Another method
Interesting read. For years, basically ever since I could get a computerized estimated total of miles and elevation, I have used the Climbing Index method. Basically, just take the total elevation gained and divide by the total miles traveled. The higher the resulting number, the harder the ride. The flaw is that in rides like your 17 mile ride with 400' of climbing, it does not show the route as flat but with one monster hill. However, generally speaking, it does a pretty good job of giving the overall ride difficulty.
For me, your 17.7 mile ride with 397' of climbing (Climbing Index of 22.4) would be a moderately flat ride (fairly easy). Sure, part of the ride is very difficult but the vast majority of the rest is flat so it would help make up for it. To give a comparison, say you were on a rail trail that climbed at a steady 2% for 1 mile. That 1 mile climb would have a Climbing Index of ~106 and would be considered Beyond Category. Much harder than your 22.4 Climbing Index ride overall. A two percent climb does not sound like much but if you lengthened that 1 mile trail to 10 miles without descending, you would now be climbing 1056 feet for the same Climbing Index of 106. However, if you then turnaround and go back down the 10 miles to where you started from, the Climbing Index drops to ~53, which is still Mostly Hilly. You can start to see it how it is both the distance and the total elevation gained that mostly effects the difficulty. If the Climbing Index were zero or less (only going downhill), the ride would be super easy. The Climbing Index gives an overall average difficulty, not a spot specific difficulty.
My point if you keep track of the Climbing Index for routes you have ridden and assign your own personal "difficulty rating index", you will very easily be able to tell how difficult OVERALL a future mapped route will be by calculating the Climbing Index. You can map your ride on RWGPS and it will give you the total miles and total climbing so you can calculate the Climbing Index easily.
Tailwinds, John