The primary quantum mechanical equation of motion entails that measurements typically do not have determinate outcomes, but result in superpositions of all possible outcomes. The paradigmatic dynamical collapse theory - the GRW theory - supplements this equation with a stochastic collapse function, intended to collapse the superposition of outcomes into just one. But these collapses are imperfect in a way that leaves the superpositions intact. This problem is the "tails problem", and is usually formulated as an inconsistent triad between (i) GRW, (ii) determinate measurement outcomes, and (iii) the traditional interpretive principle. Consequently, many solutions revise (iii). I argue that these solutions are misguided, in part because the usual formulation misses the point. Wallace (2008) argues similarly and distinguishes the "bare tails problem" from the "structured tails problem". The latter points to the structural isomorphism of the GRW ontology with the many-worlds ontology. I argue that since GRW collapses distort the tails, a third tails problem should be distinguished, which points to the structural isomorphism of the GRW ontology with merely possible multiverse ontologies. This problem applies to alternative formulations of the collapse function. I conclude that the prospects of collapse theories are bleak.