Delphi

Open

The Delphi technique utilizes a consensus-based approach to estimating involving multiple experts. The experts selected for a Delphi approach have subject domain experience and specific application knowledge. Experts conduct discussions in a structured group process designed to produce a consensus. The Delphi method of estimation in software development is the process whereby a group of experts identify the task to estimate, provide an estimation method, discuss the application of the method, and arrive at a consensus regarding the level of effort needed. The experts conduct the approach using multiple voting rounds that provide results that can be evaluated and summarized. The experts may repeat the process of estimate revision until the experts reach a specific number of rounds, reach a consensus, or until the results are stable and the answer is satisfactory. The Delphi approach to estimation makes an estimate available based on experts' collective agreement.

The Delphi approach utilizes expert assessments and involves the coordination of the team and elaboration of requirements for the members of the team to do their estimations anonymously. Estimations with a high level of variation are discussed further and re-evaluated. The results are distributed to the group for further discussion after each round to reach an agreement and review the agreement for relevance. The Delphi method captures factors from several experts and provides a defined practice in the assessment. The Delphi approach is the collective assessment of experts to establish an agreed-upon estimation.

Here's how it works:

1. Define the problem and select the experts: The problem or task that needs to be estimated is clearly defined, and a panel of experts is selected. These experts should have relevant knowledge and experience, but they do not need to be part of the project team.

2. Round 1 - Individual estimation: The experts are asked to provide their estimates independently. This is typically done through a questionnaire. They are also asked to provide justification or explanation for their estimates. The process is anonymous, to avoid any influence or bias.

3. Feedback summarization: The results of the first round of estimates are compiled and analyzed. The mean or median estimate is calculated, and the range of estimates is noted. Any justification or explanation provided by the experts is also summarized.

4. Round 2 - Review and revise: The experts are then given the opportunity to review the summary of the first round of estimates. Based on this feedback, they can choose to revise their original estimates. This process can be repeated in several rounds until a consensus is reached.

5. Consensus Decision: The final estimate is derived after several rounds when there is a consensus among the experts.

The Delphi method is particularly useful when the task being estimated is complex or unprecedented, and when there's a lack of hard data to inform the estimate. It leverages the wisdom of the crowd, allowing for a wide range of views and helping to avoid any one individual's bias dominating the process.

However, like all estimation techniques, it's not perfect. It can be time-consuming, particularly if multiple rounds are needed. The quality of the estimates also heavily depends on the expertise of the panel selected. And even though it seeks to minimize bias, it cannot eliminate it entirely.

Wideband Delphi

Wideband Delphi is an estimation technique that extends the classic Delphi method by adding more interaction and iteration between the participants. It's called "wideband" because it broadens (or widens) the amount of interaction and communication among the team members involved in the estimation process.

Here's how the Wideband Delphi process typically works:

1. Problem Definition and Team Selection: The problem or task to be estimated is clearly defined, and a team of experts is selected. These experts should have relevant knowledge and experience.

2. Individual Preparation: Each team member prepares for the estimation meeting by reviewing the task or problem and any associated documentation, and possibly by making some initial estimates based on their own knowledge and experience.

3. Estimation Meeting: The team then meets (either in person or virtually). The coordinator presents the task or problem to be estimated, and each team member presents their initial estimates and the assumptions they made.

4. Discussion: The team discusses the estimates, focusing particularly on areas where there is significant disagreement or variance. This discussion helps to uncover any misunderstandings, missing information, or differing perspectives.

5. Re-Estimation: After the discussion, each team member makes a new estimate based on the shared understanding developed during the discussion.

6. Repeat Discussion and Re-Estimation: The team repeats the discussion and re-estimation steps until a consensus is reached, or until the coordinator decides that enough convergence has been achieved.

7. Final Estimation: The coordinator then aggregates the final estimates into a single estimate for the task or problem. This could be a simple average, or it could be a more complex calculation.

The Wideband Delphi technique has several advantages. It allows for a broad range of perspectives and helps to uncover and resolve differences in understanding or assumptions. It also tends to result in more accurate estimates, as it leverages the wisdom of the crowd and minimizes individual biases. However, it can be more time-consuming than other estimation techniques, particularly if multiple rounds of discussion and re-estimation are required.