I will describe a new non-parametric Bayesian method for reconstructing the
evolution history of the dark energy equation of state, based on applying a
correlated prior for w(z). When applied to the latest supernova, CMB, redshift space
distortion and the baryonic acoustic oscillation measurements, it mildly favors a
dynamical dark energy model which evolves across w=-1. An examination of the
Bayesian evidences does not show significant preference of this dynamical model over
a cosmological constant. However, we show that the best fit models for current data
will be well distinguished from the LCDM model by future observations.