Some of these issues can't be solved easily. For example, we have many "crossover" on and off ramps in Mass. (I'm not sure of the situation in other states and provinces, although I don't see that many outside of this area.) These are interchanges where exiting traffic must cross over into the same lane in which other cars are attempting to merge onto the highway. So you have cars going 65 mph trying to slow down and slip past cars speeding up from 20-30 mph (depending on the curve of the ramp). The worse of these is the intersection of I-93 and I-95 north of Boston. Nothing like playing "bumper cars" at high speeds. Eliminating these intersections is very expensive and difficult, especially given the need, in many cases, to expropriate land near the intersection to make the changes. Rotaries are another great example. (By the way, rotaries work great with the right traffic volumes and people who know how to drive!)
But there are also many extremely inexpensive things that could be done to help reduce conflicts. Poorly marked or outright non-existent lane markings come to mind. Drive along Mass. Ave. in Arlington, for example, and you'll see almost no indications that there are two lanes in either direction, and yet there is obviously plenty of room for two cars side-by-side. Most people drive as if there are two lanes but in the absence of lines, some people aren't aware of this and drive in the middle, forming a kind of NASCAR mosh pit behind them.
There are many intersections where the lane markings on one side of the intersection don't line up with those on the other side. Given inertia, a natural tendency to drive straight (a generally desirable quality at an intersection!), plus the average person's general lack of precise control over their vehicle, and you end up with cars straddling lane lines on the other side, lurching suddenly one way or another as the drivers realize they're not exactly in the lane. How difficult would it be to line these up properly? I'm no traffic engineer but this doesn't seem like rocket science.