Think about it this way: you can't have 1 baby with 9 pregnant women in a month, but you can have 9 babies in 9 months with 9 pregnant women.
In this analogy, the pregnant woman is the node and the baby is the processed data (of course).
MapReduce is all about dividing tasks as small as they can be and then executing those tasks in parallel in several nodes. Most examples are of counting words because it's very straight forward to explain that you can give one page of a book to N people, have them count the words, and afterwards merge sort the sums into M counters and just sum it again.
In this analogy, the pregnant woman is the node and the baby is the processed data (of course).
MapReduce is all about dividing tasks as small as they can be and then executing those tasks in parallel in several nodes. Most examples are of counting words because it's very straight forward to explain that you can give one page of a book to N people, have them count the words, and afterwards merge sort the sums into M counters and just sum it again.