Richard de Grijs expects Roman to make searching archives the core skill in astronomy

The hundredfold figure describes sky area in one pointing, not survey speed ^[7]. Mapping the same patch with Hubble would take at least 100 separate pointings ^[21]. How much sky Roman actually maps in a year depends on how long each pointing has to stay open, and de Grijs's account does not state exposure times or data rates for the instrument ^[20].

During the launch broadcast, de Grijs heard NASA administrator Jared Isaacman say Roman will study the "ecology of the universe" ^[8]. Isaacman drew a contrast between kinds of camera: some give you an overview of a forest, others zoom in on individual birds and leaves, and Roman does both ^[9]. The design claim underneath the analogy is that the detailed view and the population view now come out of the same observation, so an object and the environment it belongs to sit in one dataset ^[10].

Gaia is the precedent with numbers attached. The ESA mission ran for 11 years, from 2014 to 2025, and took more than three trillion observations of roughly two billion stars and other objects ^[17]. Divide one into the other and each object was measured about 1,500 times ^[18]. Spread across 11 years, that averages near 750 million observations a day ^[19]. The Milky Way structure work came out of those public catalogues ^[17].

Against that, de Grijs sets out the workflow the field grew up on: a proposal, a time-allocation committee, a manageable quantity of data, and exclusive access during a proprietary period while the first paper was written ^[13]. "The observer knew why each exposure had been taken," he wrote ^[14]. A survey archive reverses the order. The exposure exists first, and the question that makes it useful can arrive later, from someone who never proposed it ^[12].

Targeted observing keeps its place. Hubble and Webb remain powerful because they can study a chosen object in fantastic detail ^[15], and Roman's own coronagraph will test active optics that suppress starlight far more effectively than existing space-based coronagraphs, enough to image giant planets orbiting nearby stars ^[16].

I would state the change more narrowly than de Grijs does. For a growing share of infrared results, the limiting skill will be knowing how to interrogate a public map, and how large that share gets depends on whether Roman data carry the proprietary period the older model assumed ^[13].