I'm curious as one who has never stimped a green before, much less derived a rating for a course.
But if one is looking for a valid assessment, shouldn't this consist of multiple measurements on multiple greens with at least some of those being 3-4 measurements in the exact same spot as a control?
A handful of data points can be wildly disparate, but if you have 40 or 50, that would seem to be far more statistically accurate. I would think if I were doing it to take all the measurements, throw out the outliers, and average the rest for a "course rating".
That is how it's done. The method is three rolls in one direction, take the average and roll back from there. I just roll back from the average spot as it's quicker, but measuring it in both directions and taking the average is technically more accurate - although the difference is minor, if at all. It should be a straight roll and the balls should all stop within 8" of each other and within 16" of both directions. If using the x2 side on a shorter distance, these tolerances should be halved.
I've played with it over the years on slopes for fun and the readings are actually pretty accurate even if the difference in both directions are way off - ie a 6 up hill and an 18 downhill would average a 12, which would match the proper roll on the flat area. There is a formula out there to more accurately calculate the roll on slopes but the best way is to find a flat spot. The x2 side of the stimpmeter (which wasn't around when that video was made) doesn't need much room as the roll is halved. On a side note, if you use the x2 backwards so the ball has the longer roll down the meter, the greenspeed goes up x1.5.....
Its funny this thread went here as I spend a few weeks this summer dealing with this and working out why the readings didn't match up to the perceived speed (ie they "are not as fast as usual", even though the reading was over a ft more than normal...). The issue is that its not a very accurate device, as it relies on the users input as well as other factors. Generally, a person will use it the same every time, therefore getting a consistent result, but the issue is comparing that to another user or different conditions. Eg someone else may raise it fractionally quicker and get a slightly different reading. There's consistently a half foot difference between my assistant and my readings, on the same spot at the same time...... Even the type of ball can add an extra foot of roll.....
Jeff, I pretty much use the x2 every time as it's more accurate as it can be done on a smaller and easier to find flat spot. I don't know how many others use it but I'd say most people are in the same boat. FWIW I've seen the USGA use it also for tournaments for the same reason.