The library has grown: the facilitator’s role, choosing the panel and AI panelists
Hello,
Pronoia's method library has grown by six new articles in two weeks, and many earlier articles have gained new content. Pronoia's interface is demanding at first, which is why the guidance and articles have an important job in making it clearer to use.
The facilitator's role: the Delphi manager
In the Finnish Delphi tradition, the researcher who leads a study is called the Delphi manager. The name says something the term facilitator does not: the role is proactive. You shape the panel's interaction, decide what goes into the next round, and draw attention to discussions that open up new possible futures.
The new article “The Delphi manager: the facilitator's role and ethics” brings the manager's tasks together into one whole and draws the line between facilitation and manipulation. Motivating and informing the panel is the manager's work; the manager's own positions are not.
Who belongs on the panel? Expertise and stake
The panel matrix shows which perspectives the phenomenon under study calls for. It does not tell you whom to choose or how to get them on board. The new article “Expertise and stake in choosing the panel” covers that too.
- Two criteria. A good panelist knows the phenomenon well, and often also has something at stake in its future. In Delphi, interest-based argumentation is to be expected; there is nothing in it to reject or to be suspicious of.
- Reserve places. Besides those who know and those who act, the panel needs generalists (people who see the whole picture) and at least one credible dissenter.
- Recruit in person. Tell panelists why they in particular were chosen, what they get for their trouble and what is expected of them.
The article on the panel matrix now also describes the inner and outer panel: the phenomenon's own experts in one group, those who look at it from outside in another, and dialogue between the two.
An AI panelist is a perspective anchor, not an expert
AI panelists understandably raise the most questions, and Pronoia's library now answers them in three articles. There is of course no need to read them if you don't plan to use artificial panelists, which is the starting point for most panels. But if you do bring AI panelists in, it pays to do so transparently and with the ethical criteria in view:
- “Documenting AI panelists” explains how to record an AI panelist's sources, profile, testing and every activation, so that the reader of your report knows whose perspective the panelist represents.
- “How Pronoia's AI panelists work” describes how the system works: when AI is used, what is recorded in the log (the event record) and how every activity shows which AI model wrote it. You can cite it in your own report and record only your own settings and decisions.
- “AI usage modes and costs” has been updated to match all the other AI usage modes.
The shared message is this: an AI panelist argues consistently from one defined perspective. It can give a voice to a stakeholder missing from the panel, or raise an objection no human panelist wants to make. It is not an expert, because its participation rests only on what has been described and written about the subject in advance.
Delphi as a learning environment
The new series “Pedagogical Delphi” starts with two articles. The introduction describes how rounds, reasons and dialogue teach participants to form a reasoned view, to listen to different perspectives and to change their mind when given a good reason.
“Role Delphi in Pronoia” shows how a panelist argues from an adopted role: from the perspective of a stakeholder, a profession or a historical figure. The role creates distance from one's own fixed ideas, and it is a way to find good arguments for futures one would otherwise avoid. The series will continue.
Refinements to familiar articles
- The marks of a good thesis. “Thesis structure” now lists six marks: clear, interesting, divisive, describing change, concrete and neutral. Write the thesis like a news headline, and avoid negative statements: respondents easily confuse which side they are taking.
- Time horizon in the phenomenon description. Take the time horizon far enough that immediate interests fade. For most people, the line between the near and the distant future lies about ten years ahead.
- Three ways to end. “Consensus vs. dissensus” explains how the study's aim decides whether it ends in a recommendation for action, in options left open for decision-makers, or in scenarios.
- Sources in view. Publications from the Finnish Delphi tradition have been added as sources to the variation articles and to “What is the Delphi method?”.
Where to find the articles
All articles are available in Finnish, English and Swedish. They open from Pronoia's guidance with the ? Help button, and Ask Pronoia answers on the basis of them.
If an article leaves a question open or a topic is missing, tell us with the Share an observation button. The library grows from exactly the questions facilitators run into.
Hannu Linturi
Metodix Oy
info@metodix.fi