Public research infrastructure

Research Atlas Catalog

These are orientation layers over source literature. Every listed Atlas has a version, review date, source trail, claim ledger, declared scope, and stated exclusions. This catalog is not a journal or a claim of scientific consensus.

Educational research map · non-peer-reviewed orientation tool

The de Sitter / String Swampland Atlas

v1.2.0 · reviewed 2026-07-27

A source-first concept map and claim ledger for the de Sitter problem in the string/M-theory swampland program, built from a citation-verified literature map. An educational orientation tool that separates established results from active research, conjecture, and speculative interpretation.

Public records
12 claims · 16 concepts · 28 sources
Scope
The de Sitter problem in the string/M-theory swampland program, with a literature map, claim ledger, source trail, and a downloadable context pack.
Boundary
Does not adjudicate whether string theory admits metastable de Sitter vacua.
Expansion candidates (not public pages)
  • Add source-backed argument and response records only when the underlying literature map is expanded and re-verified.
  • Treat new camps, interpretations, and observational links as editorial candidates—not pages—until sources and status labels are reviewed.

First edition · 22 source-bounded claims

Synthetic Intelligence Atlas

v1.0.0 · reviewed 2026-07-27

A source-bounded map of large language models, agents, evaluation, and forecasting limits. Each public claim carries sources, a status, and limitations.

Public records
22 claims · 18 concepts · 33 sources
Scope
LLM mechanisms and scaling, post-training and inference-time reasoning, tool use and agents, benchmarks, construct validity, evaluation, and the limits of forecasting.
Boundary
Forecast scenarios and the local methodological audit exhibit are not in the public edition.
Expansion candidates (not public pages)
  • Resolve additional primary sources before adding any claim, source, or concept page.
  • Keep LLM/agent evaluation separate from a future quantum-computing atlas with its own evidence boundary.
  • Do not convert benchmark leaderboard changes or provider marketing claims into public pages without independently verified primary evidence.

Publication gate

  1. A public atlas requires a canonical path, semantic version, review date, explicit scope, explicit exclusions, and non-zero public claim, concept, and source records.
  2. It must publish machine-readable metadata, claim ledger, and source trail endpoints beneath its own canonical path.
  3. New pages begin as editorial candidates. They become public only after source resolution, status assignment, limitation writing, and a successful production build.