I don't have a "brand new" Garmin, but I started using GPS technology back in the 90's using a hand-held unit meant for hiking. It was great for mile marker spotting. Once the civil accuracy was improved (sometime in the 90's) finding a specific spot with them was very good.
That said, I've found them very environmentally sensitive. I think this is why some people will say a model is spot on, and another will not. I think this accounts for the "it reads for me all the time so you have to be wrong" scenario also. As I've mentioned before in the forum I used to get water as a country club near here, and the GPS track would have me zooming down the road in the opposite direction and then recovering a 1/4 mile later. Other have reported the same affect there, so there's a funky reflection going on. Going around the track near my house in the summer results in tracks that resemble a drunken sailor.
While some models may be more sensitive than others, any study should look more at the conditions where unreliable readings are likely, not just the models that aren't affected as much.
People will always say "well the military can track you within a 1/2 inch", but that's not a wrist watch, and they're not measuring distance traveled. It's an entirely different matter to measure how to get to spot x with a certain percision than it is to measure the path along a series of those spots.
What I tell people when they challenge a distance based on GPS (when I'm in a good mood) is that if the course was laid out per the map, and if the procedures for measurement were done right, the process yields an extremely accurate course and knowing all the things that can affect the measured distance on a GPS there are simply way more reasons to suspect a GPS measurement than a bicycle one.
I've done numerous head to head comparisons of Jones Counter vs Garmin 610 because I always have it on the bike when I measure. I use the tracks to speed the generation of the map. Most of the time it is very close, but it's wrong often enough to not trust the accuracy as much as many do. In addition to tall buildings, trees and cliffs are problematic, but I think there are other factors I haven't identified since the incidents by the local country club seem more than a tree could cause.