Practical Software Estimation Measurement

Are Late Software Projects a Victim of 'The Planning Fallacy'?

Software Project Planning FallacyToo many projects are late, over-budget, under-delivered, or a combination.  The problems continue despite widespread awareness and improvements in project management knowledge, tools, and process maturity.  

A recent piece in the Washington Post business section identified a likely culprit: “the planning fallacy”.  Princeton psychologist Daniel Kahneman and Amos Tversky of Stanford describe it as “the tendency to underestimate the time, costs, and risks of future actions and overestimate the benefit of those actions”.  The results are time and cost overruns as well as benefit shortfalls.  The concept is not new: the pair coined the term in the 1970s and has been researching it since.

According to the Post, cognitive biases such as optimism bias (the tendency to expect positive outcomes from one’s actions) and overconfidence can be causes of the planning fallacy. There is a growing body of evidence, collected by researcher Bent Flyvbjerg at Oxford University, that optimism bias is an important bias affecting the quality of forecasts in project planning. 

Other explanations of the fallacy include possible intentional and deliberate considerations on behalf of the planner - such as incentives, organizational pressures and strategic deception. 

The Post piece describes problems New York City's Parks and Recreation Department is having with project management - problems familiar ​to those of us who've been around software projects for a while:  late and over-budget efforts. To improve their estimation, they analyzed data from 1,800 past NYC Parks and Recreation projects and found that the size, complexity, and type of subcontractors can characterize similar projects – a reference class.  Data from this class can be used to plan and estimate future projects similar to those in the same class.

The piece uses the term 'inside estimate' for what we'd call 'bottom up' and 'outside estimate' for a parametric one based on historical data.  The ‘inside’ method for forecasting focuses on the specifics of the project at hand and is prone to optimism bias, leading to the planning fallacy.  ‘Outside’, or parametric methods, by contrast, look at historical data from past similar projects and are far less prone to bias. 

It’s interesting that a group outside the software realm sees problems familiar to those in software development.  Besides the NY Parks and Recreation projects, the research from Kahneman, Tversky, and Flyvbjerg includes examples from a wide range of project activities, not just software development or Information Technology.  It’s also interesting that the researchers are university academics, as is the author of the Post piece (University of Virginia Darden School of Business faculty), so the utility of parametric estimation using historical information is becoming recognized to the point where it is touted in the mainstream press. 

Parametric, or ‘outside’ estimation is a familiar concept to QSM since it is a central tenet of the SLIM tool and methodology.  SLIM (both the tool and methodology) uses historical data from many past projects to make planning estimates for similar projects – those in the same reference class.  While originally developed to estimate software development efforts, it applies to more general information technology activities and, with appropriate historical project information, could be further expanded for use in planning general engineering projects.  SLIM’s reliance on historical data removes the optimism bias that leads to poor bottom-up or ‘inside’ project estimates.  QSM offers the SLIM toolset, targeted training and expert consulting support tailored to an organization’s planning needs with accurate and unbiased project estimates.   

Blog Post Categories 
Risk Management Project Management