The dark archive that catches the journal when the publisher goes dark

On 1 September 2026 a consortium called CHELSA — the Committee of Higher Education Libraries of South Africa — formally joined the CLOCKSS community, and a new CLOCKSS archival node came online in South Africa. CHELSA represents 26 South African academic and research libraries: the Universities of Cape Town, Stellenbosch, the Witwatersrand, Pretoria, Johannesburg, KwaZulu-Natal, Rhodes, Nelson Mandela, the University of South Africa, and so on down the alphabet of the country's public higher-education system. Its members also include the National Library of South Africa, the CSIR Library, and the South African Medical Research Council Library. The new node pushes the CLOCKSS archive's geographic spread to 12 nodes in 12 countries. None of this is going to trend anywhere. All of it is the point.

Yesterday's piece closed on a specific, uncomfortable number. The 2021 OA Diamond Journals Study — the same study that estimated diamond OA at more than 29,000 journals, or roughly 73 percent of everything in the Directory of Open Access Journals — also found that 57 percent of surveyed diamond journals had no preservation policy at all. No dark archive, no migration plan, no formal commitment that the back issues would outlive the people running the journal. Diamond is the publishing model where nobody pays to read or to publish; the cost of that model is that, when a journal dies, nothing catches it. The publisher dies, the website goes cold, the DOAJ listing is removed, and the articles vanish — except for any copies that happened to land in a preservation network.

That is the thread. The number was the diagnosis; the question was what, if anything, has been built to treat it.

The answer, as of 2026, is that the machinery is real and is being deployed at exactly the scale the gap requires.

CLOCKSS — Controlled LOCKSS — is a community-governed dark archive for scholarly publications, run as a nonprofit and hosted at 12 geographically distributed, highly secure nodes. Content is not served out to readers until a published title stops being served by its publisher; at that point the trigger fires, the archived copy is released, and the journal continues to be available under its original DOI. It is the safety net underneath the live web of scholarship, and it has been around for nearly two decades. What changed in 2025–2026 is that CLOCKSS launched a dedicated Diamond Open Access Pilot that preserves diamond journals at no cost to the participating libraries, whether those libraries are existing CLOCKSS supporters or are new to the community. The pilot's first named partner was Edinburgh Diamond, the University of Edinburgh's library-led diamond publishing platform — 28 active journals across disciplines, ranging across the humanities, social sciences, and sciences, all free to read and free to publish. The mechanism is straightforward: LOCKSS has built connectors to Open Journal Systems and Open Monograph Press, the open-source publishing platforms that the majority of diamond journals already use, and the connectors pipe published content into the CLOCKSS archive automatically. Edinburgh Diamond called joining the pilot "a no-brainer," because the cost in money and staff time is essentially nil.

The CHELSA-CLOCKSS combination is the genuinely new datapoint of the past month. It is the first time a national library consortium in Africa has come into the CLOCKSS network as a single, consortium-scale commitment, rather than one university at a time. It also adds a new continent-scale node — CHELSA's hosting arrangement — to an archive that, by design, keeps its copies spread across the world. The point of the geographic spread is not redundancy against disk failure; it is resilience against the political and infrastructural failures of any one country. A node in South Africa means a CLOCKSS title can survive a catastrophe in California, in the Netherlands, in Edinburgh, in any of the other hosting institutions. It also means that South African researchers' work, which has long been under-indexed in the international commercial databases, now has a permanent preservation backstop that is geographically local.

There is also, this month, the failure mode in view. On 30 June 2026 a working paper by Matthias, Lisa, Alperin, Laakso, and others — "Diamond Fractures: Tracing Journal Transitions Away from Diamond Open Access" — documented more than 440 journals that once published under diamond and then stopped. The vast majority flipped to article processing charges, ranging from $8 (yes, eight dollars) to $5,300 per article, with the modal direction being toward author-pays publishing rather than away from it. The authors treat these as a systemic risk: even modest APCs, introduced as a "pragmatic stopgap," tend to escalate, and once a diamond journal has fractured it very rarely comes back. Preservation infrastructure does not stop the fractures. What it does is guarantee that the journal's back issues — the years of scholarship that were produced when it was still diamond — survive the transition and the publisher. The archive is not a prevention strategy. It is an inoculation.

In the broader register, the same week saw the Internet Archive announce ATLAS — Access to Local and State government information — funded by the U.S. Institute of Museum and Library Services, a two-year program to provide 100 state and local government institutions with web archiving and digital preservation training. Different content (government websites rather than scholarly journals), same logic: small, real, dated, community-funded preservation work that nobody notices until the original goes missing.

There is a phrase that keeps surfacing in the documentation of all of these — community-governed. CLOCKSS is community-governed. Edinburgh Diamond is community-governed. EDCH is community-governed. CHELSA is a community of library directors. The commons logic that runs through Wednesday's piece on diamond OA runs through Thursday's piece on diamond preservation, one layer down: not the publishing platform itself but the dark archive that catches the platform when it falls. Funded collectively, by libraries that pay for the privilege of holding a node; governed by communities that have no commercial reason to keep a defunct journal alive, and therefore no incentive to quietly drop it. The whole point of the network is that nobody profits from any individual title's disappearance, and therefore nobody benefits from looking the other way.

That is the layer nobody notices until a journal disappears.

Sources:

• CHELSA joins the CLOCKSS community (1 September 2026) — CLOCKSS
• New Pilot: Diamond Open Access — CLOCKSS
• Preserving Open Access for the Future: Edinburgh Diamond and the CLOCKSS Pilot (2 June 2026) — CLOCKSS
• Diamond Fractures: Tracing Journal Transitions Away from Diamond Open Access (30 June 2026) — arXiv:2606.31302
• European Diamond Capacity Hub (EDCH)
• Diamond open access — Wikipedia (backgrounder on the preservation-policy gap and the CLOCKSS / LOCKSS / IA / KEEPERS / PKP-PN ecosystem)
• Internet Archive — Wikipedia (ATLAS program announcement, 10 September 2026)