JWST has been finding two puzzling classes of compact objects at high redshift. Little Blue Dots are compact, blue, broad-line AGN powered by super-Eddington accretion — matter falling onto black holes faster than radiation pressure should allow. Little Red Dots are compact, red, and heavily obscured, with extreme broad Hα equivalent widths that don't fit standard AGN templates. Two populations, two sets of papers, two evolving taxonomies.
Madau and Maiolino (arXiv 2602.22386, February 2026) show they are the same object.
A geometrically thick accretion disk — the natural consequence of super-Eddington flow — produces anisotropic radiation. Face-on, the observer looks down the funnel into the hot blue interior. Edge-on, the observer sees the radiation filtered through the equatorial dust screen, redshifted and attenuated. The broad-line region is equatorially concentrated, so high-inclination views produce stronger Hα relative to the suppressed continuum. The “extreme” equivalent widths that made LRDs seem exotic require only a 15% global BLR covering factor — standard for Type 1 AGN.
The model predicts specific observational signatures: LRDs should show broader lines (equatorial kinematics boost the velocity width at high inclination), weaker high-ionization emission (the funnel radiation that excites these lines is geometrically beamed away from high-inclination observers), and UV-optical continua consistent with a reddened version of the LBD spectrum. Each prediction is testable. The taxonomy either survives contact with orientation modeling or it doesn't.
What makes this finding structurally interesting: the dichotomy was never in the objects. It was in the observer's position. Face-on is blue. Edge-on is red. The data was correct. The classification was correct. The assumption that two appearances require two populations was wrong.
This happens whenever a system has strong anisotropy and the observer can't choose the viewing angle. Quasars and radio galaxies were unified this way in the 1990s — same engine, different orientation, different taxonomy that persisted for decades. The JWST case is the same lesson applied to a new population. The taxonomy felt real because the data was clean. Clean data from a biased sample generates confident wrong categories.
One object. Two names. The variable was where you were standing.