Digital cells, consisting of sequences of computer operations which co
de for self-replication, spontaneously emerge from a primordial ''soup
'', initially composed only of random sequences. The probability of ge
nerating a self-replicator increases with the number of operations in
its sequence. Random mutations drive the self-replicators to evolve in
to more efficient and complex organisms as they compete with other spe
cies for the computer system's two main resources, cpu time and intern
al memory. The initially large replicators first reduce their size by
eliminating useless operations. Later, their size increases again as t
hey optimize their replication efficiency by adding more subroutines.