rsiano asks a good question. It is by no means a joke.
Here is a
paper by Robert Patlovany which was published in the premier journal in the field of risk analysis, on exactly this question. He comes to the conclusion, as rsiano asked, that random is better than (some of) the current regimentation.
I have looked rather hard at mid-air collisions between aircraft on scheduled commercial operations, before and after TCAS. Keeping in mind the growth of traffic, and that there have been two collisions in cruise between TCAS-equipped aircraft, it is not at all clear from the statistics that TCAS is a win. I tried to discuss this on this forum some three years ago, which was mostly unproductive, so I took it elsewhere (the IEEE Risk Factor blog).
The posters who suggest that some of the conventions lead to convergence are following a line of thought which led Patlovany to his study. It converts into a good argument for free flight. However, free flight requires TCAS, and that is another story, for TCAS still suffers from technical and algorithmic anomalies (although the latest RTCA version is free of the one I was most concerned about).
PBL
PBL