Task Dependencies and Probability
Flip a coin, heads or tails. The probability that it will come up heads is 50%, after all there are only two sides. Flip that same coin again, the same probability, 50%. However, if we say that success is two successive heads (or tails) that is different. The probability of two consecutive heads, is the product of the two coin tosses (50% x 50%) or 25%.
I’m sure we understand task dependencies. This is fundamental to project management and schedule. Dependencies are the connections between tasks for example, I must put socks on before I put my shoes on (assuming I wear socks, which I generally do). Of course, this is not the only type of dependency (finish-start). There are four actually;
Let us focus on finish-start for a bit. Specifically, lets talk about a portion of the resulting schedule and the task consequences on the entire project.
Each task has been assessed to have a 90% probability of being successfully completed. Both tasks are to be completed to be successful, and success is defined as accomplishing the attributes of the task, that is objective, cost, and schedule.
The probability of successful completion of the tasks at 90% may sound pretty good, that is until you use some math. Let’s say these tasks are also unique, that is one task has no bearing on the probability of the other’s success, either depending or predecessor. So the probability of these two tasks actually successfully being completed is 81%.
90% x 90% = 81%
A probability of 81% success may not sound horrible until you realize there are very few projects that consist of two depending tasks, or that have a task probability of success being 90%.
If your project plan does not have any slack, you are on what my friend Kim Pries refers to as a death march. Planning long term projects, that is project aspects that are way in the future, will especially suffer from this malady. This is one of the benefits of agile approach where planning has a clear short-term focus, which is contrary to the frequent approach to conventional projects.
We can learn much about the overall project by looking at the individual tasks. It is certain that sometimes identifying the probability of the individual tasks potential success is more of a subjective exercise. However, there are also times when subject matter experts, those that do the work, or perhaps some available process measurements add some credibility to the assessment.