Some people seem to need to have every risk quantified to the point of identifying every potential impactor and cataloguing it, before they will accept that the risk can be known. others can accept a statistical description of the risk.

For instance, I ride a motorcycle and it would clearly be of great value for me to know the location of every pothole in the road and every drunk driver (and every one of a dozen other potential game-enders for my motorcyling life). If I knew this, I could ride with less worry.

However, the task of compiling that list of all potential risks would be impossible, and would prevent me ever riding. Instead, i take a statistical approach to the expectation of meeting these dangers, and have strategies in place that I have rehearsed for "what to do if...."

In a similar way, most impact studies are based on average expectations, rather than absolute descriptions of all bodies. Sure, we don't know EXACTLY what's out there, but we have a very good idea how likely it is that an impact of a certain size will occur in a certain timeframe, and we make plans accordingly...