About the Knowledge Commons

The Knowledge Commons is a search and discovery tool that brings together the scientific output of RMBL and the Gunnison Basin into one searchable platform. It includes peer-reviewed publications dating back to 1928, community and policy documents from the Sustainable Living Library, and research datasets from multiple repositories. A knowledge graph connects these resources through shared species, concepts, research methods, and geographic locations.

The Hub is designed for researchers, students, land managers, community members, and policymakers interested in the environmental research and stewardship of the Gunnison Basin. It is equally useful for scientists looking for related work and for community members exploring how research connects to local policy issues.

Knowledge Neighborhoods are research communities detected automatically by analyzing the connections in the knowledge graph. Using a community-detection algorithm (Louvain), the system identifies clusters of tightly connected authors, publications, species, concepts, and places. Each neighborhood represents a distinct research theme — from marmot behavioral ecology to watershed biogeochemistry to federal land management policy. Many neighborhoods include AI-generated research primers that summarize the key findings and cite specific publications.

Research Frontiers are synthesized boundaries between what scientists know and what they don't, with identifiable paths to push the boundary forward. The system extracts atomic gap-statements from neighborhood research primers, clusters them by semantic similarity, and uses a language model to weave each cluster into a narrative with context, key questions, barriers, opportunities, and concrete actions categorized by category (data, experiment, model, synthesis, framework, etc.) and effort tier (near-term, ambitious, major, consortium). Each frontier links back to its contributing neighborhoods, source statements, and the strongest concepts, species, places, and protocols involved — so you can trace any claim back to the underlying evidence.

The Future Scenarios Framework (currently v0.16, version-controlled at specification/Future_scenarios_framework.md) sets the rules: what a scenario is, what fields it must include, what register the prose should be in, what is forbidden, and how multiple scenarios within a set must be distinguishable from each other. Each scenario starts as a structured YAML entry with a distinguishing thesis, a magnitude bracket, a frontier portfolio, and (for upside/downside sets) the favorable or unfavorable conditions it depends on. RMBL staff write these inputs.

Claude Opus drafts each scenario by reading the spec sections, the YAML inputs, the RMBL institutional grounding, the candidate research frontiers from the Commons, and the existing era primers. The model's output is the prose body of each scenario .md file. Stories are drafted the same way against a separate story-prompt that reads the scenario as context plus a linked Commons frontier the protagonist is pushing.

The central set (centennial-2027, twelve scenarios) is contingency-honest about realistic-bracket campaign outcomes — each scenario names what would invalidate it and what other scenarios exist as alternatives. The upside companion set (centennial-2027-upside, three scenarios) explores what becomes possible when several favorable conditions stack. The downside companion set (centennial-2027-downside, three scenarios) explores what becomes the texture when unfavorable conditions stack. Each set has its own forbidden-pattern rules — the upside set must avoid utopian register, the downside set must avoid collapse register.

Each scenario has at least one companion story — short literary fiction (1,400–1,600 words) grounded in the scenario. Characters are fictional roles, not real RMBL staff or guest scientists. The fictional voice helps readers inhabit possibilities at a register the strategic-planning artifacts cannot reach.

The prompt assembles a tiered context: ~15 landmark publications (abstracts + key findings), ~15 frontier publications from the last few years, ~60 breadth publications (one finding each), plus all linked concepts with definitions. For policy-flavored neighborhoods, a parallel prompt includes management documents and federal-register notices. The prompt asks for citation-grounded prose — every claim should be traceable to a specific publication in the input.

A primer is a synthesized map of what is in this neighborhood and how the work connects, not a peer-reviewed literature review. The cited publications are real and grounded in the Commons; the synthesis itself is the model's reading. Verification is a click away — every citation links to the publication's detail page.

The exact prompt templates Claude Opus receives live in generate-primers.ts (RESEARCH_PROMPT at L42, POLICY_PROMPT at L85). Reading them is the most direct way to verify what the model was asked to do.

The prompt assembles publications and documents whose year falls in the era's window, ranked by citation count and topic-distinctiveness. The model is asked to characterize the period's questions and methods without pretending to be authoritative — to surface trends, not write a history.

Read it as a synthesized characterization of a research period, not as an authoritative history. The publications cited are real and linked; the period framing — what was distinctive, what shifted — is the model's reading of the corpus.

The exact prompt templates Claude Opus receives live in generate-era-primers.ts (PROMPT at L87, PROMPT_CENTURY at L154 for the century-scale variant).

Five stages: (1) per neighborhood, the LLM reads top papers and emits atomic frontier statements, each with a verbatim cite snippet — statements that can't be grounded in source text are dropped; (2) the statements are embedded with Voyage AI and clustered by greedy-centroid similarity with recency weighting; (3) the LLM synthesizes each cluster into a named frontier with key questions and data gaps that carry the verbatim cite snippets through; (4) the frontier loads to the DB with snapshot history; (5) a currency-validation pass asks per question whether newer papers in the same neighborhoods have addressed it.

An earlier (“legacy”) set of 98 frontiers was synthesized from neighborhood-primer prose rather than direct paper grounding. Those are kept in the DB because they anchor the downstream planning corpus, but they're hidden from the default /frontiers view; surface them with the Include legacy filter chip.

Read a frontier as a synthesized articulation of where the literature points toward a knowledge boundary, not as an authoritative research agenda. Click any cite chip on a key question to open the source paper and verify the snippet word-for-word. The contributing neighborhoods are listed so you can trace the frontier back to its evidence base.

The grounded extractor lives in extract-frontiers-grounded.ts, the synthesizer in , and the currency validator in . Full design spec: .

Each cluster of related planning items (actions, questions, data gaps, barriers, or impacts) gets a title and summary synthesized by Claude Opus. View the prompt: describe-frontier-planning-clusters.ts.

Second-order Louvain over cluster descriptions surfaces twelve cross-lens themes. Each gets an invitational opportunity statement synthesized for planning conversation. View the prompt: describe-planning-themes.ts.

A final cross-theme synthesis stage surfaces strategic opportunities that scale beyond the basin. View the prompt: synthesize-long-reach-opportunities.ts.