First, task impact per unit of effort is technically a better measure of how worthwhile it is to complete the task than absolute task impact is.
Second, I recall some writing in the personal-finance genre which said that people often make more progress getting out of debt by paying their debts in order from small to large, ignoring the interest rate. This is more expensive than paying debts in order of high to low interest rate, but it may give you more of a feeling of progress when one of your debts disappears, even though that debt wasn't, in reality, causing much of a problem. By the time you get to more difficult debts, you're on a roll.
This suggests that there is probably a place for configuring the priority scheme; absolute impact, impact per unit effort, and absolute effort are all reasonable choices.