Delphi Method


For awhile now, I've been interested in the Delphi Method, a process originally designed by the Rand Corporation to help a group of varied experts discuss complex issues and reach consensus when only partial information is present. It works like this:

First, you have a central question that needs to be answered. The process takes place over multiple rounds. In each round, every participant writes down their initial thoughts and statements regarding the answer. The participants' statements are collected, copies made, and then re-distributed to everyone, anonymously. Then everyone reads all the thoughts, and comments on the ideas of others and the progress of the group. At any point they can revise their earlier statements, again anonymously. The process repeats for several rounds until a consensus is reached. Even after the process is concluded, all results remain unattributed and anonymous.

This method has some advantages:

Time delay: it allows the process to work semi-asynchronously, and makes it so that participants can respond to ideas more thoughtfully without the time pressure of a "debate"-like situation. Rounds can even take place over the span of a few days, using a medium like email.

Anonymity: it removes many troublesome group dynamics that arise in group decision making, such as the effect of particularly charismatic or domineering individuals, or participants who are in a position of authority. It also allows participants to more easily retract an earlier view about which they've changed their mind because it's easier to admit a mistake anonymously.

Possible problems:

Collection and redistribution of each round of results needs to be done by an objective facillitator. This is a point of possible corruption. Sometimes the thoughts need to synthesized or reformatted, which opens the door for corruption through misinterpretation (even inadvertently). This might not be a problem, as the participant who recognizes in the next round that their response was mischaracterized could send a clarification.

Having worked in a semi-large company, I've witnessed and parcipated in many instances when important decisions needed to be made jointly by a number of expert stakeholders, each with a critical piece of the puzzle, but where the decision-making proceeded very inefficiently due to the friction of interpersonal group dynamics. Such problems seem to be inevitable, so I've often idly toyed with the idea of creating a web application that allows you to create a topic question, invite participants, whereupon it manages the collection and redistribution of responses.

Originally posted here on 2006 Oct 23, by Yishan Wong.