State of the Platform

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. There is a picker at the top of this page that jumps to any of them. 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.

• October 10, 2026 currentThe stance total rose while the positions a voter can read fell; 8.4 million roll-call votes; New York City; all thirty-seven city positions approved.
• October 5, 2026Six cities researched and nothing published; a sponsor list that seated four members; the first certified reliability stratum; a quiz answer that kept landing in slot C.
• September 29, 2026Four states seated in eight days; the trivia bank audited collection by collection; a measurement that broke by being fixed.
• September 22, 2026Three states seated; Tennessee 1 to 115; the Maine misfile corrected; On the Record restarts; a monitor that failed into a void.
• September 11, 2026Treasury Tracker goes statewide in four states; Season 2 opens to voters; the first rise in the stance total in months.

What the edition of October 5 got right, and the one thing it could not have known. It said 1,645 rows of city research existed and that none had reached a voter, and it was right. The thirty-seven that proposed a position have since all been approved and published, which is the outcome it told readers to watch for. What it could not have known is that the figure it reported all of this in — the stance total — was counting stored rows rather than positions. That is corrected at the top of this page, not in the archive; the October 5 page stays exactly as published.

It also set itself a test, and the answer has not changed for a third pull running. Civic Trivia returned the same twelve people out of the same nineteen game starts. That is no longer a short window, and it is read as a finding in the Reach section below.

What a blank asserts, and what it does not

Being exact about what a withdrawal claims matters. A blank asserts that the row’s own reasoning does not support the rung it carries. It does not assert that the person holds no view. The commonest defect was an absence sitting beside an affiliation — a caucus, a committee seat, a scorecard percentage, an endorsement, a district, an ideological label — used to place someone who had said nothing on the subject. Several rows stated the method outright: “absence of activity on this topic places her at…”, “insufficient evidence to differentiate from baseline”.

The first tier was read by hand, one row at a time. A detector swept all 38,405 rows carrying reasoning and produced 317 voter-facing candidates; 307 were actionable, and a person read every one. 206 failed and were withdrawn; 101 passed and were left exactly as they were. Rows were spared where the absence rules out a neighbouring rung on top of real evidence, where the omission sits inside the person’s own document, and where a score came with the record behind it. An audit that withdrew everything it looked at would not be an audit.

Nothing was added to the certified corpus this cycle. One stratum of one evidence class is still the only measured part of it — 57 state-level, record-based stances, certified October 3 at 0.830 and 0.932 with no severe disagreements. News-sourced and statement-sourced evidence are not certified, nor is the local tier. The audit above is the manual counterpart to that measurement, not a substitute for extending it.

Thirty-seven rows, four days, and the first closed loop

The question this programme has never been able to answer is whether anything it produces reaches a voter. Now it has. The six cities researched last cycle produced 1,645 rows, of which thirty-seven proposed an actual position; the rest were scope findings, proof that a city may not decide the thing at all. Every one of the thirty-seven was reviewed by a person and published: Duluth 18, Saint Paul 12, Charlotte 5, Miami 1, Bradenton 1. Tallahassee produced none that survived review.

The shape of the result is the finding. Minnesota was chosen third precisely because it reserves less to the state than North Carolina or Florida do, and it held: Duluth and Saint Paul produced 30 of the 37 between them, from 17 people; Charlotte 5, Florida 2. A city with the power to act leaves a record; a city without it leaves a citation to the statute that stopped it. Both are worth publishing, and only one shades a map.

Two Washington towns went from nothing to voter-facing inside the same five days — Redmond and Duvall: 16 city seats, 13 headshots, coverage chips and banners, budgets loaded from their own filings, and 12 published positions across 7 people. The constraint this page named last cycle was the review queue, and the queue was cleared. It will be the constraint again the moment the next tranche lands. That has not been solved; it has been paid off once.

Follow the money

The third leg, alongside votes and stances. The platform holds 29.1 million individual campaign-finance contributions totaling $8.6 billion, covering 2,250 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. Unchanged this cycle: the widened matching and the 33 disputed surname links were last cycle’s work, and nothing has arrived since.

Stance research across the map

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. The map moved where the last edition said it would. Minnesota crossed from Seeded to Growing, 7 researched officials to 24, as the thirty-seven city positions were published. Florida went 12 to 14 and North Carolina 124 to 128 — real movement that does not cross a tier line — and Washington rose 161 to 168 with Redmond and Duvall. A tile shades only when an answer is published, which is why it stood still for two editions while the research was being done. Puerto Rico is an outlined tile: 161 seats seeded, no stances yet, and it was deliberately held out of this cycle’s legislative-vote load.

Hover any state for its exact count. The seeded tier is mostly governors and statewide officers; deeper tiers add legislatures, counties and cities.

Treasury is still unmeasured rather than low — but the half that matters moved again

An earlier edition published a rate here — seven people in a quarter, as 5% of 152 visitors — and withdrew it. It stays withdrawn. The events were sound; the denominator was not, because 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 people produced 160 category drills, 100 entity selections, 100 line-item views and 56 year changes.

The two ways of counting still disagree by more than three times. 82 people reached a budget address; only 22 fired any past-the-front-page event. The split explains it: 59 landed on our own budget, one click from the donate page — a transparency visit, which selects nothing and so fires nothing. That leaves 23 on some other government’s budget, and that half has now risen twice running, 16 to 19 to 23, while the donate half went 47 to 55 to 59. The day the non-donate half outgrows the donate half is the day this section gets a real denominator. It is not that day.

The purpose, mapped

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.