The Half-Life of an Arrival Count: Dating Every Layer Before It Is Read

In the first week of November 2026, a county health office in the upper Midwest published an interactive map of refugee arrivals. The work was careful: clean geometry, a legible legend, a short methods note. It was also, in a narrow sense, a portrait of the past. The arrival counts came from a resettlement release issued in the spring. The neighborhood characteristics beneath them came from a five-year census estimate weighted toward 2022. The rental-cost layer had been refreshed in the third quarter, but its underlying survey was older still. Every figure on the page had been true at some moment. None of them described the county as it stood that morning.

This is not a failure of diligence. It is the ordinary condition of humanitarian mapping, and it deserves to be treated as a design problem rather than a secret. The refugee-settlement map is a composite instrument. Its layers move at different speeds, and the moment they are joined into one image, the viewer is invited to assume they share a single date. Most of the time, they do not.

Three Clocks Inside One Map

Every settlement layer obeys its own clock. The administrative clock governs resettlement figures: monthly and quarterly releases that describe arrivals already processed, usually with a lag of weeks to a few months, and subject to revision after the fact. The survey clock governs the census products that supply rent burden, household composition, language spoken at home, and neighborhood poverty; the five-year estimates that carry most small-area detail are rolling averages, and their effective center of gravity trails the publication date by years. The market clock governs rents, vacancies, and listings, which move fastest and are documented least completely.

A map that stacks these three clocks without comment does something quietly misleading. It takes a figure that is eighteen months old, a figure that is four years old, and a figure that is three weeks old, and renders them in the same palette at the same zoom. The user reads simultaneity. The data does not contain it.

Where a Stale Layer Becomes a Wrong Answer

The distortion is not abstract. In housing work, a rent-burden layer built on a five-year estimate can understate pressure in a metro where asking rents moved sharply in the past eighteen months; a family budgeted against the map arrives at an apartment it cannot afford. In school planning, enrollment projections built from a prior year's counts can miss a cohort of arriving children by a full grade level, and the bilingual staffing decision that follows is made against the wrong number. In health access, provider directories and clinic panels are refreshed on schedules that rarely align with either arrival data or census vintages, so a map can show a clinic as available months after its intake capacity closed.

None of these are errors of arithmetic. They are errors of implied date.

Revision Is Not Correction

Resettlement counts are living records. Agencies amend them as cases are verified, transferred, or reclassified, and a figure published in the spring may be restated by autumn. Analysts who treat the first release as final will build a map that is both accurate on its publication day and wrong by the end of the year. The same applies with greater force to census products, where every small-area estimate carries a margin of error that is often larger than the difference between two adjacent neighborhoods.

There is also a subtler problem: the difference between no arrivals and no data. Suppressed cells and unreported tracts appear on a choropleth as zero, and zero reads as absence rather than ignorance. On a settlement map, that distinction matters to anyone deciding where to place a welcome center.

A Five-Step Vintage Protocol

First, stamp every layer with two dates: the data-collection window and the release date. A publication date alone is not enough; a monthly release published in August may describe July.

Second, carry uncertainty into the legend. If a small-area estimate has a margin of error of several percentage points, say so in the legend rather than in a footnote few will read.

Third, refuse to interpolate forward. Do not stretch a 2024 estimate to 2026 for visual continuity. Where a range is defensible, display the range.

Fourth, publish a difference layer. Show what changed between the last release and this one. Change is often the more actionable quantity, and it is the one a static map hides.

Fifth, version the map. Keep a changelog, date each archived version, and link to it, so that a planning document citing the map can cite a specific vintage.

A Legend That Tells the Truth

Consider what an honest legend sounds like. Arrivals, January to June 2026, released August 2026, subject to administrative revision. Renter cost burden, five-year estimate covering 2020 to 2024, margin of error plus or minus four percentage points at the tract level. Clinic intake status, verified by phone in October 2026, not a live feed. This is unglamorous text, and it is the difference between a map that can survive a public meeting and one that cannot.

Why Open Aging Beats Quiet Aging

Advocates are sometimes tempted to present a settled picture because funders and councils respond to certainty. The temptation is understandable and the trade is poor. A map that states its vintage is harder to dismiss when a critic points out an out-of-date figure, because the map already said so. A map that conceals its age converts an ordinary limitation into a credibility problem at the exact moment the argument matters most.

Refugee Maps exists to document routes and settlement conditions for people who will act on them. That purpose is better served by a map that admits it is a photograph of several different pasts than by one that pretends to be a window. Date the layers. Show the intervals. Let the map age in the open.

A count is not a current condition until someone has said, in writing, when it was counted.