It’s also impossible to infer the logical process from a superficial observation of the tests being done - that would be like inferring the code architecture from what’s displayed on an output device, in rare cases it might be possible, but usually not