LAST EDITED ON 08-20-11 AT 02:43 PM (EST)Good analysis and you raised some points that I had thought about when I posed the question, especially the 'human' side to the game in the way players react.
While the math is sound based on certain assumptions, these assumptions, as you clearly showed, are anything but certain. A simple example is whether anyone would 'throw' the PoV (purely human factor) and also who has the best chances of winning it (this could statistically be based on past performances, but there's a too small sample to be reliable).
Let's take the most extreme example and say Dani is so good she has a 100% chance of winning PoV (which obviates any 'throws' too). Let's say players A & B are her allies and player C is a floater.
If initially nominated with player A, she removes herself and then player B gets the replacement nom. She is safe, but will lose one of her allies A & B. Good for HoH, bad for Dani. Dani loses two supporting votes.
If players A & B originally nommed, then she can get her choice of which A or B is saved, and then the HoH has to resort to nomming Player C. Not good for HoH, good for Dani. Dani only loses one supporting vote and has the supporting vote of A or B who she saved.
By this logic, then, it is clear that HoH should have initially nominated Dani, with the assumption that Dani has a very high percentage chance of winning PoV.