"Using Maupertuis' form of the principle of least action, Annila has calculated that the brightness of light from Type 1a supernovae after travelling many millions of light-years to Earth agrees well with observations of the known amount of energy in the universe, and doesn't require dark energy or any other additional driving force."
The full paper is available without paying from http://www.helsinki.fi/~aannila/arto/light.pdf. Me personally, unless my eyes are wonky, in the graph plotting distance against redshift the least-time path doesn't sit in the middle of the data points at the top, and I can't find a R squared goodness of fit in the paper. Nevertheless, there is still remarkable agreement without needing to add mass.
It's good to see a testable theory eliminating the need for dark energy — the article suggests that new data from Gravity Probe B should be enough to test this approach. Another testable result using the same least action approach is Mercury's precession where it varies very slightly from GR (http://arxiv.org/abs/1009.1571)."