I don't think that discreteness of machines is a strong argument against the possibility of creating a thinking machine (here with thinking I mean human thinking). It seems that your argument is a bit similar to Turing's argument on page 15 of [1]. He states that since the nervous system is continuous it cannot be modelled by a discrete machine (Turing then continues by writing that a continuous or discrete system is not of importance to the imitation game since you can approximate continuous systems by discrete ones). In terms of current scientific development this argument is not very strong since nowadays we can create pretty good models of isolated nervous systems (at least on a microlevel) this opens the gate to making a brain model.