Covering Human Behavior in Software Project Estimation

Several techniques for estimating the software development effort are known and commonly used. Most of them do not take into account the variable most difficult to estimate: human behaviour and its impact on the software process. People have various motivations, qualifications, workloads, abilities to communicate with their peers or supervisors, working environments, strategies for seeking help, etc. All those factors can create additional workload somewhere in the project organisation which has to be covered by a realistic estimation. Our long-range vision is that managers will design their work processes and organisations by synthesising, analysing and evaluating alternative “virtual prototypes”. We introduce VitéProject, a tool for analysing work processes and organisations. VitéProject conceptualise organisations as consisting of a web of communication channels wherein information is processed at the nodes (i.e., people or actors). Thus far, the human actors have been viewed as information-processing agents, whose part of the process has been to administer a sequence of tasks and take decisions based on their role as actor, and their skill level. We would like to present ideas, based on our experiences with applications of theory and knowledge of individual differences and group processes, helping to avoid the most common problems faced while developing software: Not delivering according to schedule, budget and/or the requested quality. By using personality taxonomies in extending VitéProject to account for human variation in personality, simulations may be made much more realistic. Following the “Big 5” personality taxonomy, there are five mutually independent personality traits that account for most of the useful variation in personality. Given this premise, simulating personality differences becomes a manageable task.