My argument against requiring a second ride in this scenario is that it is extra work that will probably never serve any purpose. In order for the second measurement to matter, three very unlikely events have to happen for the same expired course:
1) The course has changed since the original measurement without anyone realizing it. (Unlikely, but it happens. That's why we expire courses. Let's say it happens in 1 of 4 expired courses that are being remeasured.)
2) The measurer rides the course inaccurately. (Let's say this happens in 1 of 10 measurements.)
3) Assuming #1 and #2 both happen (a 1 in 40 chance) then the inaccuracy of the measurement has to exactly match (nearly) the amount the course has changed distance. Let's say something has changed about a 5k course so that it is actually 4950 meters long now. The bad measurement would have to end up being between 46 and 54 meters long for him to come up with a measurement of 4996-5004 meters. If his inaccuracy is not within that range, he will have to measure the course a second time anyway. What are the chances that his inaccuracy will match, to within that small range, the amount the course has actually changed? 1 in 10? 1 in 20?
So the chance that all three of those things happening, which would result in an inaccurate course if we don't require a second ride, is 1 in 400 cases, at worst. How many re-measurements of expired courses do we have in a year?