MP at Level

Neither can be taken as straight code, the idea behind the first is like that:
4 int, random number between 0 and 2:
4 + 0.4 + 1.7 = 6.1 -> 6
4 + 0.4 + 0.2 = 4.6 -> 4
etc.
Giving it a difference in probability for the 3 numbers:
0.6/2 = 4
1/2 = 5
0.4/2 = 6

The way I wrote it, it would just be simply 1/3 chance of each MP value, and there would be no difference between 21 and 29 int (since both round to 2), whereas with that other method, it would be
21 int:
0.9/2 (0.45) = 6
1/2 (0.5) = 7
0.1/2 (0.05) = 8
29 int:
0.1/2 (0.05) = 6
1/2 (0.5) = 7
0.9/2 (0.45) = 8

I suspect this method is right but I'm not certain. I brought up the other one because my Paladin gained 8 MP last level, and I only have 20 int (should be impossible with that method)

i hadnt really considered your suggestion. the only basis i have is people claiming to gain 4, 5, or 6 mp with different probabilities at low values of int. what i quoted works for low-ish int... more or less. however,

Hi guys.

4 + floor(int/10) + min(floor(rand(0, 3) + int % 10 / 10), 2)

tc;du you'll gain an avg mp of 5 + int/10 per level (no floor peaches)

i like this one much better because it looks like it accounts for the fact that a friend of mine went from 864 - 937 mp in 1 level with 100+ int. i was wondering about that inconsistency with the one i usually quote. it would account for gaining 8 mp at 20 int as well.
*shrug* better than 2-4.