Domain vs Path Links is the probability that a link on a synthetic page goes to another domain rather than to a deeper path on the same one. The choices run from 0 to 1.0.

At 0 every page sits under one host, http://synth.web/, and pages differ only by their path, so the whole crawl is one site. At 1.0 every page is its own host, such as http://1-2.web/, so the crawl is as many sites as pages. In between, each link is decided by the probability, the same way every time for the same crawl.

This setting is what makes a synthetic crawl exercise the parts of a crawl that depend on there being many hosts: robots files, host address lookups, and the crawl's division of its attention among sites.