Recent experiments on a wide GaAs quantum well report a fractional quantum Hall state at filling factor ν=1/2 with an unusually large activation gap, a sharp maximum of the gap as a function of density, and a possible transition from the non-Abelian Moore-Read Pfaffian to the Abelian Halperin 331 state. We analyze this system theoretically, with self-consistent subband calculations, exact diagonalization, and a reanalysis of the published transport data. We find that the well has two relevant subbands, that with the second subband frozen out the well most likely has no incompressible state at ν=1/2, that a strict bilayer at the maxima of the charge density does not produce 331, and that an active second subband is essential: the ground state is a paired state of electrons in both subbands, on a continuous path between the Pfaffian and 331, with a neutral gap much larger than that of a one-subband Pfaffian. The position of the state on this path is controlled by electrostatic energies that the accessible system sizes contain only in part, and the standard overlaps do not identify the state at these sizes. Our results are more consistent with a Pfaffian-like state than with 331, but they are not definitive. The transport data do not support the reported magnitude of the gap; the gap we extract is smaller and comparable to the calculated two-subband neutral gap. A benchmark calculation for the Laughlin state at ν=1/3 in the same well provides an experimental test of the electrostatic treatment.