[btrettel]

I haven't read this page too closely, but the better strategy is similar to an approach I've seen called the "ranking theorem" mentioned here: http://www.prioritysystem.com/math.html

> One result is the ranking theorem: If independent projects are ranked based on the ratio of benefit-to-cost, and selected from the top down until the budget is exhausted, the resulting project portfolio will create the greatest possible value (ignoring the error introduced if the portfolio doesn't consume the entire budget).

This is speaking more generally of a "budget" which could be time, money, or something else. It's an approximation because when the resource is nearly spent it becomes a more complicated optimization problem, selecting projects which fit, which doesn't necessarily pick projects with the highest rate.

Obviously this becomes more complicated if the projects are not independent, e.g., one is a prerequisite of others.

[roboy]

Ranking Theorem requires two assumptions, which are both violated in this context: 1) The projects are independent 2) Not all project need completion to achieve success (i.e. you can select a subset of projects)

Here, the failure of one part leads to the failure of the whole project, and all parts of a project need completion at some point in time. The task at hand is thus to select the order of parts of projects to maximize the overall gain over many projects.