📄 Papers — Index & Guide

Overview

All academic papers processed for PUMA. Use Template-Literature-Note-Paper for new entries. Full SLR process: WF-SLR-Pipeline


Sub-folder Organisation

FolderPapers
Triage-Benchmarks/Papers on issue classification, bug triage, defect prediction
Effort-Estimation/Papers on story points, software effort, COCOMO derivatives
LLM-Agents-General/General LLM capabilities, agent frameworks, prompt engineering
PM-AI-Convergence/Surveys and overviews of AI in project management
Reproducibility-SE/Papers on reproducibility, open science, experiment design in SE

Quick Start: Adding a New Paper

1. Find paper via Perplexity / Semantic Scholar / IEEE / ACM
2. Verify in primary source (DOI, arXiv, or publisher)
3. Add to Zotero with tag #puma-pending
4. Use QuickAdd (Alt+L) → select "Literature Note Paper"
5. Fill YAML frontmatter (copy from template)
6. Read + Cornell notes + DRCA
7. Change tag to #puma-include or #puma-exclude
8. Link to relevant MOC: [[80 - MOC/81 Topic-Maps/MOC-Literature-Review]]

Current Papers Index

TABLE authors, year, read_status, prisma_decision, relevance
FROM "20 - Literature/20.1 Papers"
WHERE type = "literature-note" AND subtype = "paper"
SORT year DESC, relevance ASC


id: LN-Book-Flyvbjerg2023 title: “How Big Things Get Done — Flyvbjerg & Gardner (2023)” type: literature-note subtype: book tags: [literature, book, project-management, uniqueness-trap, planning-fallacy] authors: [“Flyvbjerg, Bent”, “Gardner, Dan”] year: 2023 publisher: “Crown Publishers” isbn: “978-0593239513” zotero_key: “Flyvbjerg2023” chapters_read: [“Introduction”, “Chapter on Uniqueness Trap”] puma_relevance: “Provides the Uniqueness Trap concept — the third problem dimension justifying PUMA’s historical-pattern learning via few-shot prompting.” read_status: processed created: 2026-03-01

How Big Things Get Done

One-sentence takeaway: Project managers systematically fail to use historical data from similar projects — the Uniqueness Trap — and LLMs with few-shot historical examples are a computational implementation of the prescribed fix (Reference Class Forecasting).

Core Argument

Flyvbjerg & Gardner analyse thousands of large projects and find systematic failure driven by: optimism bias, strategic misrepresentation, and the Uniqueness Trap. The solution is “outside view” forecasting using reference classes of similar past projects.

Key Concepts for PUMA

ConceptDefinitionPUMA Connection
Uniqueness TrapTreating every project as sui generis, ignoring base ratesFew-shot prompting provides “base rates” as examples
Reference Class ForecastingAnchor estimates to historical distribution of similar projectsFew-shot k=3/6 implements this computationally
Outside viewUsing statistical evidence rather than project-specific reasoning alonePUMA’s H2: few-shot > zero-shot for estimation

Cornell Notes

QuestionNote
What is the core prescriptive recommendation?Use reference class forecasting before inside-view planning
How does PUMA implement this?Few-shot examples = reference class; zero-shot = inside view
Is this empirically validated?Yes — analysis of thousands of infrastructure projects globally

Summary: Flyvbjerg demonstrates statistically that managers systematically under-use historical evidence. PUMA’s few-shot prompting strategy is a direct computational implementation of the prescribed solution. This grounds PUMA’s PM relevance in established management science.

đź”— Connections

Permanent note: 30 - Permanent/31 Concepts/PN-Uniqueness-Trap Cited in: PR-PUMA-Ch1-Introduction (context) Supports hypothesis: EX-Hypotheses-H1-H2 (H2 rationale)



id: LN-Book-Allen2001-GTD title: “Getting Things Done — David Allen (2001)” type: literature-note subtype: book tags: [literature, book, gtd, productivity, workflow] authors: [“Allen, David”] year: 2001 publisher: “Viking” isbn: “978-0142000281” zotero_key: “Allen2001” puma_relevance: “GTD framework used for task and project management within this vault.” read_status: processed created: 2026-03-01

Getting Things Done (GTD)

One-sentence takeaway: Capture everything → process ruthlessly → organise by context → review consistently → engage with trusted system.

GTD Principles in This Vault

PrincipleImplementation
Capture10 - Inbox — everything starts here
ClarifyDaily review: what is this? What’s the next action?
OrganiseTASKS-Active by phase and context
Reflect95 Reviews — daily + weekly templates
EngageDo the next action, not the project

đź”— Connections

Implemented in: 90 - GTD (entire folder) · 00 - Home (daily access) Related: 20 - Literature/20.2 Books/LN-Book-Ahrens-Zettelkasten



id: LN-Book-Ahrens-Zettelkasten title: “How to Take Smart Notes — Sönke Ahrens (2022)” type: literature-note subtype: book tags: [literature, book, zettelkasten, note-taking, knowledge-management] authors: [“Ahrens, Sönke”] year: 2022 publisher: “Sönke Ahrens (self)” isbn: “978-3982438801” zotero_key: “Ahrens2022” puma_relevance: “Zettelkasten method implemented in 30 - Permanent folder. Atomic permanent notes + linking = the knowledge architecture of this vault.” read_status: processed created: 2026-03-01

How to Take Smart Notes — Zettelkasten Method

One-sentence takeaway: Writing one permanent, atomic, linked note per idea forces understanding at the moment of capture and builds a network of knowledge that generates insights through unexpected connections.

Zettelkasten Principles in This Vault

PrincipleImplementation
Atomic notesEach PN- note = exactly one idea
Own words onlyNever copy quotes into permanent notes
Links over foldersConnections via wikilinks, not hierarchy
Evergreen notesNotes improve over time with maturity levels
Fleeting → Literature → PermanentThree-stage pipeline in vault structure

đź”— Connections

Implemented in: 30 - Permanent (entire folder) · Template-Permanent-Note Related: 20 - Literature/20.2 Books/LN-Book-Allen2001-GTD