Galaxy assembly is an unsolved problem, with ΛCDM theoretical models unable to easily account for among other things, the abundances of massive galaxies, and the observed merger history. We show here how the problem of galaxy formation can be addressed in an empirical way without recourse to models. We discuss how galaxy assembly occurs at 1.5 < z < 3 examining the role of major and minor mergers, and gas accretion from the intergalactic medium in forming massive galaxies with log M* > 11 found within the GOODS NICMOS Survey (GNS). We find that major mergers, minor mergers and gas accretion are roughly equally important in the galaxy formation process during this epoch, with 64% of the mass assembled through merging and 36% through accreted gas which is later converted to stars, while 58% of all new star formation during this epoch arises from gas accretion. We also discuss how the total gas accretion rate is measured as Ṁ = 90±40 M⊙ yr−1 at this epoch, a value close to those found in some hydrodynamical simulations.