Correct, Earth formed through accretion of a disk -- all which happen due to friction. If you have no friction and the velocities of particles are isotropic, th

Correct, Earth formed through accretion of a disk -- all which happen due to friction. If you have no friction and the velocities of particles are isotropic, then you should get something like a uniform cloud of particles (instead of a protoplanetary disk, which later condenses in planets). I guess the same dynamics apply on the scale of galaxies, which should explain the violation of the expected rotation curve. It should still be possible that those weakly interacting particles would be captured by some body; if there is enough material (plus some other assumptions?) however, they should behave more like a almost uniform fluid (and not lumps like moons or planets), making it undetectable at small scales. It'll be interesting to see how this hypothesis pans out.