The answers are entirely correct.
The Garmin algorithms rely on the known data for each road on the map. Garmin rely on third parties to supply that data, for obvious reasons. For instance, if Garmin’s map data tells the algorithm that a one mile stretch of road A to B is a 30 mph limit it will route the rider along it in preference to an alternative road of say 1.5 miles A to B that is also a 30 mph limit. If the algorithm does not know that the 1 mile road is a cart track, where the rider would be lucky to make 10 miles an hour is beyond the algorithm’s control.
You can see the same effect in towns and villages where the algorithm will create a diversion to cut a corner as it is 100 yards shorter (and therefore ‘quicker’) when the sensible option is to ignore the instruction to cut the corner (down a side street) and ride to the next crossroads instead. Sometimes you just have to be cleverer than the dumb machine on the day..... and check the route before you set off!