Always good to have backups, especially for something as much built on smoke and mirrors as GPS technology. I've had a few pretty spectacular failures, like the time it tried to steer me into the river. Evidently there was a ford there when the water was low, but it had to be under at least 4 feet of water at the time!
Less explicable was the time it led me to a cul-de-sac, and kept saying "turn right, turn right" as I followed the turnaround to the left. Of course there were no roads to the right, and never had been. Not so much as a cut in the curb where a driveway was planned. As I was going out the way I'd come in (under it's guidance), it said "recalculating".
Another time, it led me down a country road to an apparently abandoned country club, saying "Continue on highway 20". I even got out and looked, there were only fields behind the building, no signs that there had ever been a road there (nor reason for one to be there). And no highway 20 in that part of the state.
I worked in the GPS field for a lot of years; seen some weird stuff. Units that swore they were tracking satellites that didn't exist; satellites broadcasting the wrong signal (with the flags set saying it was good), and navigation solutions on the wrong side of the world. Of course, when we found that kind of stuff, we always put in patches to try to detect and make sure it didn't happen again; but they didn't always work and likely we didn't see everything that can go wrong.
And our competitors weren't always so careful (or just didn't see the same issues we caught). GPS is kind of a high-tech guessing game anyway "the signal is probably here" and most units err on the side of reporting a position even if they aren't reasonably certain it's the right position. All it takes is poor geometry and one signal bouncing off of something (like the side of a building) to get a position off by a mile or more, even if it is correctly tracking all the signals it can see.
My favorite example was a test we did on a competitor's unit. We laid some model train track through the parking lot, with a long straight section and then a curve. Set the unit on a flatbed, with a model train engine pulling it along the track. Just before the curve, we blocked the signal by clapping a cookie tin over the antenna as it went by. The unit faithfully reported it was still moving in a straight line, even as it sailed around the curve and stopped!
Not so bad for a consumer grade unit perhaps; but this was supposedly survey quality accuracy, good to just a couple of inches.