Galaxies used to be smaller. At fixed stellar mass, massive spheroids at z near 2 have effective radii four to six times smaller than their local counterparts, and disks shrink systematically with redshift as well, following a power law R_e proportional to (1+z) to roughly the -1.0 to -1.5 (Trujillo et al. 2007; van der Wel et al. 2014). The size evolution is among the best-measured structural trends in galaxy populations.
Explaining it strains the standard toolkit at both ends. The compactness itself is awkward: assembling 10^11 solar masses of stars into a kiloparsec-scale object by z of 2 requires dissipative collapse more extreme than hierarchical models naturally produce. The growth is equally demanding: inflating those objects by factors of several without adding proportionate mass requires a diet of carefully arranged minor dry mergers, depositing stars in the outskirts while barely touching the center, with merger rates and orbital parameters tuned to deliver the observed power law simultaneously for spheroids and the more modestly evolving disks (Hopkins et al. 2009; Conselice 2014). Each ingredient is plausible; the requirement that they conspire to produce one clean scaling across mass and type is what resists derivation.
The standing is mature data awaiting a mechanism: the scaling is settled, the question is what single process makes galaxies both born small and grown large on this particular schedule. JWST has extended the compactness measurements to still earlier epochs, sharpening the birth side of the question.