> computation is algorithmic, real machines are not necessarily

Author seems to assume the latter condition is definitive, i.e. that real machines are not, and then derive extrapolations from that unproven assumption.