Empowered Vote · Progress Briefing · November 20, 2026
Where the Empowered Vote platform stands, in numbers anyone can check. This edition covers seven days. Tuesday’s edition was written and never published, so the site has carried the November 13 one all week.
We recorded 710 more winners from the November ballot. Five state legislatures account for nearly all of them, and a voter in any of those states can now look up the person who won their seat. The county tier opened too, which it had never done before.
The Compass moved much less. We wrote 1,655 positions and 1,635 of them went into the next season of research, which no voter can open yet. Twenty reached the site, for fourteen local officials. That is the whole of what a reader gained from a week of research.
Treasury Tracker closed its largest gap. Budget figures with no link to the document behind them fell to 298. Massachusetts is two thirds of the repair and California most of the rest, and the two were done to different standards.
Three readings of the same 52 boxes. The boxes never move; only the fill changes. Put a tab beside another and the gap is the work left to do.
Shading shows how many officials each jurisdiction has on file, researched or not. This is the figure the next tab is a fraction of. Three tiles gained officials and none changed shade. California went from 2,052 to 2,094, which is the November intake arriving. Florida went from 242 to 261, with Broward County and Fort Lauderdale. Arizona gained one.
Hover any state for its officials on file and how many we have researched. The 52 tiles hold 10,369 officials. We have researched 4,958 of them. Federal seats are on none of these maps, and the table further down counts them. Another 1,525 researched people sit on no tile at all. Most of those hold a federal seat or are standing for one, and 61 are neither.
Hover any state for its exact count. This map counts people we have researched, whatever season their answers sit in, so it is not the headline figure. Nine jurisdictions crossed 100 researched officials this week. Minnesota went from 27 researched officials to 187. Florida went from 23 to 171. Kansas reached 145, and North Dakota and Colorado both reached 113.
Indiana is the one tile that fell, from 135 researched officials to 109. Thirty people we had researched left office on November 3, and this map places a person by the seat they hold today. We still hold their answers. Of the 4,958 people on these tiles, 2,793 have a position a voter can see. Pennsylvania is the widest gap, with 254 researched and 20 of them public.
Hover any state for its exact count. No tile moved. Three budget filings arrived and all three are Massachusetts towns we already held, so this map is exactly what it was on Friday.
The 52 tiles hold 11,193 governments between them. The federal government makes 11,194, which is the roster counted in the Treasury section below.
All three maps were queried in the same run as every other figure here. Our refresh script checks the coverage table below against the map, jurisdiction by jurisdiction, and the two agreed on all 51 state rows again.
What this edition fixes in the one before it. A published edition is never edited; the next one says what it corrects.
We are withdrawing three Associated Press figures we printed on Friday. That edition said AP’s own state files hold 5,759 state legislative races and that AP had called 5,619 of them. It also said AP reported 319 as certified. We read all 51 of those files again for this edition, and every one answered. Between them they hold 644 races, and not one of them is a state legislative contest. What AP can settle is the US House. It has called 429 of its 435 House races. Fifty-six of those are certified.
We also said that two certified results were the first we had been able to write about. That is true of the November ballot and wrong about our records. Our records hold 491 certified results from this year, the earliest of them from an election in May. The two South Carolina congressional seats are still the only certified November results we hold.
The ballot basis changed again. Friday’s edition counted 2,343 races in the 53 elections dated November 3. The same elections now hold 2,303. Two records name 29 Oregon races that came off. Fifteen are state senate seats Oregon did not elect this year, and fourteen are duplicate rows for contests that elect several members on one line. We cannot account for the other eleven. Figures on either side of that change do not compare row for row.
The portrait licence gap widened. We reported 694 portraits with no usable rights statement, and it is 743 today. Portraits whose licence we record as unknown rose from 174 to 224, which is the whole of the change.
No edition went out on Tuesday. The job that writes this page wrote one and never published it, so empowered.vote/briefing has carried the November 13 edition all week. This edition covers the whole seven days, and every figure in it was measured again tonight.
Open to everyone, no login. Evidence-based, engaging, anti-partisan.
| Feature | Status | Journey | Where it stands | What's next |
|---|---|---|---|---|
| Empowered Compass | Live | alpha |
Where it standsA season is one round of research that voters can answer. The next one is still not public. It now holds 4,629 answers for 3,525 people. It held 2,994 of them on Friday. Gun policy is 3,131 of them and voting rights 1,448. The library carries 64 topics and the open season asks 60, which the live site confirms. One commit landed, a shared-component upgrade. | What’s nextOpen the season. 27,821 of the positions a voter can see still come from the season that closed in September, which is 88 per cent of them and has barely moved. We took no position back this cycle. No answer was written as a withdrawal, and none was rewritten into one. |
| Empowered Essentials | Live Alpha | alpha |
Where it standsType a ZIP code, get your representatives, with their funding, votes and positions. The roster of named officials grew by 49 people to 14,169, which is Broward County, Fort Lauderdale and a block of Indiana legislators. The pool of discovered names waiting for review reads 77,730. We hold 9,648 portraits, 65 more than Friday. Fifty-five of the new ones went to people who won this month. Ten commits landed, including county photographs for Los Angeles and Racine and catalog pages for four more places. | What’s nextPortraits for the 70 Puerto Rico mayors without one, owed and unchanged for a twelfth cycle. We re-derive it every time: 78 municipios carry a mayor, eight have a portrait and 70 do not. The island’s other 83 seats all have one. Then the licences, which got worse again. 743 portraits have no usable rights statement, 49 more than Friday. |
| Treasury Tracker | Live | alpha |
Where it standsEighteen commits, and the pile they were aimed at nearly cleared. Budget figures carrying no address a reader could open fell to 298. Of those, 153 are federal rows and 65 Californian. Three filings arrived, all Massachusetts towns. No payment record has been created since November 8, and the table holds 7,819,398. | What’s nextBerkeley is the oldest unfixed thing on this page. A scheduled import has refused to write a zero-dollar budget 128 times since August 30, sixteen more than Friday. It refused again last night. The refusal is the import working and the column mapping is still wrong. Dallas moved by one run each, to eight and seven empty returns, the last on November 15. |
| Civic Trivia Championships | Live | alpha |
Where it standsThe bank reconciles to the question for a third cycle. It reads 3,314, up from 3,305. We wrote 71 questions and retired 62. 43 of 46 collections are active, and 494 live questions carry a seven-mark scale. Across the questions with no scale, the right answer is spread almost evenly across the four slots, which the method note sets out. Four commits landed, every one a dependency upgrade. | What’s nextTwo collections are nearly the whole of this cycle’s movement. World News took 34 new questions and lost 32 to expiry. Climate Change gained 30 and lost 20. The War in Iran collection gained seven and lost ten, so it is shrinking. |
| Read & Rank | In Development | build |
Where it standsA blind taste test for candidates, where the question rather than the race is the unit of comparison. The ranking-screen redesign moved out of the design folder and into the app on November 18. It brings smaller figures, an alignment table and a Compass panel for people who have calibrated. That ends a run of eight editions with no product change. | What’s nextTwenty-one people revealed a ballot in the quarter, one more than Friday, on 21 race starts. Fifteen of them did so in the last 30 days. At this volume that is noise rather than a trend. It is still gated on a pool of positions that gained twenty in a week. |
| Fallacy Finders | In Development | build |
Where it standsv1.1 is 4 of 6 phases done and built the Fallacy Sleuth, which proposes timecoded fallacy tags at curation time, blind to the speaker. Quiet for a fifteenth cycle, with no commit since August 12. It is still the one repository with no remote, so its state can only be read from the machine this report is written on. | What’s nextThe thing it depends on is quiet again. On the Record took in four US House floor sessions on November 14 and nothing since. |
Connected accounts, one human one voice. No longer parked.
The pillar took 40 commits this cycle, against 172 in the Inform table. Thirty-one of the 40 are Empowered Listening, which finished a phase and then spent the week on what a rehearsal found. Civic Spaces took five. The remaining four were dependency upgrades.
| Feature | Status | Journey | Where it stands | What's next |
|---|---|---|---|---|
| Civic Spaces | Live | alpha |
Where it standsGeo-scoped communities, from neighborhood to federal. Five commits, and the one a member would notice is about counties. Somebody whose coordinates gave us no county now gets it from their address instead, so a county space loads for them. | What’s nextPhase 19 is still the named next step and still has no commit against it. It splits a slice that has grown too big, moving the most recently joined half into a new one. |
| Empowered Quests | Live | alpha |
Where it standsFact-verification quests with consensus scoring; verified facts feed Essentials. Two commits, both dependency upgrades. | What’s nextv1.6 planning. |
| Focused Communities | Paused | build |
Where it standsv1.0 shipped April 17, with per-topic forums wired to the Compass. Two commits, both dependency upgrades. | What’s nextv1.1 proper: badge ratification. |
| Empowered Listening | In Development | build |
Where it standsStructured debate infrastructure. Phase 5 finished on November 15 with the speaker rebuttal checklist. The team then ran a three-person rehearsal and spent the rest of the week on what it turned up. The timer counts down on the server’s clock rather than the browser’s. The transcript writes one row per utterance instead of one per word. A debate with no video now runs instead of showing a dead player. | What’s nextPhase 6 is deferred rather than planned, and the team wrote that down. It was voting and badges. Common Grounds waits on it, so that chain has no dated next step. |
| 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, the politician database, geofencing and elections. 137 commits, more than every other repository put together, and most of them were election results. Massachusetts, Maine and Maryland gave us 556 winners between them, with Oregon and Nevada behind. The total from the November ballot is now 1,945.
The county tier is no longer empty. It held no winner on Friday and holds four today. Two are Multnomah County in Oregon, read from a ranked-choice results report. Two are Racine County in Wisconsin, read from the county clerk’s own summary. We hold 103 county races, so four is the honest shape of it.
The winners come from three kinds of source. 1,725 of them are Associated Press calls. Ballotpedia gives us 114 and election officials 106. Twenty-one carry no recorded timestamp at all, which is a gap in our bookkeeping rather than in the result.
Most of the rest of the week went to gun-policy research that reads a legislature’s own roll-call record. Fourteen states ran this cycle, among them Minnesota, Florida, Kansas and Ohio. None of it is public, because all of it lands in the next season.
The stored roll-call record did not move. It reads 11,557,651 legislative votes, the same figure as Friday and as the Tuesday before that. Recording a winner and recording how someone voted are different jobs, and only the first one happened again.
We re-ran the contradiction check for this edition. It reports 340 groups covering 2,347 rows. Friday’s run found 337 groups, and nothing has adjudicated any of them. It reads 35,320 rows a voter could meet and can key only 9,328. The check reports disagreement rather than error, and its own documentation says a mapping that is wrong for everyone who cites it produces no disagreement at all.
No version number is printed on this card, because nothing in the repository produces one.
Speaker-attributed, searchable transcripts of public meetings. Four US House floor sessions arrived on November 14, and nothing has arrived since. The totals read 203 meetings and 23,696 transcript segments. We have identified 1,666 speakers, and Los Angeles City Council carries 7,200 recorded votes. 21 meetings landed in the last 30 days. The repository took one dependency upgrade, so the intake moved without the code moving. Segments carry no creation date, so dated meetings are the evidence. The two drops earlier this year are still unexplained. Built by volunteer Chris Andrews.
The invisible work. This is what “evidence-based” costs, and we paid it.
Eight of the sixteen named rows are unchanged, and that is the check this table exists for. When they hold, the order that assigns a person to a place has not shifted under us. The national row took the largest movement, gaining 73 people. Utah gained 54 and Tennessee 37. The remainder row took 885 people, which is Minnesota, Florida, Kansas and the Dakotas arriving. The row for candidates with no seat fell by 43, the first fall we have printed there; some of that is winners taking office and moving to their own state. The table counts answers we have stored rather than positions a voter can see.
| Coverage area | Politicians | Stances |
|---|---|---|
| US Congress + national | 626 | 10,507 |
| 2026 candidates & discovered names (no seat) | 899 | 5,622 |
| California (state + local) | 386 | 5,337 |
| Massachusetts (state + local) | 302 | 4,285 |
| Texas (state + local) | 273 | 2,768 |
| Maryland (state + local) | 192 | 2,564 |
| Maine (state + local) | 186 | 1,796 |
| Virginia (state + local) | 166 | 1,178 |
| Oregon (state + local + county) | 154 | 826 |
| Wisconsin (state + county + local) | 254 | 793 |
| Utah (state + city + county) | 118 | 784 |
| Washington (state + county + city) | 179 | 608 |
| North Carolina (state + local) | 194 | 597 |
| Arizona (state + local) | 132 | 531 |
| Pennsylvania (state + local) | 254 | 439 |
| Tennessee (state + local) | 165 | 396 |
| 37 more jurisdictions (incl. Minnesota, Florida and Kansas) | 2,003 | 4,565 |
Both columns reconcile against the headline counts of 6,483 people and 43,596 stored answers. 2,534 positions stand withdrawn. Note what this table cannot show you. A position we take back adds a row here rather than removing one. It also counts the next season, which is why Pennsylvania has 439 answers behind only 20 people a voter can read. A row here is research we have done. It is not yet a position anybody can see. 2,722 sources were re-checked in a sweep that finished in April 2026, and that figure has not moved since, because every row in it was written then.
We certified nothing new, for a tenth cycle running. One group of 57 state-level positions read from the public record is still the only measured part. It was measured on October 3 and passed, with no severe disagreements between coders. The rulebook it was measured against has had two revisions since. Positions read from news reports and from politicians’ own statements are not measured at all, and neither is the local tier.
Treasury Tracker’s database holds budget data for 11,194 government entities across all 50 states, spanning fiscal years 1976 to 2028.
| How well checked | Rows |
|---|---|
| Self-reported, unaudited | 533,391 |
| Unknown | 23,304 |
| Compiled from audited figures | 19,590 |
| Not applicable | 7,958 |
| Audited to government accounting standards | 1,398 |
| Audited to another recognised basis | 28 |
Budget figures carrying no address a reader could open fell to 298 this week, from a pile four publisher families had made. It is the largest single move we have recorded on this question. A separate fix mattered as much. A figure cited without a fetch date used to show no source link at all, and roughly 24,500 of them were in that state.
Massachusetts is the half we would show anyone. The state’s own reporting service answers per town, per year and per kind of figure. So the 16,829 rows we rewrote now carry 16,829 distinct addresses, and opening the source on one town’s accounts for one year lands on that town, that year.
California is the half that is not finished. 7,684 rows now cite one of seventeen statewide files, one per year, each shared by about 480 cities. A reader who opens that source downloads every city in the state for that year. The link answers. It does not show the row.
The four statewide dashboards have not moved either. 8,622 California budgets still cite one of them, every one dated between 2017 and 2025, exactly as on Friday. We said then that we did not know why those rows were left behind, and we still do not.
We found the California problem by counting distinct addresses rather than by checking that links resolve, and the same shape shows up elsewhere. A single New York address is cited by 274,428 of our rows.
The third leg, alongside votes and positions. The platform holds 29.6 million individual campaign-finance contributions totaling $8.9 billion. They cover 2,289 federal politicians and candidates across 24 election cycles, 1980 to 2026. Those four figures now come from a totals row refreshed once a day rather than from a count over the whole table, and the row we read was written on November 18. The word federal is still the limit on everything this leg can do. Of the 956 state legislators who won this month, 27 have any campaign-finance record at all. Of the 428 US House winners, 425 do.
Who actually uses what we build.
1,079 people used the platform in the last 90 days. They ran 1,605 sessions and 4,077 page views. Visitors rose by twelve, while sessions fell by 56 and page views by 119. The average session is five minutes eight, up from five minutes two. The share who do not bounce is 57.1%, against 58.9% on Friday. A rolling window does this. A busy visitor from 91 days ago drops out and takes their sessions with them.
The main site gained twenty, its highest reading in this series, and Treasury gained six. Essentials was flat. The Compass, Civic Trivia and Read & Rank each slipped a little, and the table beside this one carries every count.
Our “clicked into an app” figure counts three events together. Two of them fired for exactly the same 126 people, which we checked rather than assumed, so they are one signal and not two. The third adds six more. 132 of the 517 people who reached the main site clicked through. That is 26%.
Does the tool get used as a tool?
Everyone who arrived, against the people who did the one thing that app exists for. Bars are the 90-day rate; counts are people.
Essentials leads the Compass for a ninth pull, 52 people to 35. Read & Rank has the highest rate on the board, on the smallest base, and fifteen of its 21 finishers did so in the last 30 days. The Compass is still much the deepest funnel. 137 people opened the quiz and 35 finished it.
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 three, and Essentials, Compass and Civic Trivia on two each.
We re-ran the hostname breakdown first, as we do every cycle, and found no new host. The same fourteen served somebody, and two known ones served nobody and stay in the filter anyway. 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.
Somebody drilled into a budget chart yesterday evening, which is the most recent use we can see. Over the quarter those few people produced 158 category drills and 85 line-item views, with smaller counts for entity selections, year changes and chart switches.
This is unmeasured rather than low, and the reason has not changed. Treasury pushes its navigation into the address bar without recording a page view, so every visitor is filed against the bare domain. A second query splits the quarter’s arrivals. 164 of them touched financials.empowered.vote, which rewrites its own address to our own budget as it loads, and 217 used the explorer hosts. 128 reached a budget address at all. Of those, 84 reached our own budget and 52 some other government’s. We are still not publishing a rate.
How much further we have to go.
4,629 finished answers are sitting where no voter can reach them, against 2,994 on Friday. Publishing them is still the single largest thing we could do for a reader. The research behind them reads a legislature’s own roll-call record, which is the one kind of evidence we have ever been able to measure.
Open Season 3, asked in four editions since November 6. Settle the 743 portraits with no usable rights statement, rising for a third edition. Measure a second group of positions, open since October 3. Fix the Berkeley import, failing since August 30. Ship Civic Spaces phase 19, named four cycles ago. Give Fallacy Finders a remote, open for nine editions. Decide whether to add state campaign-finance disclosure data, or to say on each page that we do not hold it.
929 of the 956 state legislators who won this month can show no funding picture at all, because our contributions come from federal filings. No amount of position research changes that. A reader looking at one of those pages today cannot tell whether the money is absent or simply not held by us.
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.
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.
inform.seasons was read again: Season 1 closed, Season 2 open, Season 3 still a draft. No cutover happened this cycle.source_url is null or blank; it does not test whether the address resolves, which is a separate check.www.empowered.vote and read-rank-frontend.onrender.com served nobody and stay in the filter anyway.