The question was posed, how much mode share would non motorised transport need before we see an overall reduction in the road toll? With non motorised transport currently more dangerous per distance travelled than motorised transport any small shifts away from cars to bicycles or walking would result in an increase in the road toll, how can it be made safer?
Non motorised transport is inherently safe, the ability for a pedestrian or cyclist to kill one another without intent and even in the event of accidents is at least an order (likely 2-3 orders) of magnitude lower probability compared to a car, or worse a truck, bus, or other heavy vehicle. People are accustomed to bumping into one another while walking without a second thought for the risk of injury or death, but as vehicles get faster and heavier both pedestrian risk and fear grows. The theoretical extreme of 100% pedestrian transport is observed routinely inside shopping centres where few people would be concerned for their safety and no special traffic controls are needed, logically as once the dangerous forms of transport are removed from the environment the remaining risk of injury or death is exceptionally low even in the chaotic movement of uncontrolled traffic. Its only once the reduction of motor traffic then crates less risk for pedestrians and cyclists that the road toll could fall by shifting to these modes, and we now have an estimate.
From the previously discussed modelling method the rates at which different modes interact was estimated at 50%, each mode spends half of their travels interacting with the same vehicles/mode and the other half of the journey interacting with all the modes split equally by their mode share. Unbundling can be done explicitly with exclusive lanes or paths, or happen organically through filtered permeability. Starting at an estimate of 50% for this unbundling we can look at how many car journeys would need to be shifted to bicycles to reduce the road toll along the dark blue line:
For the existing situation it would require an unrealistically large proportion of cars existing 70% mode share to be moved to bicycles, with an overall reduction only occurring once bicycles reach a 36% overall mode share. Considering a case of no segregation (-100%, the light blue line) makes it impossible for bicycles to be a viable mode in the interests of safety. The other extreme of complete segregation has both immediate benefits and eliminates the initially rising trend, it appears that more separation of transport modes is the key to unlocking a reduction in the road toll. Looking at only a more realistic range of mode shares:
The slopes are all very slight once we assume a base level of unbundling, and small improvements in the segregation of all modes (+25% taking the fraction of exclusive sharing up from 50% to 66%) suggest a substantial 10% reduction in the overall road toll. Vulnerable non motorised modes benefit the most from unbundling with the above reductions having the majority of lives saved in the pedestrian and cyclist groups, even before considering changing mode shares. At the current rates non motorised transport modes account for 16% of the deaths on the roads, while representing less than 4% of the mode share by distance travelled, an entirely inequitable share that needs addressing.
On a positive note the sustained reductions in risk for cyclists in Victoria are almost to the point where increasing the mode share would result in reductions to the road toll, once past that point there will be a strong public interest to further increase the cycling mode share for the benefits of all.