It may be that the old-fashioned tensor calculus has guided Einstein to some wrong assumptions on the field equations. For example, the construction of eqn.(28) of the above paper (covariant derivative and contraction with respect to a value index) is quite unnatural in a differential forms context.
A criticism that has been given by Cartan (I am grateful to Jose G. Vargas for directing my attention to it) is that the field equations proposed by Einstein may be a weaker condition than teleparallelism in the sense that these equations dom not imply the first Bianchi identity. Also, they are a weaker condition that teleparallelism because, even if they implied the first Bianchi identity, this identity would not imply that the curvature is zero, as the same Bianchi identity as in teleparallelism can be obtained with less stringent conditions.