David Relman

Record year1990

summaryIt used molecular phylogenetic information to identify candidate microorganisms that were difficult to culture or previously unknown, extending discovery workflows to objects conventional culture struggled to cover while making the interpretive limits of sequence evidence explicit.

M rationaleOriginal-paper reviews and CDC open full text support the formation of this paradigm and its use in identifying specific unknown microorganisms. M6.1 is assigned for its testable knowledge boundary, extension to difficult-to-culture objects, and methodological reuse, rather than accumulated by pathogen count or clinical consequences.

attribution rationaleRelman 0.22; Falkow 0.12; early coauthors 0.18; prior molecular methods 0.19; clinical pathology sequencing teams 0.12; institutions 0.05; unresolved 0.12.

overlap boundaryThe knowledge and methodological outcome, represented in 1990–1998 by culture-independent molecular identification of difficult-to-culture and previously unrecognized microorganisms; it counts only the identification paradigm and knowledge increment, and does not equate sequence presence with causal pathogenicity.

Evidence gapsAuthor order, experimental design, and independent verification for each specific discovery have not been closed paper by paper. A systematic sample of successes and failures from candidate identification to disease causation has not yet been made.

Record year2000

summaryUsing longitudinal, multi-site community data, it characterized individual differences, temporal change, and incomplete recovery after disturbance, treating the human microbiome more explicitly as a dynamic ecosystem.

M rationaleStanford research summaries and representative papers support actual research deliveries on oral communities, temporal change, and recovery from disturbance. M5.8 is assigned for establishing knowledge of dynamic human microbial ecology and a long-term methodological example, without adding credit for the label “pioneer” or imagined health applications.

attribution rationaleRelman 0.20; Bik/Dethlefsen representative coauthors 0.16; laboratory and coauthors 0.22; prior ecology and microbiome work 0.19; institutions 0.06; unresolved 0.17.

overlap boundaryResearch from 2000–2020 on community structure of the native human microbiome across body sites, individuals, and time, and on stability and recovery after disturbance; it counts only the ecological knowledge outcome, and does not equate correlation with disease cause, diagnosis, or treatment.

Evidence gapsThe specific coauthors and cohort contributions in different representative papers still require paper-by-paper decomposition. Causal strength among community description, disturbance mechanisms, and health associations has not been systematically calibrated. Project-by-project records are lacking for privacy, sample governance, and participants' actual research contributions.

Record year1996

summaryIt organized high-level conditions such as tissue distribution, host response, consistency, reproducibility, and alternative explanations into a causal-review framework for sequence evidence, making clear that a single sequence detection is insufficient to establish causation.

M rationaleThe open review clearly distinguishes sequence detection from causal attribution, and the Fredericks–Relman framework forms an epistemic outcome different from discovery technology. M5.3 is assigned for the method's general review value; its scope is narrower than an actual discovery system and does not duplicate the latter's general detection value.

attribution rationaleRelman 0.28; Fredericks 0.28; prior classical causality work 0.20; research-community experiential input 0.10; institutions 0.04; unresolved 0.10.

overlap boundaryCriteria and epistemic limits for attributing causation from sequence-based microbial evidence in 1996–1998; it counts only a rule framework that moves from “detected” to “conditionally supports causation,” without duplicating the preceding discovery method.

Evidence gapsThe early intellectual sources of each criterion and their subsequent empirical performance have not been traced item by item. Its effect in reducing false attribution across different fields has not yet been systematically compared.