The Nancy Grace Roman Space Telescope — named for NASA’s first chief astronomer, the woman NASA’s own mission page calls the “mother of the Hubble Space Telescope” — is in space. It launched on 30 August 2026, and by 8 September Space.com was running the photo of the returning SpaceX booster as its shot of the day. So this is no longer a rendering. Roman’s headline number, per NASA’s mission page, is a field of view at least 100 times larger than Hubble’s, enough to catch light from up to a billion galaxies over its lifetime. It carries a coronagraph that blocks a star’s own light so it can, as Space.com‘s exoplanet explainer laid out this week, directly image planets and planet-forming disks rather than infer them — and its survey job is to run a statistical census of planetary systems and settle open questions in dark energy, exoplanets, and infrared astrophysics.

The trade is worth stating plainly: a field 100 times wider than Hubble’s buys coverage at the price of the deep single stare. Roman is a census-taker, not a portraitist — the two telescopes are complements, not rivals.

What makes the census possible now is the same thing that makes it hard to store: scale. WIRED reported this year on the largest map of space yet assembled — a 5.6-trillion-pixel image covering roughly 75 percent of the sky and cataloguing about 4 billion celestial objects. Roman feeds directly into that regime. And the moment you tabulate 4 billion objects on a curved sky, you inherit a problem architects have carried for centuries without naming it: you are flattening a sphere onto a plane, and the sky does not flatten for free.

This is where the frontier line runs straight onto your desk. Every all-sky survey has to project the celestial sphere into a 2D atlas, and every projection distorts something — area, angle, or distance — because a sphere is not developable. It is precisely the map-projection tax you already pay when you drop a curved slice of the Earth into a flat site plan. Same geometry, different radius.

There is a second, slower complication the surveyors know and most of us forget. The Earth’s axis is not fixed; it traces a slow cone. That is why star catalogues carry an epoch — J2000, for instance — and why a coordinate quoted without one drifts. The full loop of that axial wobble, the precession the ancients half-glimpsed and Hipparchus first measured, is often called the Great Year. It is not the ~2,300 years of folklore; take the IAU general precession rate of 50.29 arcsec/yr, run the arithmetic, and it comes out near 25,772 years for one complete turn. Awe is fine. Awe under arithmetic is the desk’s rule.

←TODAY: Roman is on orbit since 30 August 2026, a 100×-Hubble field feeding a sky census of a billion galaxies. →3012: those catalogues become the base layer of every model — the universe stored the way we store a building. Fulcrum: the same projection math that flattens the sky flattens your site, and both carry a datum epoch that quietly expires.

PAZ has kept an eye on this orbit before — from the THEMIS plasma-wave work NASA turned into audible data for citizen scientists in the HARP project, to the CRS-24 cargo run we read as a lesson in constraint hierarchies. Roman belongs on the same shelf: a frontier instrument whose real inheritance to the building trade is a discipline, not a gadget.

Atelier: The office lesson is about provenance, not astronomy. A survey — of the sky or of a plot — is only as trustworthy as its stated reference frame and epoch, and a coordinate handed over without one is a claim you cannot check. Monday move: open your project template and make the geodetic datum plus its epoch a required field in the model’s project-origin sheet, so no survey enters the BIM coordination without both stamped on it.

Hack: Compute how long the Earth’s axis takes to trace one full circle — the number behind every catalogue epoch. The point is that a “fixed” sky coordinate has a shelf life you can put a figure on. Take the IAU general precession rate and divide a full turn by it. Three lines settle the folklore.

arcsec_per_yr = 50.29                      # IAU general precession rate
great_year = 360 * 3600 / arcsec_per_yr    # arcsec in a full turn / rate
print(f"{great_year:,.0f} years per axial precession loop")  # ~25,772

The habit Roman rewards is the one that survives it: read the number that tells you how sure a claim is before you read the claim. Pull up your last site survey, find the datum and its epoch, and if either is missing, send it back before it becomes the invisible origin of everything you draw on top.

Source: science.nasa.gov