It is confusing to the general public and, for the most part, arbitrary.
Courses could be measured to the actual distance, and validated if the course proves to be at least 99.9% of the stated distance.
An American record from an old course (current certification process) could only be broken on a new course if the time run is at least 0.1% faster. It's unlikely this would ever be an issue since American road records are almost always broken by more than that.
It's been quiet on the board lately, so I thought I'd rattle a few cages.