Sotomayor, J., dissenting
Constructors, Inc., 515 U. S., at 237). Indeed, the Court gives the game away when it holds that, to the extent respondents are actually measuring their diversity objectives with any level of specificity (for example, with a “focus on numbers” or specific “numerical commitment”), their plans are unconstitutional. Ante, at 30–31; see also ante, at 29 (Thomas, J., concurring) (“I highly doubt any [university] will be able to” show a “measurable state interest”).
3
The Court also holds that Harvard’s and UNC’s race-conscious programs are unconstitutional because they rely on racial categories that are “imprecise,” “opaque,” and “arbitrary.” Ante, at 25. To start, the racial categories that the Court finds troubling resemble those used across the Federal Government for data collection, compliance reporting, and program administration purposes, including, for example, by the U. S. Census Bureau. See, e.g., 62 Fed. Reg. 58786–58790 (1997). Surely, not all “ ‘federal grant-in-aid benefits, drafting of legislation, urban and regional planning, business planning, and academic and social studies’ ” that flow from census data collection, Department of Commerce v. New York, 588 U. S. ___, ___ (2019) (slip op., at 2), are constitutionally suspect.
The majority presumes that it knows better and appoints itself as an expert on data collection methods, calling for a higher level of granularity to fix a supposed problem of overinclusiveness and underinclusiveness. Yet it does not identify a single instance where respondents’ methodology has prevented any student from reporting their race with the level of detail they preferred. The record shows that it is up to students to choose whether to identify as one, multiple, or none of these categories. See Harvard I, 397 F. Supp. 3d, at 137; UNC, 567 F. Supp. 3d, at 596. To the extent students need to convey additional information, students can select subcategories or provide more detail in
