/magi

Six Thinking Hats Reference

Purpose: Edward de Bono's parallel-thinking protocol that separates a single deliberation into six discrete cognitive modes (one "hat" at a time, the whole group wears the same hat together). Replaces argumentative cross-talk with sequenced single-mode passes, surfacing facts, feelings, risks, benefits, alternatives, and process control without the role-fixation that plagues classic debate.

Scope Boundary

• magi sixhat: parallel-thinking protocol applied to a proposal already on the table. Group cycles through six hats in a planned sequence. Output is a hat-by-hat synthesis feeding the standard Logos/Pathos/Sophia VOTE.
• magi decide / tradeoff / arbitrate / strategic (default lanes): independent three-perspective evaluation. Use when a structured verdict is the goal, not divergent surfacing.
• magi devil (elsewhere): formal opposition. Black Hat is one mode of six; Devil's Advocate is an entire structured red-team protocol with rotated dissenter, not a single-mode pass.
• magi delphi (elsewhere): anonymous multi-round expert convergence. Six Hats is single-room, named, parallel — not anonymous, not iterative across days.
• flux (elsewhere): reframes the problem (axes, lenses, paradigm shifts). Six Hats keeps the problem fixed and rotates evaluation modes.
• flux (elsewhere): iterative dialogue idea-deepening. Six Hats is single-pass per hat with strict mode discipline.
• omen (elsewhere): enumerates failure modes via FMEA/RPN. Black Hat overlaps but is qualitative, time-boxed, and embedded in a six-mode rotation.

Workflow

SETUP    →  facilitator picks Blue Hat opener; selects sequence (eval vs solve)
         →  state proposal in one sentence; agree time box per hat (3-5 min typical)

BLUE-1   →  define purpose, scope, success criteria for this session
         →  declare hat order; participants commit to mode discipline

ROTATE   →  whole group wears same hat at the same time
         →  no cross-hat interjections; out-of-mode comments parked by Blue

CAPTURE  →  scribe logs each hat's outputs as separate columns (W/R/B/Y/G)
         →  no confidence scores yet; surface-mode only

BLUE-2   →  summarize, choose action, route to Magi VOTE or extend a hat
         →  if Yellow+Green dominate Black: green-light; else loop

The Six Hats

Sequencing Protocols

Time-Boxing Defaults

Total typical session: 20-35 min for evaluation, 30-50 min for problem-solving.

Hat-Switching Rules

• Whole group, same hat: never split (e.g., "you take Black, I take Yellow") — that's role-fixation, not parallel thinking.
• No cross-hat in mode: a Black comment during Yellow is parked by Blue and surfaced when Black returns.
• Blue owns transitions: only the facilitator (Blue) calls the next hat; participants don't self-switch.
• Symmetry rule: every Black session must be paired with a Yellow session of equal time, to prevent bias toward criticism or optimism.
• Red is brief and unjustified: Red feelings are stated, not defended. "I feel uneasy" stands; "I feel uneasy because X" is Black in disguise.
• Green is judgment-free: ideas float without evaluation; evaluation happens only when Black/Yellow return.

Facilitator (Blue Hat) Script

• Opening: "We're deciding [proposal] in [time]. We'll use [sequence]. Hat order: [list]. I'll call hats and time-keep."
• Mid-session: "We're now wearing [Hat]. The question is [hat's prompt]. Out-of-mode contributions get parked."
• Park: "Good point — I'll park that as a [Black/Yellow/Green] item for when we get there."
• Closing Blue: "Summary: [White facts] + [Yellow benefits] + [Black risks] + [Red gut] + [Green alternatives]. Recommended action: [verdict]."

Tie-In With Magi's Three Perspectives

• White outputs feed Logos (facts, evidence base).
• Red and Pathos overlap (emotion, human impact) — Red is collective gut; Pathos is structured stakeholder modeling.
• Black informs Logos risk register and Pathos team-impact concerns.
• Yellow and Green inform Sophia (business value, alternatives).
• Blue maps to Magi's facilitator role at SYNTHESIZE.
• After Six Hats: pass hat outputs as evidence into the standard FRAME → DELIBERATE → VOTE workflow. Six Hats produces structured surfacings; Magi produces the verdict.

Anti-Patterns

• Splitting hats across people ("you're Black, I'm Yellow") — that's classic debate with color labels, not parallel thinking. Group must wear the same hat together.
• Skipping the second Blue — without a closing Blue, outputs scatter and no decision is named.
• Over-running Black without matching Yellow — produces a critique session masquerading as deliberation; pairs of equal time.
• Letting Red turn into argued opinion — "I feel X because Y, Z, and W" is Black in red costume. Cap Red at 60-90 seconds, no justification required.
• Using Six Hats on a problem with no proposal — switch to problem-solving sequence (Green-heavy) or run Flux first to reframe.
• Treating Six Hats as a verdict mechanism — it's a surfacing tool. Always feed outputs into Magi's VOTE phase for the actual decision.
• Running Six Hats with one person — it's designed for parallel group thinking; solo use collapses into ordinary checklist review and loses the parallelism benefit.
• Mixing hats in a single round (Black-then-Yellow ping-pong) — destroys mode discipline; use the planned sequence and hold the line.

Handoff

• To Magi VOTE: hat-column outputs become structured evidence for Logos (White, Black-risks), Pathos (Red, Black-team-impact), Sophia (Yellow, Green).
• To Flux: if Green Hat outputs reveal the original frame is wrong, halt and route to Flux for reframing before resuming.
• To Omen: Black Hat outputs that warrant quantitative scoring (RPN/AP) escalate to Omen for FMEA-grade enumeration.
• To Flux: if Green Hat surfaces an idea worth deepening rather than deciding, route to Flux for iterative expansion before returning to Magi.
• To human: if Blue closing reveals no consensus and Red gut-checks diverge sharply from Black/Yellow analysis, escalate as a values disagreement, not a fact dispute.