Coverage vs. Depth argues that scholarly production can optimize different goods.
The depth-architecture typically offers:
The coverage-architecture typically offers:
The paper does not claim that one architecture is superior or that depth is unimportant. It argues that applying monograph standards as the only standards produces an evaluative-frame error.
Six alternative evaluation categories are proposed:
1. internal consistency across the corpus; 2. cross-node predictive traction; 3. explanatory parsimony; 4. combinatorial product quality; 5. provenance integrity; 6. self-corrective capacity.
The paper situates coverage scholarship beside encyclopedic, archaeological, network-mapping, distant-reading, and longue-durรฉe traditions. AI assistance is treated as an accelerating technique, not as the definition of the architecture.
Its empirical appendix is based on a full Zenodo-community scan performed on May 23, 2026. Title and keyword regular expressions assigned 705 records to 28 domains. The paper openly notes possible false positives and false negatives and says the classification lacks inter-rater validation.
The central product is the map: relations and patterns across nodes that no single monograph can produce at the same scope and speed.