And they don't like to discuss them because they might disclose some proprietary advantage to a corporate spy.
One of them SEEMS to be that the Nuvis have a prejudice for right turns over left turns. I have seen it create a route any number of times (to different places) where the Nuvi sends me "too far" and makes right turns to reach the destination, when the short and direct route would be just a little shorter--but require left turns, crossing traffic.
If you contact Garmin support and upload the specific route to them, they MAY be willing and able to tell you which "feature" is creating those puzzling routes.
I'm not sure how many other unpublished "rules" exist in the logic of the Nuvi, but they definitely are doing things based on factors that they are not disclosing to the users.
Sometimes I'll compare routing from AAA, Google Maps, DeLorme, and the Nuvi, and all four rarely agree on the same route. Sometimes, some of them even show the real best route.<G>
There's more to the dual route issue than just what other posters have noted. Think about what these units do -- they constantly communicate with dozens of earth-orbiting satelites, talk to you (some let you talk to them), know what time it is, know what the speed limit is on most roads, and with FM traffic can even incorporate that constantly changing data into route planning. They are pretty advanced gadgets now-a-days and have a lot of programming jammed into them.
That said, there's no doubt that time of day and assumptions made about the correlating traffic patterns on a given type of road have an impact on route calculation, regardless of whether an FM traffic receiver is inolved, although that certainly adds more complexity.
I think there are calculations based on whether you are turning into oncoming traffic and the resulting delay that causes. As a result the route there could be different than the route back as the machine will attempt to eliminate as many left hand turns as possible thus speeding your travel time.
On a recent trip to Moncks Corner, SC my 765T took me a shortcut via SC 178 but on the return trip it took me back through Columbia, SC. It was a confusing mess.
A point to keep in mind is that if the GPS unit considers two routes to be similar (according the road type details in its database and the your route preference selections) it will select the shortest route from where you are when you initiate a route calculation. A difference of only a few metres is significant. In many cases a good minor road will evaluate the same as a goat track... in other words GIGO - garbage in = garbage out. The routing functions are only as good as the info they work with. In my experience they have improved considerably over the last seven years.
It also pays to keep in mind, that we ALL know that these GPS's are NOT foolproof, and it pays to have maps available so that you have a basic knowledge where you wish to get to, and HOW to get there.
Remember, these are not the same quality that the defence forces use and the mapping and programming is far inferior to the costly GPS systems used by those organizations.
You get what you pay for..