Some response after all
Slynky, your adendum to Geo's formula is not bad, but it still gives a bit strange results. What my little table do is to mix the logaritmic of Geos formula with some linear components. By changing some simple factors I can get it to give (almost) the results I want. The computation of the actual game results could be done just the same way you propose using your formula (with a little fix of the table), calculating the knocked out players points first and dividing the result among the remaining players.
Just som examples (remember the table can be twisted to give both higer values and more/less spread):
3 Players
Slynky/Geos formula: 2,60 / -0,87 / -1,73
Primitives table: 1,51 / -0,08 / -1,44
4 players:
Slynky/Geos formula: 3,67 / -0,33 / -1,33 / -2
Primitives table: 2,04 / 0,29 / -0,74 / -1,59
10 players:
Slynky/Geos formula: 8,95 / 2,62 / 1,04 / -0,01 / -0,80 / -1,44 / -1,96 / -2,42 / -2,81 / -3,16
Primitives table: 3,73 / 1,58 / 0,71 / 0,17 / -0,25 / -0,60 / -0,91 / -1,20 / -1,48 / - 1,75
Anyway, Its your call Slynky. If you want me to make a table I will do so. If you will go with the formula I'm okay with that too