A roundabout way to address traffic problems

I realize that the roundabout is the latest fad in traffic control, and that it presents a low maintenance way to avoid accidents at intersections. Unfortunately, in practice they are being used in places and ways which don't seem appropriate, if benefit to to motoring public, the taxpayers, the voters, are taken into account.

In places where the traffic is light, is roughly the same from all incoming roads, a roundabout is just a really annoying way to cause traffic delays as people slow down to negotiate the turn, coming to a full stop or crawling pace when there is snow or ice on the road. In bad weather it presents the winter motorist a fine chance to slide off the road negotiating a turn which has been introduced into an otherwise safe straight road.

At heavy traffic times such as rush hour, as volume increases people coming from one road and going to another half or three quarters of the way around the circle will occasionally dominate the circle and enter the roundabout in a steady stream of traffic with no gaps for access by motorists entering from other roads. If this persists briefly it's intended behavior, but if it lasts for minutes it results in a backup on those other roads, sometimes all the way back to another roundabout, which in turn becomes totally deadlocked.

Eventually the driver at the front of one of the backlogged roads gets frustrated, perhaps urged on by the people behind leaning on their horns, and that front driver tries to pull into a small or nonresistant gap, which at best means a driver in the roundabout, who has the right of way, will have to brake and give every car behind a chance to have a rear end accident. At worst, the non-gap will be occupied by several cars proving the law of impenetrability, that two objects can not occupy the same space at the same time. This doesn't seem to bother traffic engineers, one was quoted in local news coverage saying that "accidents are usually harmless fender benders with no injuries." If accidents actually at the roundabout are not higher, I have to wonder how many drivers frustrated by a long wait try to "make up the time" and drive aggressively for the remainder of their trip while doing so.

In most fields of engineering there is a technique called "worst case analysis," which would predict behavior in the worst possible case and the worst case which occurs regularly, such as twice a day in morning and evening rush hour. That might lead to some better choices of where to use a roundabout, spacing between roundabouts, and where a signal is a better choice. Since removing under-performing roundabouts would be an expensive solution to an occasional but frequent problem, perhaps augmenting the roundabouts in problem locations with a traffic signal, controlled by backlog sensors and active only at problem times, to prevent prolonged exclusive flow from any one source.

In a perfect world motorists would note a backlog problem and take action on their own to allow smooth traffic flow. In the real world, particularly at the end of the work day when people are tired and want to get home, motorists don't care if someone waits a long time to proceed, as long as it is someone else.  Proponents claim that roundabouts work fine if drivers only knew how to use them. This is somewhat like a software package with a poor user interface and a vendor suggesting training the users instead of fixing the problem. Users will turn to other software, and I know from discussions I have heard, drivers state that they are already taking alternate routes to avoid roundabouts. Shifting the traffic from main roads to neighborhoods and secondary roads is not the solution, it's time to rethink and improve the "user interface" of problem intersections