When they were common here in Phoenix, most people would just throw the ticket directly in the garbage. Law enforcement knew they were pretty unenforceable but if they got a small percentage of suckers I guess it was worth it.
Rear end crashes are much more survivable than t-bones though, and this study doesn't consider pedestrian injuries. To me it sounds like the cameras should also be fining people who are stopping too quickly, as it implies they were going at an unsafe speed for the condition.
Anyway, relatively few people run red lights. Speed cameras would likely have a much higher safety return.
Unless you move them around drivers will just slow down at the camera and then commence speeding. What works is strategically placed cameras that measure average speed over a distance. We have these a few places in Norway and most people stick to the speed limit or a bit below between the cameras.
Reading your first sentence, I immediately thought "this is easy to deal with using checkpointed cameras." And then I read the next part. Curious how well that works out, all told.
it would also be relatively straightforward to require cars to self-report speeding (keep a speed/position log, have it be read out at annual inspection). There are some issues around tunnels and poor GPS geolocation in urban areas, but it would work great for things like highways...
I had that exact thought as I was typing my post! :D
That said, I can see many reasons that is not liked. Amusingly, as you add more and more detection systems to cars for stuff like this, you are backing into autonomous vehicles.
Detecting out of spec driving is a much simpler (and cheaper to solve) problem than autonomous driving, and may have higher safety benefits? Yes there are privacy concerns but the same concerns exist with autonomous driving.
Motorists hate them because they're a money grab from local municipalities.