I still think there are other viable possibilities besides the ones mentioned. I mentioned briefly on another thread the idea that there might be emergent dynamics at play for gravitational systems with large numbers of constituents; these wouldn’t show up in calculations unless one actually derived an exact numerical solution for a discrete system (n-body with large n) rather than a continuum simplification. Unfortunately we don’t have the computational power to do this, not even in principle.
In this picture, DM would be an artefact of our own computational limitations, and not any kind of new physics - the discrepancy that leads us to postulate DM would arise because we don’t have the means to use GR correctly for this scenario. After all, the universe is under no obligation to be computable (given current computational power), even if we do know all relevant laws of nature. There are similar problems in QCD as well.