Step is a variation of what Alan Turing called a choice machine.¹  When we run a Step program, the machine makes a series of calls, and for each call it may have a choice of several different methods.  For example, here's a slight variation on our Sentence generator:

# Try: [Sentence]
[randomly]
Sentence: The cat slept
Sentence: the cat [Verb] the dog
Sentence: the dog [Verb] the cat

[randomly]
Verb: ate 
Verb: chased 

When running [Sentence], the system must choose:

• Which method of Sentence to use, and
• Which method of Verb to use, if our chosen Sentence method calls Verb

We can diagram the choices and the outputs they produce.  The rounded boxes are calls, the green square boxes are the final outputs.  And the arrows are labeled with the different methods that can be chosen for the call:

graph TB
   s("<b>[Sentence]</b>") -- "Sentence: The cat slept" -->  m1[The cat slept]
   s -- "Sentence: the cat [Verb] the dog" --> m2("the cat <b>[Verb]</b> the dog")
   s -- "Sentence: the dog [Verb] the cat" --> m3("the dog <b>[Verb]</b> the dog")

   m2 -- "Verb: ate" --> v11[The cat ate the dog]
   m2 -- "Verb: chased" --> v12[The cat chased the dog]

   m3 -- "Verb: ate"--> v21[The dog ate the cat]
   m3 -- "Verb: chased"--> v22[The dog chased the cat]

   style m1 fill:green
   style v11 fill:green
   style v12 fill:green
   style v21 fill:green
   style v22 fill:green

We'll call this a choice tree.  Different sequences of choices – different choice paths – produce different outputs, shown here in green.  We'll use green in these diagrams to indicate the ends of successful choice paths, and red to indicate failed ones.  For the moment, all paths are green because there's nothing that can fail.

Notes