That does add some better potential to developing a protocol. But, it also introduces a statistical possibility: you may want to consider that one reason your success shrinks as the number of uncontrolled, random people shrinks, is that the more people are around, the higher the possibility that someone will randomly look toward you, or that your fast head movement will catch their eye, or that you will have more chance to ignore times when you look around and there is no one looking. The fact that you have noticed that your success shrinks as this crowd, and hence these conditions shrink, seems to suggest you have a feeling that you need all this randomness to have a hope of success, and when that randomness is removed, and the conditions become controlled (no matter what those controls wind up being), your success drops.
If this is true, then you should notice that you will resist any controls people suggest that weed out this randomness. This can be self deception.
But if this is not true, than coming up with a controllable protocol, based on your statement above, should now proceed rapidly.
!.
