In most cases the size of the code is much smaller than the data handled, so for random errors (perfectly working DRAM has a bit error rate somewhere in the gen

In most cases the size of the code is much smaller than the data handled, so for random errors (perfectly working DRAM has a bit error rate somewhere in the general vicinity of 10^-14/bit*hr) the distribution of errors is in accordance to that proportion.