Tim Blair criticises the idea that small reductions in speed can greatly reduce the probability of a car accident. I think on this occasion he is incorrect.
I forget much of my high school physics but isn’t it true that impact is proportional to mass and speed squared? Thus increasing speed from 60 to 65 km per hour - by about 8% - increases the collision impact by 17%. This is quite apart from difficulties of stopping a faster car, problems with lower reaction times and so on. Taking all these into account it has been found that increasing speed as suggested does in fact double the accident probability.
The impact of speed on accidents is discussed here. Their suggestion is that reducing a car’s speed from 60 to 50 km per hour reduces the probability of killing a hit pedestrian by half although you are still almost certain to injure them. A more ccomplete bibliography is here.
Of course reducing car speeds to zero would reduce traffic accidents to zero. The question then is what are appropriate speeds from the perspective of trading off reduced travel times against reduced accident risks. There is quite a literature on this (see here for example) . My understanding is that current speed limits on urban residential streets get it about right.