Methodology

How Denver Works Better knows what it publishes

Every figure on this site traces to a public record. Each one carries a label saying what kind of claim it is. Before it appeared here, we checked it against something the publisher printed itself. This page is that record, generated from the pipeline rather than written about it.

01What this is — and isn’t

Denver Works Better is an independent, public-records-based effort to help people see how the City and County of Denver raises and spends money. It uses official documents and clearly-labeled analysis. It is separate from the City and County of Denver and from every party and campaign, and it does not speak for any agency.

One word recurs on every page here. An appropriation is money the legislature authorized to be spent. It is never a record of money actually spent, and the two are kept apart throughout this site.

Some public records are incomplete, delayed, or inconsistent with each other. Where that is true, it is said on the page rather than papered over. The job is to make the public record legible — not to replace it, and not to claim more certainty than the record supports.

This page is generated from the same validated records the figures come from. So it cannot describe a source the site does not read, and it cannot omit one it does.

02How every number is labeled

Three different kinds of claim, kept apart so you always know which one you are looking at. Every figure on this site carries one of these chips.

FACT Taken from an official record

The number is copied from a public record you can open, with the record named and the date we read it. Nothing about it is ours except the reading.

53 of the 80 figures behind this locale.

CALC Arithmetic on facts, with the formula shown

We did the math, and the math is printed next to the number: what went in, what was done to it, and where each input came from. A CALC you cannot check is not a CALC.

26 of the 80.

PENDING Blank, on purpose, with the reason on file

Never estimated. The field is real and empty. Nobody has published the number, or we have not read the source yet. We leave it blank rather than guess. A blank has more than one cause, and the blanks section tells them apart.

1 of the 80.

NOT PUBLISHED is the sharpest kind of pending, and this locale shows it. The publisher could report the figure and does not. The field stays blank, and no estimate goes into it.

NOT ELIGIBLE is not a pending at all. The figure cannot exist at this level, so no amount of work would close it. Pending would promise something nobody can deliver.

The Honesty Framework, the open standard behind this site, defines all of these.

03Source confidence grades

Every source carries a grade, and the grade governs what the figure is allowed to do. A media-graded figure may be displayed and is excluded from every total.

A — Official record, current cycle

The publisher’s own document for the period in question, read directly.

7 sources here.

PEND — Not published

Left blank. Never estimated, never carried forward from an adjacent year or an adjacent stage.

1 figure here.

04What this locale refuses to publish

A refusal is a finding, not a gap in the work. Each of these is asserted by the harness, so it cannot be dropped by a later edit without the build failing.

Publishing any Denver citywide total without naming its quantity, or swapping one of the five for another. — the dashboard says so in these words: “five true citywide totals”, “labeled with its quantity”.
Filling the People incarcerated /100k tile from anything but a machine-readable Denver custody count. — the dashboard says so in these words: “no jail or custody population dataset”, “346 datasets, read 27 August 2026”.
Using the state-prison imprisonment series for any Denver custody figure. — the dashboard says so in these words: “Jail is not prison”, “publish a state basis under a city label”.
Publishing a vehicle theft rank for Denver, or any rank built by ordering a chosen set of cities. — the dashboard says so in these words: “no city is a member of that set”, “NOT ELIGIBLE”.
Adding the Denver Health operating-agreement jail health lines to the Sheriff’s appropriation, to each other, or subtracting them from anything. — the dashboard says so in these words: “not established from the budget book alone”, “components a publisher does not sum”.
Dividing the open-data checkbook’s payments into any appropriation figure, or presenting payments as budget. — the dashboard says so in these words: “Payments are not appropriations”.
Deriving any education figure from the city budget, adding district money to a city figure, or implying school money is inside any city total on this page. — the dashboard says so in these words: “Denver Public Schools is a separate government”, “None of its money is in any city figure on this page”.
Publishing an education figure under Denver’s name without the block-level proof that Denver Public Schools and the City and County of Denver are the same territory. — the dashboard says so in these words: “all 10,149 blocks in the City and County of Denver are served by Denver Public Schools”, “if it ever finds one, these measures get renamed for what they actually cover”.
Reading the October membership count as a count of Denver’s school-age children, or stating a total for the reverse direction that the publisher does not print. — the dashboard says so in these words: “prints no total for it”, “not children living in the city”.
Drawing a Total Program or per-pupil history for Denver Public Schools, or filling one from a document that names no stage. — the dashboard says so in these words: “One appropriated year, so no series”, “a proposal, not an enactment, under a neutral label on a current page”.
Publishing, estimating or implying an amount spent against any Denver capital project. — the dashboard says so in these words: “publishes no spending by project”, “NOT PUBLISHED”.
Showing the sum of the project rows as Denver’s capital budget, or showing it without the city’s own printed capital total beside it. — the dashboard says so in these words: “published as fund totals and no finer”, “A part of the capital budget, not the whole”.
Adding any figure in the city’s capital section to any figure in the State of Colorado’s capital section directly below it. — the dashboard says so in these words: “no figure here is ever added to a figure there”.

05Every calculation, with its formula

A CALC shows its arithmetic. These are the formulas the pipeline actually ran, not a description of them.

den-receipt

Each fund group’s share of every $100 of the 2026 all-funds GROSS budget: the group’s Exhibit 2 total divided by Total Expenditures (All Funds), times 100. The six groups sum to the printed gross EXACTLY in the publisher’s $-thousands precision, and the cents are allocated by largest remainder so the six shares sum to exactly $100.00 — rounding each share independently leaves the column a cent out. Drawn over the gross rather than the net because the publisher prints the fund breakout against the gross; the $7.64 of every $100 that is internal duplication is named on the page beside it.

den-passthrough

The Denver Downtown Development Authority’s 2026 line (project 38383, the Final book’s Capital Projects Funds summary), divided by Total Net Expenditures (All Funds), times 100: $60,000,000 ÷ $5,411,388,000 = 1.11%. Over the gross, $60,000,000 ÷ $5,858,807,000 = 1.02%. The Authority’s line is its own tax-increment revenue, so it is counted as passed through. The Denver Health and Hospital Authority Tax Fund ($67,659,469, Ballot Measure 2Q of 2024) is not: it is a tax Denver’s voters levied on themselves and dedicated by a vote, so it is counted as Denver’s own money. Denver Public Schools’ property tax is collected on the same bill but is not in either total, so it is not in this share. The page prints these shares to two decimal places; the dataset carries four.

den-capital-at-project-grain

The five capital funds the Final book publishes project by project, added: the printed Grand Total of each fund’s own “Projects by Agency” schedule. Each Grand Total is reconciled twice before it is added — against the rows printed beneath it, and against the same figure printed again as that fund’s Expenditures/Appropriations row in its financial summary. THE CITY PRINTS NO TOTAL FOR THESE FIVE FUNDS TOGETHER, which is why this is a CALC and not a FACT, and why the quantity is named in full at every occurrence: it is capital appropriated to individual projects, not the capital budget. The capital budget is larger and is published as its own figure beside this one.

den-per-resident

A 2026 citywide total divided by the Census resident population estimate for the City and County of Denver (Vintage 2025, the county row, which is the city). Three totals are divided, and each is named: Total Net Expenditures (All Funds), Total Expenditures (All Funds) gross, and General Fund Total Appropriations. Residents are not the only payers, and a per-resident figure is a scale, not a bill.

den-gf-rank

A rank over every General Fund item in the enacted Annual Appropriation Ordinance (CB25-1813): the items are sorted by their printed 2026 amount and the first is the largest General Fund line. The set is the one that sums to General Fund Total Appropriations to the dollar, so no item is missing from it. A tie at the top is refused. The rank covers the General Fund only; special revenue, enterprise and capital lines are not in it, so the result is never called the largest line in the budget.

den-crime-mvt-rate-fbi

Annual offense count reported by the Denver Police Department, divided by the population served by the department itself, as published in the same FBI response, times 100,000. The agency endpoint publishes no per-agency coverage series, so the 90% gate the state locales apply cannot be applied here; a year whose reported count falls below a non-report floor of 500 would be published as a gap instead — no Denver year does. Deliberately NOT the same calc_version as the state rate, because the denominator is a different quantity — an agency’s population served, not a state’s FBI matched population.

06Where the numbers come from

Every source below is public, and every link opens in a new tab. This is the reader’s list. The full register covers every system read across all locales, including the ones out of reach and why. It is generated from the pipeline and kept in the repository’s source register.

07What is measured about you

Nothing that identifies you. No cookies, no identifiers, and no storage of anything you type, except what you choose to send us through the forms described below. The one disclosed exception to that is aggregate, cookieless page-view counting, first-party only.

This site counts page views using first-party, cookieless analytics: aggregate counts of pages viewed, referrer, country and device class. No cookies, no fingerprinting, no identifiers, no cross-site tracking, and never any content you type. Why: watching aggregate patterns for anomalies, and for signals about what to improve.

We also measure how quickly pages load for real visitors: the timing of this page rendering in your browser, and with each timing the browser, device class, device operating system, network speed and country. No cookies, no identifiers, nothing that follows you from one page to the next, and never any content you type. Why: a page a reader cannot load is a page that publishes nothing, and we cannot fix what we cannot see.

What we count when you use the page. We also count a few interactions, not just page views: when a citation is followed out to its publisher; when the Honesty Framework is opened and from which page; when a data table is opened; how the question player on one prototype page is used; which question in that player is starred, as a total for each question and never tied to who starred it; which link is followed, and from which menu or part of the page; how far down a page a reader gets, by section; how far the guided tour is followed; what kind of answer a ZIP code typed on this site got (one county, a county line, another state and which one, a territory, a code from outside the country, too few digits) and the shape of the entry — never the code itself; when the link asking us to come to your state is opened, and which state; that updates were asked for, and from where; which folded section of a project page is opened; when the rating on a project page comes into view, and that a rating was sent — never which answer, and never the comment; how a county was opened on the Near you page (a ZIP code, the county list, the map, a saved county or a link), and whether a county picked on any page was saved on this device or forgotten, never which county; when the Help menu is opened and which of its lines is taken (a video, the guided tour, the Near you page or the help page); when the panel saying who runs this site is opened, and whether the corrections log or the About page is opened from it; the first step on a page after arrival: a second page, a ZIP code answer, a county, a project, a follow form or a rating; the first follow or rating started on a page; that a follow, a sign-up or a rating reached us, and for how many items, never the address and never the answer; when a follow did not go through, and why (the network, a refusal, or an email address that could not be used), never the address; which version of the home page you saw, while we test two versions; and which broad region you are reading from, and whether you came by a link we printed, emailed or posted. Four rules bind all of it. Every event has a name from a fixed list, and anything not on that list is dropped. Nothing you type is ever recorded — not a ZIP code, not a search, not the words of a comment; where we count one of those, we count only that it happened. Location stays coarse: a county, a region or a state — never a postal code, never coordinates, never an address. A county is named only where your postal code sits wholly inside one and enough people live there for the answer to be about a place rather than a person. Location is worked out where your request first reaches us: a county where your postal code sits wholly inside one and enough people live there, otherwise a group of counties, more often a whole state, and outside the country nothing but that. The closer location your network reports is read there and thrown away in the same breath — it never reaches this page, it is never written down, and nothing about it is kept between one page and the next. If your browser sends the Global Privacy Control (GPC) signal, or the older Do Not Track (DNT) signal, we honor it. Nothing about your visit is counted: no page view, no page-speed timing, no interaction. We do not ask our edge where you are, so no region and no county is ever known. The two counting scripts are not loaded at all. We sell nothing and share nothing with anybody, under any setting. The address bar is stripped of anything after the path before a count is sent, so a value you typed cannot reach us even by that route. Nothing is counted while somebody from this project is previewing a feature that is off for the public: those visits make no request for a location, send no event, and are not included in the page-view count.

How we handle what you send — the full disclosure. The sign-up on the dashboard, and Follow and Feedback on each project page, send what you give them to a store this project controls, and to nothing else. Ratings and comments are stored with no name, address or identifier. A rating is counted into a total, so the people who govern this place can see how residents read their spending. No rating and no total is published on this site, and a comment is never published on it. We may publish summaries of what residents tell us about a place. Those are our own writing, never your rating or your comment, and never a count, average or distribution of anybody’s. Names, emails, phone numbers and street addresses are stripped from a comment before anyone reads it. A sign-up or a follow keeps your email address and what you chose to follow, and nothing else about you. A confirmation link is sent before anything else, every email carries a one-click unsubscribe, and no email is sent until you click. We do not record IP addresses, and we do not track whether an email was opened or a link was clicked. Nothing you send is used to train or tune any model. The store is controlled by this project and by no other company. The day these forms opened is recorded in the corrections log.

Some sources are person-level: a licensing register, a custody feed. We read those as a count and nothing else, and we publish only the total.

08Updates and corrections

When a published figure turns out to be wrong, we correct it and log it publicly on the dashboard. The log says what it was, what it is now, and why it moved. It is part of the product, not an apology for it.

Read this locale’s corrections log