“Does the explanation explain the observed events and does it explain all of them?; he always writes out what the solution is expected to make happen—make automobile accidents disappear, for instance—and then tests regularly to see if this really happens; and finally, he goes back and thinks the problem through again when he sees something atypical, when he finds phenomena his explanation does not really explain, or when the course of events deviates, even in details, from his expectations.
