Quote from the Blane Bachelor article on course measurement in Runner's World, posted here by Justin Kuo:
"There’s the Short Course Prevention Factor to figure in, too. According to USATF and the International Association of Athletics Federations, all races must be measured one-tenth of one percent of the total distance long, which ensures that the course cannot possibly be found to come up short in case of a record."
Well, leaving out Blane's statement "...cannot possibly be found to come up short in case of a record.", which is not true, I have understood that the purpose of the SCPF is to compensate for known and unknown reasons that any particular calibrated JOC measurement may vary from a (theoretical) equivalent steel tape measurement. "Known" being greater wobble during measurement than during calibration, "unknown" being a difference between tangents measured and tangents run in the corresponding race.
The article would seem to me to inform the layperson that we add in this random extra distance to account for the "unknowns", so that we have a small "fudge factor" in the event of a record and subsequent validation. I am sure that there must be a lot of material on this subject in our archives. I have not read it. But I was taught that measurement rides are often or typically subject to more wobble than calibration rides. I certainly agree with this. My cal rides are typically on quiet streets with no stopping and no guessing tangents. There is little or no distracting traffic on these cal rides, so wobble is at a minimum. Not so for me during certification rides, especially when there is traffic around. I am sure I am wobbling slightly more during measurement, and I know that wobble can add up and can contribute to any given measurement being short of the stated distance.
Maybe I am misinformed. I haven't read the rulebook that Dave Katz referred to about this.