discussion-document-writing

Discussion Document Writing

Produces Takshashila-style discussion documents: longer analytical pieces that map a policy problem, synthesize evidence, identify trade-offs, and frame the debate — without necessarily committing to a single recommendation.

Purpose

Discussion documents differ from policy briefs in that they:

• Are written for an informed audience (researchers, practitioners, policymakers)
• Explore multiple perspectives and trade-offs more fully
• May present a range of recommendations or leave open questions for debate
• Are more suitable for complex, contested, or early-stage policy questions

Structure

Abstract (150–200 words)

• What question does this document address?
• What is the scope?
• What are the key findings or arguments?

1. Introduction

• Frame the policy problem.
• State why it matters now.
• Describe the structure of the document.
• 300–500 words.

2. Context and Background

• Regulatory, institutional, historical context.
• Key actors and their interests.
• Existing policy landscape.
• 500–1000 words.

3. Problem Analysis

• Disaggregate the problem into its components.
• Use frameworks where useful (e.g., market failure types, principal-agent problems, coordination failures).
• Draw on evidence: data, case studies, comparative examples.
• 800–1500 words.

3b. Causal Analysis (mandatory)

• State the central causal chain as a single sentence: "X → Y → Z"
• Identify the weakest link — the one step that, if broken, collapses the argument.
• Name the key reinforcing and balancing loops (e.g., R1 "ecosystem flywheel", B1 "budget constraint").
• Map cross-connections between loops: where does one loop undermine or amplify another?
• If the system has stacked interdependencies, name the layers and show which layer the intervention targets.
• Flag unsupported causal links explicitly.
• Use causal-loop-analysis skill to produce a Mermaid diagram and full loop inventory.
• 200–400 words of prose + diagram.

4. Policy Options / Debate

• Present the main positions or approaches in the literature/debate.
• For each: evidence base, assumptions, trade-offs, feasibility.
• Be fair to positions the author may not personally favor.
• 500–1000 words.

5. Analysis and Implications

• Author's own assessment.
• Which approaches are most promising? Why?
• What are the key uncertainties?
• 400–800 words.

6. Conclusion

• Summary of key findings.
• Open questions for further research or debate.
• 200–300 words.

References

• Comprehensive; all claims sourced.

Process

1. Ask the user: what is the core question this document addresses?
2. Ask: is this meant to conclude with recommendations, or is it primarily mapping the debate?
3. Pull sources from Zotero; supplement with web search.
4. Draft abstract and section headers — share structure with user before drafting body.
5. Draft section by section.
6. Apply writing-anti-ai pass.
7. Offer to export to Google Docs.

Quality Checklist

• Abstract can stand alone
• Problem clearly framed with evidence
• Causal chain stated explicitly; weakest link identified
• Key loops named with descriptive names (not just R1/R2 labels)
• Cross-connections between loops noted where they exist
• Multiple perspectives fairly represented
• Trade-offs explicit
• Evidence cited throughout
• No AI writing tells
• Length 2000–6000 words

Length Norms

• Short discussion note: 2000–3000 words
• Standard discussion document: 3000–5000 words
• Extended paper: 5000–8000 words