corrupted three rows before it was caught. |
| Empowered Essentials | Live Alpha | alpha |
Where it standsZIP-to-representatives with funding, votes and stances. Seven more communities got a banner — Grand Forks, Lexington, Wichita, Detroit, Aberdeen, Biloxi and Akron. Two of them are their state’s banner photograph, and the same view cannot appear twice on one page: Michigan gave its banner up and took Pictured Rocks, while Wichita took a street on the river rather than the skyline the state banner already shows. | What’s nextElection pages got the cycle’s quiet repairs: registered write-in candidates are marked, returned contributions are shown apart from total raised, interim appointees appear on the tab for the seat they hold, and no elected judge is hidden by the wrong tab. Next: portraits for the 70 Puerto Rico alcaldes still without one. |
| Empowered Compass | Live | alpha |
Where it standsSeason 2 is what the Compass asks: 60 topics from a library of 64. Season 3 is still a draft, assembled in the open. No change under the voter this cycle — which is worth saying, because a season opening is a launch the numbers do not announce, and we now check for one every time. | What’s nextThe research standard behind every stance was written down this cycle as a single document — how a position is found, judged, sourced and refused — so it can be handed to someone who was not in the room. The number that proves it was needed: on one pilot wave, three research agents returned 38 rows, 14 looked evidenced, and 8 survived someone reading the bills. |
| Treasury Tracker | Live | alpha |
Where it stands8,144 governments, unchanged this cycle. The work was speed, and the symptom was variance — which is why a single fast measurement kept clearing it. The same page, minutes apart, took 12.1 seconds and 3.2 seconds to put a figure on screen. | What’s nextEvery page load was downloading the entire list of governments. Measured live today: that call is 3.05 MB and 1.5 seconds; the one the app now makes is 7.4 KB and 0.13 seconds. A city page renders none of the panels that needed the list. Still owed: the same five-run before-and-after on the real page, not just the call. |
| Read & Rank | In Development | build |
Where it standsA blind taste test for candidates: the question, not the race, is the unit of comparison. No commits for a third cycle. It is gated on the stance pool the research waves are filling, and the four states above added rosters rather than stances. | What’s nextCandidate quote sourcing for 2026 races, once the pool for those races is deep enough to draw from. |
| Fallacy Finders | In Development | build |
Where it standsv1.1 is 4 of 6 phases done and built the Fallacy Sleuth: curation-time AI proposing timecoded fallacy tags, with names stripped before the model sees a word and every clip re-run with the speakers swapped. Quiet for a fourth cycle. | What’s nextPhase 10 finishes the US House floor clip end to end, then rights-clean clip sourcing at volume through On the Record — which is running again. |
Connected accounts, one human one voice. Parked during the election push; resumes after November.
| Feature | Status | Journey | Where it stands | What's next |
|---|---|---|---|---|
| Civic Spaces | Live | alpha |
Where it standsGeo-scoped communities, neighborhood to federal. A second busy week in a parked pillar, all of it interface: a thread now has a URL of its own (and so a Share button again), the ten screens with no dark mode got one, and the location banner carries the population of the area you are standing in. | What’s nextPlace names now resolve offline, from a table we hold, because the Census endpoint the app called started redirecting instead of answering. A dependency that returns a redirect looks exactly like our own bug. |
| Empowered Quests | Live | alpha |
Where it standsFact-verification quests with consensus scoring; verified facts already feed Essentials. Red gem quests unlock at VR 90+. | What’s nextv1.6 planning. |
| Focused Communities | Paused | build |
Where it standsv1.0 shipped April 17: per-topic forums wired to the Compass. Waiting on the post-election Connect pivot. | What’s nextv1.1: badge ratification and argument mapping. |
| Empowered Listening | Paused | build |
Where it standsStructured debate infrastructure; 5 of 6 phases done — timers, transcription, audience notes. | What’s nextPhase 6: voting and badges. Then Common Grounds can build on it. |
| Emparks | Design | design |
Where it standsPlatform-wide annotation system, five mark types. Full design doc; no build yet. | What’s nextAwaits Connect-pillar capacity. |
| Common Grounds | Design | design |
Where it standsAdversarial-collaboration mediation. Full design doc; depends on Empowered Listening and Fallacy Finders arbiters. | What’s nextBuilds after its dependencies land. |
The shared spine every app plugs into.
Auth, tiers, XP and gems, politician database, TIGER geofencing, elections. This cycle it spent most of its effort on seating four states one seat at a time, and the rest on removing things that were never true. 10,932 districts are now mapped, up 1,743, and 470 of those are the four new legislative maps. 121,178 legislative votes and 19,622 bills on record. Every one of the 635 new officeholders carries a dated term, and where a city publishes no dates the date came out of the council’s own minutes: Wichita’s published appointment date for one member is the day of the vote, not the day the term began, and the two are a week apart.
Speaker-attributed, searchable transcripts of public meetings, synced to the original video. The restart held for a second cycle — five more meetings landed, the most recent on September 26 — and the dataset now reads 169 meetings, 21,080 transcript segments and 1,370 identified speakers, including LA City Council with 7,200 recorded council votes. The caution from last time is repeated rather than retired: segments carry no creation timestamp, so the dated meetings are the evidence here, not the segment count — and the two drops earlier this year are still unexplained. A total climbing past its old high is not an answer about why it fell. Built by volunteer Chris Andrews.
A 2026 backfill gave every US Senate candidate an open-ended term on a “Candidate for U.S. Senate” office. An open term reports its holder as the current occupant for ever. So eighteen people who had lost a primary, withdrawn or simply stopped running were still recorded as sitting members of Congress — returned by address search for every address in their state, and read as incumbents by the campaign-finance matcher. Two were caught because their federal filings went dark; the other sixteen were invisible that way, because a campaign that ends without a final filing leaves a committee that still looks alive. Those sixteen were confirmed against each state’s own election authority instead, and their terms were closed.
This shows up in our own coverage table below as a loss. The US Congress + national row fell by 15 people and 153 stances, and not one stance was retired to do it — those people simply stopped counting as federal officeholders. We would rather publish a row that fell for a good reason than carry eighteen phantom senators.
The invisible work. This is what "evidence-based" costs, and we paid it.
36,438 stance positions across 4,196 politicians and 64 topics, every one carrying reasoning and cited sources; 1,846 of the 2026 candidates already have some. One figure keeps its date said out loud: 2,722 sources were independently re-checked in a sweep that finished in April 2026, and the number has not moved since, because every row in it was written then. It is a dated achievement, not a running total.
| Coverage area | Politicians | Stances |
|---|---|---|
| US Congress + national | 556 | 9,494 |
| 2026 candidates & discovered names (no seat) | 934 | 6,038 |
| California (state + local) | 372 | 4,767 |
| Massachusetts (state + local) | 302 | 3,355 |
| Maryland (state + local) | 188 | 2,228 |
| Texas (state + local) | 270 | 2,073 |
| Maine (state + local) | 186 | 1,794 |
| Virginia (state + local) | 166 | 900 |
| Oregon (state + local + county) | 154 | 824 |
| Utah (state + city + county) | 57 | 650 |
| Wisconsin (state + county + local) | 208 | 628 |
| Washington (state + county + city) | 161 | 489 |
| Arizona (state + local) | 129 | 436 |
| North Carolina (state + local) | 124 | 406 |
| Tennessee (state + local) | 119 | 231 |
| 38 more jurisdictions (incl. Florida and DC) | 270 | 2,125 |
Eighteen new answers were written this cycle, and four rows moved. The table is rebuilt from a single query run, every politician with at least one answer lands in exactly one row, and the rows sum exactly to the headline totals — a check that catches a missing row but not a row filed in the wrong place, since a misattribution moves a count between buckets without changing the sum. The movement this week is almost all re-plumbing, not research: the closed Senate-candidate terms above, and Utah’s county row officers being split out of the chambers named for their county commissions — ten counties in which the Sheriff was filed as a member of the County Commission.
How a stance gets seated. A member is placed on a topic because of a bill they sponsored, read in its session-law text — not a sentence someone else wrote about them. That standard now exists as a document rather than as judgment held in one person’s head: what counts as a chair, what counts as a source, ten worked judgments including the refusals, and the rule that a new researcher’s first wave is read before it is pushed.
Treasury Tracker’s database holds budget data for 8,144 government entities across all 50 states plus DC, spanning fiscal years 1976 to 2028:
Two findings are worth more than the defect counts. The first is that a citation can return HTTP 200 and be worthless in five distinct ways — a soft 404 that serves a full page, a JavaScript app carrying none of the text cited to it, a moved site, a bot wall returning 403, and a challenge page that answers 200. A link checker passes all five. One step catches all five: fetch the page and search it for the question’s own figure. One collection turned out to be cited almost entirely to pages nothing automated can read; another’s explanations carried figures present in no source at all, one contradicting a live question beside it. A supported answer is not a supported question, and the explanation is what the player reads afterwards.
The second is that a collection can be about the wrong things and pass every rule. Every check asked whether a question was wrong, duplicated, leaked or stale; none asked whether it was about civic life. One city’s bank had been generated from a department website rather than about the city — the room number of the Auditor’s office, its opening hours, the colours in its logo — with none asking which river runs through it. Another carried eighteen questions about its state inside a city collection, and ten of them were labelled with the state’s own name, which made that one visible in a database query before anybody read a word.
The third leg, alongside votes and stances. The platform holds 29.1 million individual campaign-finance contributions totaling $8.6 billion, covering 2,248 federal politicians and candidates across 24 election cycles (1980 to 2026). Every record traces to its original FEC filing, and superseded lines from amended filings are retired individually rather than counted twice. The politician count nearly doubled this cycle because the matching widened, not because new filings arrived — and the same pass disputed 33 links made on a surname substring and retired the matcher behind them. A wrong link here puts someone else’s donors under a real person’s name.
Every state has at least one researched official, and all 50 congressional delegations are covered nationally. Shading shows the depth of state and local research. This week’s four states barely move it, and that is the honest reading: they gained 635 seats between them and almost no stances, because seating a legislature and researching it are two different jobs and this map shades by the second. Puerto Rico is an outlined tile — 161 seats seeded, no stances yet.
Hover any state for its exact count. The seeded tier is mostly governors and statewide officers; deeper tiers add legislatures, counties and cities.
Who actually uses what we build — measured, not guessed.
633 people used the platform in the last 90 days, across every app, over 1,656 sessions. What they do once they arrive is good — sessions average six minutes twenty and 7 in 10 don’t bounce — so the constraint is arrival, not engagement. 98 of the 314 who reached the main site clicked through into an app, 31% in the quarter, the same share as last time on a larger base.
Read every figure here as an order of magnitude, not a trend line. At this volume a couple of people either side is the entire swing, and the quarter window slides forward, so an app can “fall” because a busy week in June rolled out of it. Nothing this cycle was aimed at arrival, and it rose anyway: over six pulls the main site reads 306, 316, 312, 297, 300, 314. The hostname map was re-run before any of these figures were pulled and found no new host, which is what makes this pull readable against the last five.
Everyone who arrived, against the people who did the one thing that app exists for. Bars are the 90-day rate; counts are people.
Two of these are start problems, not finish problems. Only 18 people started a trivia game in a quarter, on 111 arrivals; 12 finished. Read & Rank is the same shape: 20 races started, 10 ballots revealed. The Compass is the opposite and is the healthiest funnel we have: 118 of its 131 visitors opened the quiz and 37 saw it through. The trivia audit above is aimed squarely at the fourth row, and it landed too late in this window to have moved it — the next pull is the one to read.
Unique visitors per app, 30 days and the full quarter. Each app is counted once across every hostname that serves it — Treasury drew traffic on four this quarter, the main site on two.
Treasury Tracker is still the second most-visited thing we run and holds the platform’s largest dataset — which is why a page that took twelve seconds to show a figure was worth a week. Civic Spaces is absent from this table, and that is a measurement gap rather than a result: it serves no traffic our analytics can see.
An earlier edition said seven people in a quarter opened any of Treasury Tracker, as 5% of 152 visitors. That rate was not sound and it stays withdrawn. The events were fine; the denominator was not — Treasury pushed its in-app navigation into the address bar without recording a page view, so every visitor was filed against the bare domain. The direction was wrong too: this quarter those few produced 126 category drills, 91 entity selections and 68 line-item views.
The fix has been recording since August 16, and the two ways of counting still disagree by nearly five times. 61 people reached a budget address; only 13 fired any past-the-front-page event. The split explains it: 47 of them landed on our own budget, one click from the donate page — a transparency visit, which selects nothing and so fires nothing. That leaves 16 who opened some other government’s budget, unchanged from 16 a week ago while the donate half grew from 41 to 47. Neither figure is published as a rate, and Treasury stays unmeasured rather than low until the half that is not ours outgrows the donate traffic. The rule it bought applies to every app here: a low count is a claim about instrumentation until you have read the code path.
How much further we have to go, honestly.
Research the seats just loaded — seven states now hold full rosters and almost no stances. Finish the 2026 candidate waves, and land Read & Rank candidate quotes once the pool can carry them. Write the one trivia collection an audit cannot fix, and clear the last two collections still under the expiring-content floor. Success in November looks like one thing: any voter in a served area can look up every name on their ballot and follow the money behind it.
Reactivate the parked pillar: Empowered Listening phase 6, Focused Communities v1.1, Civic Spaces v4.0 — whose shell, shared theme, thread links and sibling-shard view are groundwork done early. This is where the platform stops being a reference tool and becomes a community, and where our sustaining-donor model lives.
Designed but not built: the Gems stipend economy, Awareness Exchange, Badges, Symposium tournaments, Emparks, Common Grounds, and tools for citizens to run for office on the strength of their solutions. Roughly a third of the founding vision, waiting on the first two pillars.
A volunteer-built platform, visible in the commit history.
An interactive Purpose mind map walks the whole vision — the thesis, and the Why, How, Who and What — with a short explainer on every concept and pillar feature.