My recollection of the USATF/RRTC thinking on this is this:
Using the USATF 2000 Competition Rules, the latest edition I have, our rule 185 (3) reads:
“3. Road running performances will not be accepted if the remeasurement shows that the actual course distance was shorter than the stated distance”
Now for semantics: We took the word “shows” to mean “shows convincingly.”
If a 10 km course remeasures to exactly 10,000 meters we know that it is about right, but cannot say that it is either long or short. There is uncertainty in the measurement. If the course remeasures 9999.9 meters, we are still within an area of uncertainty. To deal with this we established an “allowance for error in the validation measurement” of 5 meters in 10 km. In other words, if the course remeasures to less than 9995 meters we feel that shortness has been convincingly established.
Without this allowance we could be faced with a marathon remeasurement of 42194 meters, and have to disallow the record. This would be seen as ridiculous by anyone knowledgeable in measurement theory. Thus the allowance.
It is true that a 10k that remeasures to 9995 meters was probably not well-measured on layout, but our goal in validation is to establish, as best we can, the length of the course as it exists, without regard to how it got that way.
