The Unread Shelf: American Doctoral Research's Long Detour Around Peer Review
Photo: Oskar Tauber, CC BY 4.0, via Wikimedia Commons
The dissertation is the longest document most scientists will ever write, and by a comfortable margin the least read. Each year American universities award tens of thousands of research doctorates — the National Science Foundation's Survey of Earned Doctorates has counted roughly 55,000 in recent cycles — and nearly every one of those candidates deposits a bound or digital manuscript. It is catalogued, assigned a persistent identifier, and then, in most cases, left alone.
An Archive That Behaves Like an Attic
Institutional repositories were built in the late 1990s and 2000s on a defensible premise: a university's intellectual output should be open, permanent, and locally held rather than licensed back from a commercial publisher. The premise held. What was never built alongside it was a readership. Repository records are inconsistently indexed by the major scholarly search tools, described with metadata that varies from campus to campus, and stored in formats that resist text mining. A dissertation on kinase signaling and a dissertation on municipal bond markets occupy the same undifferentiated shelf.
The consequence is a body of scholarship that is technically public and practically invisible. It is not peer reviewed in any formal sense — committee approval being a different instrument from external review — yet it carries the authority of a doctoral credential. It is citable but rarely cited. It is, in the vocabulary of information science, gray literature: produced outside commercial channels, difficult to discover, and largely excluded from the apparatus that synthesizes evidence.
The Duplication Question
If dissertations were merely redundant with the published record, the neglect would be harmless. They are not, or not entirely. Doctoral work typically reaches the literature in fragments — a methods paper here, a results paper there, a review chapter later rewritten as a book. What survives into journals is filtered by editorial preference, and what editors tend to prefer is the confirmatory finding rather than the abandoned arm of the study.
Scholars who have sampled individual fields report that anywhere from a quarter to two-thirds of dissertation content eventually appears in some published form, with wide variation by discipline. Chemistry and molecular biology skew high; education, public policy, and parts of the humanities skew considerably lower. The variation matters because it correlates with the fields in which systematic reviews and meta-analyses are thinnest. In those areas, the gray literature is not a supplement to the evidence base. It is a substantial part of it, and reviewers are trained to exclude it on quality grounds without ever having assessed it. Standard systematic review protocols, including the widely used PRISMA framework, generally treat dissertations as a secondary search tier at best.
What the Shelf Actually Holds
A candid inventory would include three kinds of material the published record does not capture well.
The first is negative and null results. A doctoral candidate who spends four years on a hypothesis that does not hold still has to write it up; a journal editor has no equivalent obligation. The dissertation, that is to say, is one of the few venues in American science where a well-executed failure is recorded in full rather than compressed into a sentence in someone else's methods section.
The second is methodological detail. Journal word limits compress procedures into supplements that authors frequently deprioritize. The dissertation chapter often remains the only complete description of an instrument, a sampling frame, or a coding scheme that later work depends on.
The third is local and applied research — school district evaluations, regional water quality surveys, county-level public health assessments — conducted with access that no external team could obtain and of interest to practitioners who do not read journals at all.
The Case Against Rescue
Any proposal to formalize access to dissertations runs immediately into a reasonable objection: the median dissertation is not very good. It is a training document produced under supervision, on a timeline set by a committee rather than an editor, and it is frequently too long, too cautious, and too poorly organized for a reader outside the field. Peer review exists partly to correct exactly these problems, and a system that indexed dissertations as equivalent to reviewed articles would degrade the literature rather than enrich it.
A second objection is labor. If the roughly 55,000 annual deposits were routed to reviewers at anything like journal rates, the reviewing workforce — already strained — would collapse under the additional load. The arithmetic does not support universal treatment.
What Triage Would Require
A workable model, then, would be selective rather than universal, and it would need three components.
First, better metadata. Repositories could adopt a shared minimum schema — discipline codes, method type, data availability, whether a related article has been published — that would make dissertations discoverable by the search tools researchers actually use. Several consortia have piloted this; none has achieved adoption at scale.
Second, a nomination pathway. Departments, not central administrators, are best positioned to know which dissertations deserve an external audience. A standing departmental right to nominate one or two theses per year for review, with the candidate's consent and the adviser's endorsement, would route scarce reviewer attention toward work with a reason to be read.
Third, a journal-side format. The article-length distillation of a dissertation — shorter, restructured around the central finding, stripped of the literature review that no one needs reproduced — is a distinct genre from the dissertation itself. A handful of journals have experimented with a dedicated submission track for recent graduates, and the results suggest that the bottleneck is not author willingness but editorial bandwidth.
None of this is glamorous infrastructure, which is precisely why it has stalled. Repositories are funded to deposit, not to curate. Journals are funded to review what is submitted to them. The gap between the two is where a decade of American doctoral research currently sits, indexed well enough to be found by accident and not well enough to be found on purpose.