Brainwriting

Quick facts

Mechanism: Generates ideas the way silent parallel brainwriting does, producing several independent idea streams that build on each other without anchoring on the first voice, then consolidates them into a shortlisted idea pool.

Produces: idea-pool

Evidence: S · Domain: Divergent Ideation

Use when you need breadth of options and want to avoid the anchoring and conformity that make ordinary brainstorming underperform.

Verbal group brainstorming reliably underperforms: only one person talks at a time (production blocking), and the room anchors on the first speaker. Brainwriting removes those losses by generating ideas silently and in parallel, then building on each other’s written ideas. This skill adapts that for solo-plus-AI: produce several genuinely independent idea streams (distinct angles, generated without contaminating each other), then run a build-on round and consolidate into a shortlist. The output is an idea pool. The key is keeping the streams independent first; collapsing into one stream throws away the mechanism that does the work.

When to Use

• You need breadth of options and want to avoid premature convergence.
• Anchoring or conformity would otherwise narrow the ideas (a single brainstorm).
• Early divergent generation, before any selection.

When NOT to Use

• When you need to converge, rank, or decide (use a decision skill).
• When deep single-expert reasoning beats breadth.
• If the result would be volume with no build-on and no selection.

Instructions

When asked to brainwrite, follow these steps:

1. State the prompt for ideation in one line.
2. Generate independent streams. Produce 3 to 4 separate idea streams, each from a distinct angle, persona, or constraint, generated as if blind to the others. Keep them visibly separate. Aim for a quota per stream.
3. Build on, across streams. Now combine and extend: take ideas from one stream and push them with another’s logic. This is where brainwriting beats parallel lists.
4. Consolidate and de-duplicate. Merge near-duplicates; keep the distinct ideas.
5. Shortlist. Select the strongest ideas to carry forward, noting why.
6. Emit the idea pool per references/TEMPLATE.md.

Output Format

Use the template in references/TEMPLATE.md. The deliverable is the separate streams, the build-on round, and a consolidated shortlist, not prose.

Quality Checklist

Before finalizing, verify:

• The streams were genuinely independent, not one list relabeled.
• A build-on round combined and extended across streams.
• Near-duplicates are merged; distinct ideas kept.
• A shortlist is selected with reasons.
• The output is the idea pool artifact, not prose.

Evidence

Tier S. Across decades of replicated studies, silent parallel generation (nominal groups, brainwriting) outproduces traditional verbal brainstorming in quantity and quality of ideas, because it removes production blocking and anchoring (Diehl & Stroebe 1987; Mullen, Johnson & Salas 1991 meta-analysis; methods from Rohrbach 1968 and Delbecq & Van de Ven 1971). The strong evidence is for human groups; the solo-plus-AI adaptation transfers the mechanism (independent parallel streams plus build-on) but does not inherit a measured AI effect size. Full grading: evidence/dossier.md.

Examples

See references/EXAMPLE.md for a completed idea pool.

Deep dive: worked example

A full worked run (the shared Northwind scenario)

Idea Pool - Worked Example

A completed run of think-brainwriting, on the shared Northwind scenario. This is the quality bar a generated pool should meet.

Northwind is a B2B SaaS. Here the skill generates growth ideas with independent streams to avoid everyone anchoring on “free tier.”

---

Prompt

• Ways to generate qualified pipeline for the Q3 target.

Independent streams (generated as if blind to each other)

Stream A (the growth marketer):

• Self-serve free tier, ICP-gated
• Reverse trial (full features, then downgrade)
• Referral loop with in-product incentives

Stream B (the sales leader):

• Outbound to a tight ICP list with free pilots
• Re-engage closed-lost from the last 12 months
• Partner co-sell with an adjacent vendor

Stream C (the product lead):

• Fix trial onboarding so existing trials convert
• Expansion motion inside current accounts (more seats/use cases)
• A high-value template/integration that pulls in qualified teams

Build-on round (combine and extend across streams)

• ICP-gated free access (A) offered only to outbound-qualified prospects (B) - free as a sales-assist, not open top-of-funnel.
• Fix onboarding (C) and instrument it as the test of whether packaging was ever the problem before building free (A).
• Expansion motion (C) plus closed-lost re-engagement (B) - growth from accounts that already know us, no new funnel.

Consolidated shortlist

1. Fix-onboarding-first (C): cheapest, most reversible, and directly tests the free-tier premise. Do now.
2. Expansion + closed-lost (B+C): qualified pipeline from warm sources; low cannibalization risk.
3. ICP-gated free as sales-assist (A+B): keeps the free upside while limiting non-ICP cost; pilot it.

---

Note: the value is that three independent streams surfaced options a single “should we build a free tier?” brainstorm would have anchored away from. The build-on round produced the strongest idea (free as a gated sales-assist) by crossing the marketer’s and sales leader’s streams. Carry the shortlist into a decision review.

Grounding: the full evidence dossier

What the research does and does not show, with graded sources

Evidence Dossier: Brainwriting (6-3-5 / Nominal Group Technique)

Single source of truth for the brainwriting skill. The SKILL.md, sidecar, and evals derive from this. This is one of the library’s strong-evidence anchors.

Skill	thinking-framework-skills.brainwriting (installable name think-brainwriting)
Family	divergent-ideation
Evidence tier	S (strong; replicated experimental evidence)
Confidence	High - this is among the best-evidenced ideation findings
Status	draft (authored 2026-05-31 from the discovery corpus)

1. The mechanism (what actually does the work)

Verbal group brainstorming reliably underperforms because of production blocking (only one person talks at a time, so ideas are lost or suppressed while waiting), plus anchoring on the first speaker and social conformity. Brainwriting removes those losses: people generate ideas silently, in parallel, in writing, then build on each other’s written ideas. 6-3-5 is one protocol (6 people, 3 ideas each, 5-minute rounds, pass and build); Nominal Group Technique (NGT) is the close cousin (independent silent generation first, then round-robin sharing and ranking). The active ingredients are silent parallel generation, independence before discussion, and structured build-on.

Adaptation for solo-plus-AI: the agent simulates several genuinely independent idea streams (distinct angles or personas generating without seeing each other first), then builds on them and consolidates. This captures the mechanism (parallel independent generation + build-on) even without multiple humans. Be honest that the social diversity of real participants is only approximated.

2. Lineage

• Brainwriting 6-3-5 (Bernd Rohrbach, 1968). Nominal Group Technique (Delbecq & Van de Ven, 1971).
• Evidence base: Diehl & Stroebe (1987) demonstrated production blocking in brainstorming; Mullen, Johnson & Salas (1991) meta-analysis found nominal/brainwriting groups reliably produce more and better ideas than interacting verbal brainstorming groups.

No trademark. Named descriptively.

3. What the evidence shows, and what it does NOT show

Strongly supported (the S): across decades of replicated studies, silent parallel generation (nominal groups, brainwriting) outproduces traditional verbal brainstorming in both quantity and quality of ideas. As the discovery research puts it, “brainstorming the activity has consistently failed replication for nearly 40 years; the underlying mechanisms - silent parallel generation, idea quotas, structured build-on - work robustly.” This is a genuine strong-evidence result and a credibility anchor for an evidence-graded library.

Boundaries (still honest): the evidence is about humans in groups. The solo-plus-AI adaptation transfers the mechanism; it does not inherit a measured effect size for AI use. And brainwriting improves generation; it is not a selection or decision method.

4. Transferred-evidence flag

The strong evidence is for human groups. The AI adaptation (simulated independent streams) is transferred, not AI-validated. The AI value is real regardless: parallel independent generation plus build-on is a structured counter to a model’s tendency to produce one anchored stream of ideas.

5. When it works / when it fails

Works best when: you need breadth of options and want to avoid anchoring and conformity; early divergent generation; when a single brainstorm would converge too fast.

Fails or misleads when (poor-fit / anti-patterns):

• Collapsing into a single idea stream, losing the parallel independence that does the work (the central failure mode for the AI adaptation).
• Generating volume with no build-on and no selection.
• Treating it as a decision or convergence method.
• When deep single-expert reasoning beats breadth.

6. Output artifact

An idea pool: several independent idea streams generated in parallel (kept visibly separate at first), a build-on round that combines and extends across streams, and a consolidated, de-duplicated shortlist of the strongest ideas.

7. Sources

1. Rohrbach, B. (1968) - Method 6-3-5 (brainwriting).
2. Delbecq, A., & Van de Ven, A. (1971) - Nominal Group Technique.
3. Diehl, M., & Stroebe, W. (1987) - production blocking in brainstorming.
4. Mullen, B., Johnson, C., & Salas, E. (1991) - meta-analysis: nominal groups outperform interacting brainstorming groups.

Verification status: these are well-attested, frequently-cited findings. The “S” grade is justified for the human evidence; keep the AI-adaptation caveat (mechanism transferred, effect size not measured for AI) visible in any public claim.