I think the ring road is the most accurate; it shows the density fluctuations that appear and is a good match to actual traffic. The other situations have left out some details — I’ve noticed that when a lane closure is upcoming, where people will change lanes varies greatly, even when traffic in both lanes has slowed to a crawl. When that happens, you’re supposed to go all the way to the merge point, and then merge, but many people will force their way into the other lane well before, and some people will simply not let a car into the gap in front of them. In the uphill grade I didn’t see much of the behavior I observe on I-81 in northern Pennsylvania: a truck going 55 mph passing another truck going 54 mph, and screwing a long procession of cars behind them, all of whom want to (and can) do at least 65.
Still, it’s a simulation, and I see there’s a “politeness factor” slider, which I presume controls letting people in rather than someone in the simulation flipping a virtual bird or honking a horn.