In a CrossTalk article Risk Management for Dummies, Tom DeMarco speaks about late software projects and the approaches for the solution. The notion of early start as the solution for late finish needs to address the core question, but fails to address the root cause of lateness - no schedule margin.
How soon should we have started to finish on or before the needed finish time?
The answer to this is simple:
- If we have a schedule for the project, which contains the needed work efforts, in the planned sequence with an estimated duration for that effort - the Most Likely duration - then we can model the probability of a completion date with Monte Carlo Simulation.
- The question is what is the range of possible durations an activity can have, before we actually start performing that work?
Good question. And of course like all questions about some future activity we'll need to Estimate those values. We can't actually manage - in any credible manner - without estimating. Anyone suggesting otherwise has likely only encountered trivial projects where the Value at Risk was low enough that no one cared is you were late or over budget.
An Example of Managing In The Presence of Uncertainty
Columbus proposed a voyage to Ferdinand and Isabella in 1486. The monarchs promptly set up a special commission of learned men to study the proposal, which they found vague and arcane. After four years of unsatisfactory discussions with the master mariner, the commission's finally rejected the proposal.
To their credit they did not abandon Columbus and soon recalled Columbus not to plan again, but to name his price for carrying out the voyage. The price was enormous. Columbus insisted on being knighted, appointed a grand admiral and viceroy, and given the unwillingness to compromise, the king and queen dismissed Columbus. Eventually, they relented and accepted all of Columbus's demand.
I (Norm Augustine) put forth this example because all parties involved could have benefited from risk management. Columbus could have presented his proposal containing each element of risk, but consider the payoff, well worth the investment.
Ferdinand and Isabella could have assessed the "price of doing nothing." The commission of experts could have have a framework in which they could evaluate the technical feasibility of the proposal. If the parties had studied the technical risk management, perhaps we'd be celebrating 1486, not 1492
For an simple schedule, like the Wright Brothers meeting their contractual data to the U.S. Army of August 31 1908 for the delivery of the Wright Flyer, they needed a confidence level of 80% for the on or before date. The schedule margin of 10 days provided the protection for that date.
What Does This Mean?
The first meaning, missing from the CrossTalk article is.
Schedules without margin are late on day one
It means Estimating is Risk Management. No estimates, no risk management. No risk management, lower probability of project success. For schedule risk we need irreducible and reducible uncertainties that drive the schedule. The for the reducible uncertainties, we need activities to reduce the risk from that uncertainty. For irreducible uncertainties, we need margin to protect each major deliverable.
Risk Management is How Adults Manage Projects - Tim Lister
Remember that next time you hear we can't estimate, we don't want to estimate, estimates are a waste.
Here's How to Manage in the Presence of Uncertainty