Data 0: the text

"Bereishit" (the Book of Genesis) as a sequence of 78,064 Hebrew letters.

Step 0->1: classifying the letters

Data 1: relevant letters

Each letter of the given word occurs in the text a lot of times.

Step 1->2: searching for the word

Data 2: ELS's

Equidistant letter sequences (ELS's) spelling out the given word.

Step 2->3: sorting ELS's

Data 3: families

Families of equally perturbed ELS's.

Step 3->4: how noteworthy is each ELS?

Data 4: domains of minimality

Step 4->5: comparing two words

Data 5: distance between two words

Step 5->6: correcting the distance

Data 6: the corrected distance

What for did I cut the program in six? Surely it does not make it faster. Rather, it makes it understandable and verifiable. Intermediate data, formerly invisible, became intermediate files, open to inspection. Moreover, we can test each step separately by modifying its input file manually and observing the effect on its output file.