[ RESACH · BY THINK NORTH ]
Give it a field. It finds the first paper, walks the citations forward, decomposes every argument into rated claims, audits whether each citation actually holds — and hands you the holes as testable hypotheses.
LIVE FROM THE READING ROOM · UPDATES HOURLY · OPEN-ACCESS SOURCES ONLY
[ 01 / THE PROBLEM ]
Paper count and citation count reward volume and popularity — not soundness. The system's own literature says so:
of papers are never read by anyone beyond authors, referees and editors
wasted on irreproducible preclinical research in the US alone
average peer-review wait, on ~15M hours of unpaid reviewer labour a year
a working researcher spends just finding papers — before reading one
[ 02 / THE METHOD ]
Start at the field's first paper. Move forward through citations, in publication order.
Each paper becomes ≤10 claims — premises and conclusions, rated 0–100 for soundness and validity.
Every citation edge audited: does the citing conclusion actually extend the cited claims?
Where the chain fails, the hole becomes a falsifiable claim with the method and stats to settle it.
[ 03 / FIVE INSTRUMENTS · ONE ENGINE ]
[ LINEAGE ]
A chronological spine from the seed paper forward — every argument decomposed, every claim tagged by what carries it: data, argument, citation, or assumption.
[ ONCITE ]
Eleven relationship types, one verdict per edge: grounded, citation black hole, or circular. Citing ≠ supported — now visible.
[ PAPER WEIGHT ]
Fourteen intrinsic dimensions — evidence, rigor, problem gravity, reach — scored like fundamental analysis values a stock, not like a crowd counts applause.
[ GAP FINDER & PAPERSTORIES ]
Gaps ship as falsifiable claims with study design and statistical bar. And any paper reads as a structured story that leaves you carrying its mental model.
[ 04 / STRAIGHT FROM THE WALK ]
“The 1982 journal strata remain stable over time despite the introduction of keyword schemes or other organizing mechanisms.”
found on «The journal communication system for MIS research» (1982) · to settle it: Spearman's rank correlation, effect size (ρ), n=50 journals, α=0.05, power=0.80.
See every gap →[ 05 / START ]
Every field below starts from its verified founding paper — no lucky searches. Want one we don't cover yet? Request it inside; it becomes the next curation.