9 4 in particular we show that the efficiency of our voting algorithm can be improved by choosing the parameters of the error correcting code to match the probability of the computational faults