On the first ballot, my mistakes were I think relatively common. I overestimated the strength of the top four and underestimated the strength of the bottom four.
In my prediction, I did state that I expected Martha to do better than I had projected but the others caught me off guard. It also seems I drank too much koolaid as I disproportionately overestimated Kennedy's support.
Here is where my predictions for each on the first ballot differed from reality:
Hall Findlay +2.1%
I grossly underestimated the strength of Dryden, Brison and MHF (though I allowed for that) while I overestimated the strength of Ignatieff and Kennedy. Rae, Dion and Volpe were all roughly on the mark.
The rest of my predictions are essentially garbage. It is like when you are doing a complicated math problem, if you screw up the first step, the rest of the math may be correct but you still have a fundamentally wrong answer. This is the case here, if Kennedy had placed third, I suspect my prediction may have been fairly accurate, however he didn't and thus the rest of the thinking is flawed.
However, I was relatively close in my thoughts on how an Ignatieff vs. Dion ballot would break down. In Part II, of my leadership series, I said this is how an Ignatieff vs. Dion ballot would come to pass: (bold denotes notes added now)
Ignatieff vs. Dion - Too close, slight edge to Dion
Under this scenario, Ignatieff will not bleed very much support (correct) while Dion will have placed third in the final four and picked up a substantial amount of Kennedy's support (correct). However, a good number of Kennedy's delegates west of the Ontario/Manitoba border will have been uncomfortable with another Francophone from Quebec as leader and move to Ignatieff and Rae (incorrect). However, enough will move to Dion to allow him to place second in the final three.
On the final ballot, Kennedy's remaining supporters will spilt around 55-45 to Dion over Ignatieff however a strong number of Rae's supporters going to Dion should push him over the edge possibly winning in a very tight contest over Ignatieff by about 52 to 48 (roughly correct - it was 54-46), however it could go either way.
Thus, as usual, I've missed the mark but am not totally out to lunch.