I had the toughest time figuring out the math for this, but it appears that the expected value of a poison requiring two consecutive saves to cure, with an infinite duration, is:
1/p^2 effects experienced
Where p is the probability of making a successful save against the poison.
The variable p can be spelled out more explicitly:
p = (21 - poison DC + save mod)/20
Which translates to:
expected value = 400 / (21 - poison DC + save mod)^2
Of course, p needs to be confined to the range of 1/20 to 19/20, and for very low-p situations the duration of the poison is likely to significantly reduce the actual expected value. Exactly how is something I'm still considering.