Scholarly communication · May 28, 2026 · 5 min read

Institutional repositories, twenty years on

Institutional repositories after two decades of open access: why voluntary deposit lagged, what policies and services changed it, and what works today.

Students working at long tables in a university library atrium

In the early 2000s, the institutional repository looked like a turning point. Universities would collect the papers their researchers wrote, make them free to read, and preserve them for the long term. Software such as EPrints and DSpace appeared in 2000 and 2002. The Budapest Open Access Initiative named self-archiving as one of two routes to open access in 2002.

Two decades later, almost every research university runs a repository. Many hold large and valuable collections. Yet the original hope, that researchers would deposit their work on their own, mostly did not come true. This post looks at what happened and what has worked since.

A view from 2004

The questions were already clear early on. On October 2, 2004, the MetaScholar Initiative at Emory University hosted a Symposium on Open Access and Digital Preservation at the Emory Conference Center in Atlanta. The program paired the two topics deliberately. Open access was about who could read scholarship. Preservation was about whether it would still exist in twenty years.

The speakers covered the full range. David Seaman of the Digital Library Federation gave the keynote. Panels addressed the legal and economic side, with Ray English of Oberlin College and SPARC and economist Mark McCabe; institutional partnerships, with Caroline Arms of the Library of Congress and the MetaArchive partners; and technology, with Thomas Robertson of the LOCKSS Program and John Willinsky of the Public Knowledge Project. Open access publishers, including the editors of Southern Spaces, described running journals in practice. David Shulenburger of the University of Kansas spoke as a long-time advocate for reform of scholarly communication.

Reading that program today, the striking thing is how well it framed the issues. Access and preservation still depend on each other. A repository that is open but fragile fails half its job.

Why voluntary deposit lagged

Studies through the 2000s consistently found that when deposit was optional, only a small share of eligible papers reached the repository. The reasons were not mysterious.

  • No direct reward. Promotion and grants rewarded publishing in journals, not depositing in repositories.
  • Uncertainty about rights. Many authors did not know which version of a paper they were allowed to share.
  • Friction. Deposit forms asked for metadata the author had already entered elsewhere.
  • Low visibility. Early repositories were hard to search and rarely surfaced in the tools researchers used.
  • Disciplinary alternatives. In physics and mathematics, arXiv already served the purpose, and scholars had no reason to deposit twice.

Version confusion made things worse. Authors often held only the final publisher PDF, which many publishers did not allow in repositories. The accepted manuscript, which was often permitted, had been lost in an email folder. Asking for it months later rarely worked.

Repositories did not fail. They simply could not change author behavior by existing.

Collections board
Backlog1
Broadsides: transcribe song sheetsMOSC · 0%
In progress2
Voyages: map ship records to schemaTASTD · 64%
Highlander photographs: rights reviewSPT · 40%
Ready for review1
Free Culture papers: assign stable linksevents/2005 · 90%
Preserved1
MetaArchive report: fixity verifiedpdfs · 100%
Figure 1 · interactive illustration

What changed deposit

Policies did more than persuasion. In 2008, the US National Institutes of Health made deposit in PubMed Central mandatory for funded research, and Harvard's Faculty of Arts and Sciences adopted an open access policy that granted the university a license to share faculty articles. Many institutions followed with similar policies. In the US, the 2013 White House memo on public access and the 2022 Nelson memo extended requirements across federal agencies. In Europe, Plan S, launched in 2018, set conditions for many funders.

Policies worked for a specific reason. They changed the default. Instead of asking each author to opt in, they made sharing the expected outcome and gave the institution or funder a clear right to act. Exceptions and waivers remained available, but most work flowed through.

Services did the rest. Where libraries took on the work of depositing, checking rights and harvesting metadata, deposit rose. Where authors were left to do it alone, it stayed low.

ApproachWhat it doesEffect on depositCost to the library
Voluntary self-archivingAuthors deposit when they chooseGenerally lowLow
Institutional open access policyGrants a license and sets an expectationHigher, especially with follow-upModerate, needs outreach
Funder mandateTies deposit to grant complianceHigh for covered researchModerate, compliance support
Mediated depositLibrary staff find, check and deposit papersHigh where staffedHigh, ongoing staff time
Automated harvestingPulls metadata and open versions from publishers and systemsHigh for metadata, mixed for full textModerate, technical setup

Services that matter now

The modern repository is less a filing cabinet and more a service layer. The most useful functions are rarely the deposit form itself.

  • Persistent identifiers. DOIs or handles for every item, and ORCID iDs for authors, make deposited work citable and attributable. Our post on persistent identifiers compares the options.
  • Rights checking. Staff who know publisher policies and can tell an author which version to share remove the biggest barrier.
  • Theses and local publications. Repositories are often the only home for dissertations, working papers, datasets and campus journals.
  • Interoperability. Exposing metadata through OAI-PMH and other standard routes lets aggregators and discovery services find content. See our view on whether OAI-PMH is still worth it.
  • Preservation. Fixity checks, format monitoring and replicated copies, for example through a LOCKSS-based network, protect what was deposited.
Server racks beside a shelf of archival boxes
Photo illustration

What works for small teams

A small repository team cannot do everything. The institutions that get the most from limited staff tend to make the same choices.

They focus on content nobody else holds. Journal articles may also exist in disciplinary or funder repositories. Theses, local reports and digitized special collections do not. They automate metadata wherever possible and spend human time on rights and quality. They put deposit inside existing workflows, such as thesis submission or annual faculty reporting, instead of asking for a separate step. And they treat preservation as part of the service, not a later project.

They also measure the right things. Raw item counts say little. More useful questions are which share of a department's recent output is openly available, how many theses arrive on time with complete metadata, and how often items are found through search engines and aggregators. Usage figures should be read with care, since bots inflate them, but trends over time are still informative.

Finally, they plan for the platform. Repository software changes, hosted services come and go, and migrations are costly. Clean metadata, persistent identifiers and a documented export routine make the next move routine rather than risky. Content that sits behind stable identifiers can move to a new system without breaking a single citation.

Collaboration helps too. Consortia can share hosting, preservation and expertise in rights review. The MetaArchive partners at the 2004 symposium were already describing cooperative preservation among institutions that could not each afford it alone. That model still suits smaller libraries today.

The 2004 symposium put open access and preservation in the same room for a reason. Twenty years on, that pairing still describes the job. A repository earns its place when it makes local scholarship easy to find, easy to cite and still there tomorrow.

Questions

Why did voluntary deposit in repositories stay low?
Authors had little direct reward, were unsure about rights, and faced extra steps. Policies, funder mandates and library-mediated deposit did far more to raise deposit.
Are institutional repositories still needed if arXiv and PubMed Central exist?
Yes. They hold theses, local reports, datasets and special collections that disciplinary repositories do not, and they support institutional policies and preservation.
What should a small repository team prioritize?
Unique local content, rights checking, persistent identifiers, automated metadata, deposit built into existing workflows, and basic preservation with replicated copies.