Parallel thinking  

Trying to remember a phrase someone used. From thinking in series to thinking in parallel. Designing problems that can be solved by distributed processes; programming for statistical failure and feedback of many simultaneous processes—reporting back to collation—rather than one series of steps.

Consider farming a problem to Hadoop. Consider saving a single tweet to a cluster of servers. Consider asking a hundred chefs to try a recipe, though maybe that example conflates too much. Consider this:

if you delegate the same operation to three machines, get the result from all three and discard an outlier, you get both transcription services at one end and the Minority Report at the other.