The cold dark matter paradigm has been extremely successful for explaining a wide range of cosmological phenomena .
Nevertheless , since evidence for non-baryonic dark matter remains indirect , all reasonable alternatives should be explored .
One proposed idea , involving a fundamental acceleration scale a _ { 0 } { \simeq } 1 – 2 \times 10 ^ { -10 } { m } { s } ^ { -2 } , is called MOdified Newtonian Dynamics or MOND .
MOND was suggested to explain the flat rotation curves of galaxies without the need for dark matter .
Whether or not it can adequately fit the available data has been debated for almost 20 years ( and we summarise many of these studies ) , but only recently have there been studies attempting to extend MOND to larger scale regimes .
We discuss how the basic properties of MOND make it at best ambiguous to apply these ideas to cosmological scales .
We emphasize the difficulties inherent in developing a full theory in which to embed the main MOND concepts .
Without such a theory there is no obviously consistent way to discuss the early Universe and the growth of perturbations .
Recent claims that MONDian cosmology works very well are therefore not supportable .
We also provide an argument for why a _ { 0 } { \sim } cH _ { 0 } naturally , a coincidence which is often suggested as a motivation for taking MOND seriously .
We discuss other alternative theories of gravity concluding , as others have , that no metric theory extensions appear workable for explaining rotation curves as well as other observed phenomena .
The whole premise of many of these attempts is fatally flawed – galaxies are not pre-selected , discrete , isolated regions which formed monolithically and around which one can construct an axially-symmetric dynamical model in order to remove the need for dark matter .
In the modern view , galaxies are part of a dynamic continuum of objects which collectively make up the evolving large-scale structure of the universe .