Publications¶

All BDPD publications are available in the docs/publications/ directory. Source files are Quarto documents (.qmd) rendered to PDF and HTML. Presentations use RevealJS via Quarto. Blog and landing page are standalone HTML.

BDPD⁰ — Foundational¶

BDPD⁰ — A Computational Laboratory for Generative Agents, Commons Dilemmas and More

Roberto Brunelli

The foundational paper presents the BDPD framework, experimental results from both the platform and card game substrates, and LLM agent findings. It covers the Seneca Effect, Tragedy of the Commons, and wealth distribution dynamics as three interacting phenomena.

Resource	Path
Source (Quarto)	`publications/paper_00/index.qmd`
Full bibliography	`publications/paper_00/bdpd_full.bib`

Render:

cd docs/publications/paper_00
quarto render index.qmd

BDPD¹ — Governance¶

BDPD¹ — Governance: Cheap Talk and Sanctioning on a Fragile Commons

Roberto Brunelli

Closes the communication and sanctioning rows of Ostrom's design principles left as "not directly implemented" in BDPD⁰. Introduces a pact registry, a cheap-talk channel with five tool primitives, five governance metrics, and a sanction perturbation. Reports the Farrell-Rabin mini-pilot: at the canonical BDPD⁰ cliff, pure cheap talk does not save the commons — the diagnostic is silent defection. Graduated sanctioning (Ostrom principle #5) reliably preserves the commons; flat sanctioning does not.

Resource	Path
Source (Quarto)	`publications/paper_01/index.qmd`
Headline figure	`experiments/figures/pilot/d1_n5_headline__light_paper.png`
Raw pilot data	`data/pilot/d1_n5/aggregate.json`, `data/pilot/d1_cell_c/aggregate.json`
Bibliography	shared via `publications/paper_00/bdpd_full.bib`

Render:

cd docs/publications/paper_01
quarto render --to html

BDPD² — Vignettes¶

BDPD² — Governance of Nested Commons: Four Vignettes

Roberto Brunelli

Surveys the platform's governance affordances through six vignettes. Each vignette poses one research question, shows a result at N=5 seeds with both built-in and LLM players. Two recurring findings: qualitative direction of effects is robust to seed variation; LLM players consistently invert the built-in headlines.

Resource	Path
Source (Quarto)	`publications/paper_02/index.qmd`
Bibliography	shared via `publications/paper_00/bdpd_full.bib`

Render:

cd docs/publications/paper_02
quarto render --to html

BDPD³ — Seneca Polycentric Governance¶

BDPD³ — Governing the Signal, Not the Symptom: Leading-Indicator Regulation Prevents the Seneca Cascade in a Polycentric Commons

Roberto Brunelli

Promotes the S1/S2/S3 development probes into a third paper. Govern the visible signal (capital, leading) not the hidden symptom (pollution, lagging): pollution is the lagging tail of the industrial boom, so a P-reactive regulator is structurally too late, while the same cap armed on the capital stock prevents the cross-boundary cascade and rescues the emitter (coordination win-win). Includes the OFAT structural robustness sweep.

Resource	Path
Source (Quarto)	`publications/paper_03/index.qmd`
Source (Markdown)	`publications/paper_03/index.md`
Bibliography	shared via `publications/paper_00/bdpd_full.bib`
Robustness data	`data/robust/v11_s3/ofat.json`
Dev journal	`dev/v1.1_S3_exploration_journal.md`

Overview — The Why, the What, and the How¶

BDPD — The Why, the What, and the How: A narrative tour of the project

Roberto Brunelli

A no-code, no-equation overview written for the curious newcomer. Three chapters (Why / What / How) walk through the motivation, the parts of the laboratory, and the dual-substrate methodology. Includes a single mind-map figure that ties together the simulator, the card game, the agents, the governance toolkit, and the four research papers.

Resource	Path
Source (Quarto)	`publications/overview_00/index.qmd`
Bibliography	shared via `publications/paper_00/bdpd_full.bib`

Render (HTML + PDF):

cd docs/publications
./render_publications.sh --which overview_00

Pitch¶

Two RevealJS presentations in the BDPD theme (Cinzel + Share Tech Mono), each shipped in dark and light variants.

Deck	Scope	Slides
`pitch_00/`	BDPD⁰ only — the original platform-paper pitch (one-aggressor cliff, dual-mode laboratory, safe betrayal, compensator).	13
`pitch_01/`	Full arc ⁰→³ — single-deck overview of all four papers: BDPD⁰ cliff + BDPD¹ governance (talk vs ladder) + BDPD² polycentric inversion + BDPD³ Seneca leading-signal win-win.	15

Resource	Path
Pitch⁰ dark / light	`publications/pitch_00/bdpd-pitch-{dark,light}.qmd`
Pitch¹ dark / light	`publications/pitch_01/bdpd-pitch-{dark,light}.qmd`
Dark / Light SCSS (both)	`publications/pitch_NN/bdpd-{dark,light}.scss`

Render (both decks, dark + light, HTML + PDF):

cd docs/publications
./render_publications.sh --which pitch         # all pitch decks
./render_publications.sh --which pitch_01      # one deck only

Outputs are self-contained HTML (~7 MB) and PDF (~2 MB) per variant.

Blog¶

Standalone HTML articles in literary-essay register, sharing the same typographic skin (Lora serif, Source Sans 3, amber accent on a warm cream background). Each post is self-contained for hand-off / re-host.

Post	Theme	Read time
`blog_00/` — What Happens When One Bad Actor Enters the Commons?	BDPD⁰ headline. The one-aggressor cliff, the tragedy of the compensator, the framing piece for the whole programme.	14 min
`blog_01/` — Why Communication Wasn't Enough	BDPD¹ D1. Cheap talk does not change preservation rate (4/5 → 4/5) but redistributes wealth (−33% aggressor, +17% cooperators). The dominant axis is architecture, not channel.	14 min
`blog_02/` — The Slow Catastrophe Has a Tell	BDPD³ S3. Why regulators armed by lagging signals (pollution) fail while the same lever armed by leading signals (capital) rescues both the emitter and the downwind cascade. The trigger, not the lever, is binding.	16 min
`blog_03/` — What a Card Game Knows That a Differential Equation Doesn't	The dual-mode methodology. Safe betrayal emerged in CT4 on the tabletop before P11 confirmed it; the CT3/P9 noise divergence sharpens the mechanism. The card game catches strategies the simulation's heuristic library does not yet contain.	14 min
`blog_04/` — You Can't Save the Polluter by Punishing Households	BDPD³ S3 negative control: the `fine` lever fails under every trigger because the model has no demand→production channel. The diagnostic: name the channel before deploying the lever. Companion piece to `blog_02`.	13 min
`blog_05/` — The Mule, the Compensator, and the Inevitability of Late Defection	Cross-paper essay (BDPD⁰ P11 + BDPD¹ D2/D3). Why the savings-account intuition fails on a structurally fragile commons, and why graduated sanctioning is the only mechanism we found that disciplines the late-defection attractor.	15 min

Resource	Path
Sources	`publications/blog_NN/bdpd-blog-article.html` (standalone HTML)

Annotated Bibliography¶

The full BDPD bibliography with abstracts, generated dynamically from the BibTeX source.

Resource	Path
Source (Quarto)	`publications/bib_00/extended_bib.qmd`
BibTeX source	`publications/paper_00/bdpd_full.bib`
Rendered HTML	`publications/bib_00/_site/bdpd_extended_bib.html`

Render:

cd docs/publications
./render_publications.sh --which bib --html-only

Figures¶

Two figure locations are recognised by the publications system:

Location	Used for	Referenced as
`publications/pictures/`	Static figures (hand-curated)	`../pictures/<filename>`
`experiments/figures/pilot/`	Regenerable pilot plots (re-emitted by `scripts/plot_*.py`)	`../../../experiments/figures/pilot/<filename>`

The static pictures/ folder is the canonical centralised location for schematic diagrams, card images, and any plot whose source data does not live in data/pilot/. The experiments/figures/pilot/ path is reserved for plots whose canonical generator is a checked-in plot_*.py script reading from data/pilot/<pilot>/aggregate.json: keeping them in their generation location avoids stale-copy drift when a pilot is re-run.

publications/pictures/                  experiments/figures/pilot/
├── CT1_collapse_matrix__light_paper.png  ├── d1_n5_headline__light_paper.png
├── CT2_gamelength_sweep__light_paper.png  ├── d2_llm_n5_headline__light_paper.png
├── ...                                    ├── d2_ladder_sweep_headline__light_paper.png
└── warrior_vs_temple__light_paper.png     └── …

Papers referencing a regenerable plot must use the long ../../../experiments/figures/pilot/ path; render_publications.sh includes the per-paper post-processing that maps these paths correctly into the mkdocs site/ build (../../../experiments → ../../../../experiments).