I hate to disappoint, but those slides are not theory. They are processes mandated by good project management principles – take another look at the Immutable Principles of Project. Violate one of those principles and your project is in jeopardy on day one.

Only in the informal IT world does it seem it is allowed to “swizzle” the numbers while the program is underway, or start work using someone else’s money with a less than credible cost and schedule estimate, or simply “behave badly” when calling the project “managed.”

There is no excuse – other than laziness and ineptitude – for not making every attempt possible to produce a credible estimate for the project, revise those estimates as the project progresses, and hold those making the estimates and those asking for the estimates accountable for their decisions.

It’s these ad hoc approaches to project management that are the source of most of the IT project failures when you read things like the Standish Report or the Capers Jones materials.

This is not science, this is core project management. The actions of projects you seem to be describing violate all the principles of good project management. To suggest otherwise is to ignore both those principles and the consequences of ignoring.

Here in the US that’s called a self inflicted wound.

With a single point estimate – say 48 days to build the gadget. Not knowing the variance on that estimate means you have a 50/50 change of making the date driven by that duration.

Cutting that estimate in half means you have a 0% chance of making it.

The outcome of the modeling processes described in the SlideShare materials, is it determine the confidence level of “completing on or before” a target date. Now this confidence is driven by the “credibilty” of two numbers:

1. The MOST LIKLEY duration – this is the duration that is “most likely” to appear if there are multiple estimates captured, or the actual work is done multiple times (not something that can be done in one-off development, but is possible is “re-use” projects.

2. The upper and lower bounds of the MOST LIKELY estimate. These are (depending on the approach) the 0% and 100% limits, or better the 10% and 90% limits. In English this means the activities will complete 90% of the time “on or before some date,” or 90% will take this duration or less. Same for the 10%. You can see the 0% and 100% are long tail estimates and not very credible.

Allowing anyone – management or not – to cut any estimate in the absence of the statistical analysis of the impact is suicide – or at best a self inflicted wound.

Projects that do this and project managers and their teams that allows this to happen in the absence of equivalent changes in scope and/or staffing need to look for work elsewhere.

Just an opinion of a grey beard Program Manager.