Interesting discussion - thanks for the informative discussion. There was some nice work done by Kirkland, Caves, Humphreys and Pitfield in 2004 which allowed a relative probability of overrun events to be assessed. Basically the probability of an overrun can be found in terms of how much distance is left on the runway after the required accelerate-stop or landing distance.
If there are only a few hundred metres (yards) extra, then the relative probability of an overrun rises. If the runway is long and well beyond that needed for takeoff/landing, then the relative probability of an overrun drops away sharply. Nothing new there, but it is nice to be able to quantify the risk and relate that to the need for specific engineering action.
The graphic comes from a paper looking at the need for a runway surface friction layer using this methodology at
Risk_friction_layer_AMS_2009
Cheers,
Overrun