Then of course there is the self-evident fact that there is no significant collision problem in UK VFR airspace - at least, the statistics don't indicate one, but one suspects that the majority of VFR pilots are at least subliminally aware that they keep a grossly inadequate lookout and this is a means of compensating.
I'm not quite sure what you're advocating here.
The evidence is that pilots simply aren't physiologically capable of maintaining a good lookout, including professional pilots. e.g.
www.atsb.gov.au/media/32918/limit_see_avoid.pdf. The current low rate of collisions probably has more to do with the low chances of a collision occurring, and there's only limited scope for further reducing accident rates by improving pilot lookout.
That isn't to say that pilots shouldn't be looking out of the window, or even work on improving their skills, but I put more faith in strategies such as random cruising altitudes or for that matter the quadrantal/semicircular rules.
What it results in is that no one - least of all those who do fly responsibly, have any idea what direction a conflict is likely to appear from which actually makes the situation less safe, not more. But as chaos is the inevitable result of anarchy why would anyone be surprised?
The AIP only suggests (and does not mandate) that quadrantal rules be adhered to. And this only above the transition altitude. In practice this means that in much of the busiest airspace (London TMA, Manchester and Luton/Stanstead corridors) you're not allowed to fly at an altitude that would enable you follow the quadrantal rules.
So how to cope with this government-mandated 'anarchy'? If random cruising levels are one way to reduce the risk of collisions, then it's perfectly legitimate to discuss them.
It's also quite reasonable to discuss whether random cruising levels may be safer than quadrantal cruising levels. One other poster has mentioned work showing that they may actually be less safe. I remained on the fence. But matters are serious if we live in a country where we may not even question the justification of rules (or in this case, a recommendation).