OFL Glossary
Core terms and definitions used across the Open Facilitation Library.
A
Adversarial Cross-Pollination
A variant of cross-pollination that deliberately exposes participants to maximally opposing viewpoints, measured by embedding distance.
Artifact
Any output produced during a facilitation process that can be passed to subsequent phases or sessions. Examples: summaries, vote tallies, clustered opinions.
Asynchronous Facilitation
Facilitation conducted over time without requiring simultaneous presence of all participants. Enables scale and thoughtful responses.
C
Conversation Signature
A vector representation of a conversation’s characteristics based on dialogue act tagging. Used to compare conversations to known facilitation styles.
Cross-Pollination
A deliberation technique where participants are exposed to opinions different from their own to broaden understanding and foster perspective-taking.
Clustering
The process of grouping similar opinions or responses together to identify distinct viewpoints within a larger group.
D
Daisy-Chaining
Running multiple facilitation sessions in sequence, where outputs from one session become inputs to the next. Enables complex methodologies through simpler building blocks.
Deliberative
A category of facilitation focused on collective decision-making, weighing options, and building consensus.
Dialogue Act
A unit of conversational action (question, statement, clarification, etc.) that can be tagged for analysis.
E
Evaluation (Process-based)
Assessing facilitation quality by measuring characteristics of the process itself (intervention timing, question types, participation balance).
Evaluation (Outcome-based)
Assessing facilitation quality by measuring results (agreement reached, participant satisfaction, action completion).
F
Facilitation
The practice of guiding group processes to achieve productive outcomes. Distinct from moderation (which enforces rules) in that facilitation actively encourages participation and shapes discussion flow rather than just preventing violations. See Facilitation in the AI Era for practitioner perspectives.
Facilitation Primitive
A basic building block of facilitation that can be composed into larger methodologies. Examples: probing question, summarization, reflection prompt.
Fine-tuning
Training a language model on specific data to improve performance on particular tasks. Contrasts with prompt/skill-based approaches at inference time.
G
Generative
A category of facilitation focused on ideation, creativity, and producing new ideas or possibilities.
H
How (in Why-How-Who)
The process dimension of facilitation: what techniques are used, how the facilitator intervenes, and the structure of the interaction.
I
Inference-time Skills
Enhancing AI capabilities through prompts, tools, and context rather than model fine-tuning. The approach used by Harmonica.
Intervention Style
How a facilitator engages with participants: non-directive (minimal guidance), semi-directive (questions and summaries), or directive (active guidance).
M
Moderation
Ensuring orderly interaction by enforcing community guidelines and removing violations. Distinct from facilitation in that moderation is primarily reactive (responding to problems) rather than proactive (shaping discussion). The WHoW Framework provides a taxonomy of moderation actions.
O
One-to-One
Facilitation mode where the facilitator (human or AI) interacts with one participant at a time. Enables depth but requires synthesis across participants.
Opinion Clustering
See Clustering.
P
Pattern
A documented facilitation methodology in machine-readable format, including purpose, process, and participant information.
Perspective-Taking
An outcome goal of facilitation where participants gain understanding of viewpoints different from their own.
Plenary
A facilitation mode involving the entire group simultaneously, typically for information sharing or final decisions.
Preference Elicitation
An outcome goal of facilitation where individual preferences are surfaced and captured.
Psychological Safety
A shared belief that the group is safe for interpersonal risk-taking. In facilitated settings, psychological safety enables participants to share honest views, disagree openly, and admit uncertainty. Creating this safety is often cited by practitioners as foundational to effective facilitation. See Facilitation in the AI Era.
R
Reinforcement Learning (for Facilitation)
Training AI facilitators using feedback signals derived from conversation quality metrics and Why-How-Who evaluation.
S
Synchronous Facilitation
Facilitation conducted in real-time with all participants present simultaneously.
Synthesis
An outcome goal of facilitation where diverse inputs are aggregated into coherent, actionable outputs.
T
Transformative
A category of facilitation focused on conflict resolution, perspective shifts, and relationship repair.
Turn-taking
The pattern of who speaks when in a conversation. In facilitated settings, balanced turn-taking (where participants contribute roughly equally) is often a goal. Unequal turn-taking can indicate power imbalances or dominant voices crowding out others. The Fora corpus includes analysis of turn-taking patterns in facilitated dialogue.
W
Why (in Why-How-Who)
The purpose dimension of facilitation: what outcomes the method aims to achieve (agreement, ideation, perspective-taking, etc.).
Who (in Why-How-Who)
The participant dimension of facilitation: group size, roles, power dynamics, and interaction modes.
Why-How-Who Framework
An evaluation framework for encoding and comparing facilitation methodologies across three dimensions: purpose, process, and participants.