Fortress Libraries or Captured Ecosystems?
Open Access and the Democratization Paradox
Infophilia: A Positive Psychology of Information | January 31, 2026, Vol. 4, Issue 8
Intellectual Freedom | Civic Infophilia | Technophilia | Toolbox | Wellbeing
✨Welcome to Infophilia, a weekly letter exploring how our love of information and connections can help us all thrive, individually and collectively. Avant-garde research.
📌 Access & Attribution: This serial offers free previews always and occasional open access essays to keep scholarship accessible. Students and those facing financial barriers can request complimentary access. If you find value here, please cite the original work and consider supporting the public scholarship through subscription. Proper attribution sustains this work and models healthy information engagement.
What happened to openness? That’s the big question I explore today.
📚 Book Club/Office Hours: Join me Feb. 4 Wednesday 12 noon Pacific Time for a live discussion of this essay. We will also resume reading The Privatization of Everything by Donald Cohen and Allen Mikaelian. Free for all readers. Contact me for the link.
Cite as: Coleman, A. S. (2026). Fortress libraries or captured ecosystems? Open access and the democratization paradox. Infophilia: A Positive Psychology of Information, 4(8).
Fortress Libraries or Captured Ecosystems?
Open Access and the Democratization Paradox
The $1.5 Billion Question
On January 1, 2026, thousands of works entered the public domain—books, films, music published in 1930 finally free for all to use. In August 2025, Anthropic agreed to pay $1.5 billion to settle claims it had trained Claude on pirated copies of roughly 500,000 copyrighted books. And 29 January 2026 was the last day for authors to opt-out of accepting that payment. The irony continues.
Let’s do the math: approximately $3,000 per title (authors get less post-fees, see Notes).
$3,000 is the price of what books become when transformed into training data—when circulation becomes extraction, when sharing becomes capture.***
The settlement raises questions libraries have grappled with for centuries: What does it mean to make knowledge freely available in a world where some actors can extract value at scale while others simply provide the labor?
The question has taken an unexpected turn. Recently, major publishers (Hatchette, Penguin Random House, New York Times, Financial Times) began limiting the Internet Archive’s access to their content. They feared AI companies could use the Wayback Machine as a training-data shortcut. Librarian and lawyer Kyle Courtney, Harvard’s Director of Copyright and Information Policy calls this “the oldest story in the library book”: when licensing shows up, access quietly leaves the building.*****
Libraries are now blamed for AI scraping they didn’t authorize and can’t control. Publishers respond by blocking archives. AI companies profit regardless. The knowledge infrastructure has become collateral damage.
The answer, I believe, lies in a fire from 1764.
What the Harvard Fire Taught Us
On January 25, 1764, lightning (unconfirmed by primary sources) struck Harvard Hall during winter break. The library burned. Maps, scientific instruments, philosophical apparatus, even John Harvard’s original 400-book donation were reduced to ash.
Only 404 books survived.
They survived not because they were protected, but because they were missing (stored/unpacked). About 144 had been checked out by students—many overdue, some flagrantly so. Among them was The Christian Warfare, the library’s only theology text from John Harvard’s original donation, borrowed and not returned by a senior named Ephraim Briggs. His delinquency preserved it. After the fire, Briggs quietly returned the book. No punishment followed; there was little left to punish him for.
Knowledge survived not inside the institution, but in circulation.
Harvard drew the wrong lesson. The library was rebuilt with tighter controls—limited borrowing, restricted hours, safer and more closed. Access was treated as a risk to be managed, not a condition of survival.
That tension never disappeared. It simply migrated from locked doors and ledgers to licenses, platforms, and training datasets. The question facing us today: who benefits when knowledge circulates—and who bears the cost?
The Library of Congress offers an elegant case study.
The Democratization Paradox: By the People and Mass Persuasion
Participation can be widely distributed while the ability to extract value remains concentrated.
The Library of Congress runs two public projects that shine a light on this tension. By the People invites volunteers to transcribe and tag digitized manuscripts, turning fragile documents into searchable datasets. Mass Persuasion Campaigns curates historical propaganda to teach how information shapes belief and behavior.
Each serves a different purpose. Together, they reveal a structural paradox.
By the People is genuinely progressive. Volunteers contribute skilled labor. Collections once locked behind institutional scarcity become searchable and usable. Participation is democratized. Knowledge work is validated beyond academia.
But participation is not the same as power.
The datasets created through volunteer labor are released openly, often with explicit encouragement for computational reuse—including AI training. Downstream benefits accrue to actors with capital, infrastructure, and scale. Not the volunteers. Not the communities whose histories are being transcribed.
Now apply the Anthropic settlement math: if a commercially published book is worth $3,000 as training data, what’s a transcribed diary worth? A letter collection? A run of historical newspapers?
The volunteers aren’t paid. The AI companies profit.
Mass Persuasion Campaigns supplies the critical lens. It trains readers to ask: who can afford to persuade at scale? Whose interests are served? What remains invisible when rhetoric frames itself as neutral public service?
Applied here, it invites scrutiny of “openness” itself. When access language obscures asymmetries of benefit, persuasion is at work.
Openness resists enclosure at the point of access, but enables extraction at the point of scale.
The paradox isn’t new. Colonial heritage projects used the language of preservation and universal benefit while extracting labor and knowledge for metropolitan gain. What’s changed is the scale. And the speed.
The question becomes: can institutions govern circulation without retreating into fortress libraries?
Librarians champion fair use for AI training (lawful materials only—not shadow libraries). But what's happened to openness when libraries get blamed for scraping? When everyone can contribute but only a few can profit, has openness failed? Or does it demand reciprocity we haven't built?
Two models, one institutional and the other individual, suggest answers.