In 1991 the World Wide Web seemed to provide a path to a dazzling future: everyone in the world would be able to communicate, at a minimal cost, with everyone else through the Internet. In 2004 Google promised to make that future even brighter. By digitizing library holdings, Google would create a modern Library of Alexandria: everyone would have free access to all the books in existence. Digitization promised to open up the world of learning to the excluded and the underprivileged, particularly in developing countries. But it touched off an equal and opposite reaction in the form of closed access, paywalls, and monopolies. The world of learning has become a battleground between the opposed forces of democratization and commercialization.

To be sure, the complex realities of technology, publishing, libraries, and authorship cannot be reduced to a Manichaean notion of progressives battling money-grubbers, because most of the participants in the knowledge industries pursue self-interest, whether it is profit or prestige. The recent history of academic books and journals shows how these interests have collided, driven by tectonic shifts in the global order of research and publication.

As soon as it began to digitize libraries, Google became entangled in copyright law, which protects rights to a book for the life of the author plus seventy years—that is, in most cases, for more than a century. A moving wall now keeps all books published since 1927 out of the public domain, but Google has forged ahead, digitizing everything (though without making entire texts available), despite having been sued by rights holders for infringement of copyright.

By agreeing on a settlement with the Authors Guild and the Association of American Publishers, Google converted what was originally a free search service involving only snippets from books into a profit-driven commercial library. In order to subscribe to Google Book Search, research libraries, which had made their holdings available to Google free of charge, would be required to buy back digital versions of their own books at a price set by Google, one that could escalate beyond their ability to pay. In 2011 the United States District Court for the Southern District of New York declared that the settlement violated antitrust law. Its decision prevented the creation of a new kind of monopoly, one that would control access to knowledge in digital form.

The death of Google Book Search nevertheless meant that most books published during the last one hundred years cannot be made freely available to the public except as volumes loaned by libraries one copy at a time. The nonprofit Internet Archive has bravely made digital versions accessible, also with restrictions to single copies (the digitized counterparts to the physical volumes in its collection), but in March it lost a suit to publishers who accused it of violating their copyrights. The Sonny Bono Copyright Term Extension Act of 1998 (known as the Mickey Mouse Protection Act because the copyright on Mickey was about to expire) cemented the extent of copyright at the author’s life plus seventy years. According to Jack Valenti, a champion of Hollywood interests, copyright should exist for eternity minus one day.

Yet Article I, Section 8 of the US Constitution requires that copyright should extend “for limited Times” only, subject to the purpose of promoting “the Progress of Science and useful Arts.” In the first US Copyright Act, in 1790, Congress decreed that “for the encouragement of learning,” the term should last fourteen years, renewable once. That act was modeled on the first copyright law, England’s Statute of Anne (1710), which, also “for the encouragement of learning,” limited copyright to a maximum of twenty-eight years. Today most books cease to sell in large numbers within a year or two. Protection for twenty-eight years would be adequate for all but a handful of authors, and it would leave room for a greater good—the public’s access to knowledge. In deviating from its original purpose, copyright legislation has gone only downhill since the time of the Founding Fathers.

Monopolistic abuses are far worse in the publishing of academic journals. Three publishers—Reed Elsevier (now RELX), Wiley, and Springer—dominate the field, producing 42 percent of scientific journals in the English language. They sell most copies to research libraries, ratcheting up the prices so mercilessly that they cripple acquisitions budgets. The average annual subscription price of a chemistry journal is now $7,014. The current price of a full subscription to The Journal of Comparative Neurology is $40,000—the average cost of five hundred to six hundred academic monographs.

Once a university library has subscribed to a journal, it is hooked, because it cannot cancel the subscription without alienating scientists on the faculty who demand an uninterrupted flow of up-to-date information. Faculty members do the research, contribute the articles, referee submissions by others, and serve on editorial boards, almost all for free, and then expect their library to pay for the journals they have helped produce, no matter how extortionate the price. Publishers claim to add editorial value to the articles, but the published version is often nearly the same as the submitted version, so they usually add little—and to buy them libraries have to subtract sums from their budgets for monographs in fields outside the sciences.


The irrationality of the system has been obvious for decades, yet despite many efforts libraries have failed to break the stranglehold of the publishers. The system seemed to reach a breaking point in 2019, when the University of California implemented a two-year boycott of Elsevier journals on all of its ten campuses. In March 2021, after intense negotiations, UC and Elsevier reached an agreement on a four-year contract. Nearly all of Elsevier’s 2,600 journals were made available for reading by UC faculty and students, and the journals were to receive publishing charges (subsidized up to $1,000 by UC libraries) for articles by UC researchers. Such charges, levied on the author, are common in scientific journals, and they often come to $5,000 or more, which is usually covered for the most part by the original research grant.

The agreement made the articles available as “gold” open access (OA)—that is, anyone could consult them without charge—a gain for the general public; but articles by non-UC authors remained behind a paywall, making the journals hybrid (part OA, part paid for by subscriptions). Far from promoting accessibility, hybrid journals maintain publishers’ subscriptions while providing them with additional income from authors. In the end the UC libraries had to pay as much as they did before the deal, about $13 million a year.

Although the millions of dollars in library budgets have not attracted much attention outside the world of libraries, the stakes are high and the issue is political. Most scientific research is subsidized by the federal government. Shouldn’t the public have access to the results of work supported by public funds? The government committed itself to that principle in February 2013 (and in 2008 for funding by the National Institutes of Health) when it directed federal agencies that issued grants worth more than $100 million to require recipients to make their research freely available from digital repositories. But it permitted an embargo of twelve months, which allowed commercial journals to cream off the demand.

Under President Biden, the White House Office of Science and Technology corrected this deficiency in a new directive issued on August 25, 2022. As of December 31, 2025, all agencies, whatever the extent of their funding, must require immediate open access, and the requirement covers data and metadata as well as the final text. The G7 leaders took a similar stand on May 14, 2023, as did the European Council on May 23. The tide is turning in favor of unrestricted access, but the countervailing forces are so complex that the future remains cloudy.

To make sense of these issues and the vast literature surrounding them, the best place to begin is Athena Unbound by Peter Baldwin, an eminent historian of modern Europe and the welfare state who has made himself an expert on the history and current practices of scholarly publishing. In 2001 Baldwin and his wife, Lisbet Rausing, created the Arcadia Fund to preserve cultural heritage and promote open access. As a trustee of the New York Public Library and member of the advisory board of the Wikimedia Endowment, he has been a participant observer of recent attempts to free knowledge for the benefit of the general public. And in an excellent earlier book, The Copyright Wars (2016), he demonstrated his command of the legal issues, stretching back to 1710.

In Athena Unbound Baldwin takes a hard look at the world of knowledge. Have no illusions, he warns: journal publishers are gouging their customers, scholarly monographs reach a tiny audience, libraries are floundering under budget pressures, academics are pursuing careers rather than truth, and readers are not getting all the information they deserve. He expresses contempt for “cultural nostalgists,” who have a romantic view of creativity and an outdated attachment to the printed codex.

Yet he also argues that recent trends point to a favorable future. The widespread notion that we are being buried under unmanageable, expensive, and frequently false information is misleading, he claims, because the improvement of search engines will one day help us find everything we need, and the triumph of open access will eventually make all scholarly knowledge available for free.

Baldwin supports his assertions about current trends with plenty of evidence. His book contains sixty-eight pages of notes to an amazing variety of sources, and he relies heavily on quantitative arguments, many of them surprising. For example, on book publishing:

The average scholarly monograph these days sells 60 copies….

In the major Anglophone outlets, at best around 3,000 books are reviewed annually, out of 500,000 total published in the US and UK….

Of the 10,000 US books published in 1930, only 174 were still in print in 2001.

On scientific journals:


Elsevier posted robust results for 2020…with profits of over £2 billion on revenues of £7 billion….

Two physics articles from the 1990s, presenting findings from work on particle accelerators, had 406 and 271 coauthors respectively….

China issued 21 scholarly journals in 1970, but over 11,000 by 2019.

And on research libraries:

In a medium-size US university library, only 20% of books are checked out even once.

Major research libraries in North America hold almost a billion physical volumes, but only 59 million distinct titles.

The statistics have a gotcha quality, which jolts the reader into questioning the status quo and considering policies that would change the rules of the game. The greatest change is the cause that Baldwin defends convincingly throughout his book: open access, or practices that would realize the promise of the Internet by making material on it freely available to readers. In fact, OA proposals have generated an enormous literature, which goes back to principles formulated in a manifesto, the Budapest Open Access Initiative of 2002. The declaration recommended two versions of OA: “green” (researchers deposit the final version of an article in a repository, which makes it available to everyone free of charge) and “gold” (journals publish individual articles free of charge to the reader, usually in exchange for hefty publishing fees paid by researchers or their funding sources).

Although the green–gold distinction may seem esoteric, it represents an important fork in the road for university policies. For example, in 2008 the Harvard Faculty of Arts and Sciences voted to take the green route, and their repository now provides readers with a vast amount of research. But journal publishers responded by imposing an embargo, which gave them a temporary monopoly on the material, usually for one year. And because the policy was voluntary, professors often failed to deposit their work. The University of Liège solved that problem by fiat: in deciding on faculty promotions, it refused to consider any articles that were not deposited in its repository. The recent decisions by the White House, which make government grants contingent on immediately accessible deposits, indicate that green is the way forward.

Baldwin, however, seems to favor gold. He concedes that it will not work if not applied globally, but he imagines a future in which journals will give up subscription fees in exchange for payments to publish. A great deal of “flipping” to publishing charges is already occurring. But it often results in hybrid journals, which pocket publishing charges (often paid in part by libraries) for articles they make accessible for free, while continuing to collect subscriptions, also paid for by libraries. Unless all the major scientific journals switch entirely to gold, library budgets will not be able to bear the burden.

In the best of all possible worlds, Baldwin imagines a global bulletin board that would make everything instantly accessible, fusing green and gold. All content, of books as well as articles, would be financed by the producers, stored in the cloud, and downloaded gratis by consumers—or, for readers who favor the printed codex, produced by publishers and sold in the manner of trade books.

The fantasy of a universal library has attracted many dreamers. Baldwin thinks that universal digitization can make the dream come true, because the cloud has infinite storage capacity, even though storage requires servers and financing, and because search engines will inevitably be perfected. Readers need not have doubts about finding what they want in “the new Alexandria in the cloud,” because technology will sort that out. “The perfect search engine will one day provide a universal index,” Baldwin writes. That may sound Panglossian, but Baldwin points to current advances in OA as evidence for optimism.

His argument is most convincing when applied to the publication of scientific articles, where the costs are greatest and the technology most advanced. The Public Library of Science, founded in 2000, publishes a dozen first-rate open access journals. The OA repository arXiv has successfully published more than two million “e-prints” in physics and several other fields, with some preliminary assessment but no peer review. For-profit scientific journals increasingly rely on publishing charges in place of subscriptions, thereby making their content free to readers.

But the prospect for open access looks less promising in the humanities and social sciences. Authors in those fields rarely receive grants that cover publishing charges, and their careers often depend on publishing monographs rather than journal articles. Baldwin makes a strong case that their work, along with that of scientists, should be treated as a public good like clean air and highways. Moreover, he sees vanity and careerism as the driving forces of academic authorship. Academics outside the hard sciences receive salaries, yet feel entitled to royalties on the relatively rare occasions when they publish a book. Baldwin considers such incidental income unjustified because it is earned on company time. The publications of nonscientists should be treated as work for hire, he claims, and should be made virtually free to the public.

That proposition accompanies a general assessment of the scholarly world’s infrastructure, which Baldwin considers antiquated and wasteful. His vision of the future therefore turns into an indictment of the present, and he carries his argument to such extremes that it loses the power to convince.

What, he asks, would be the role of book publishers in a fully digitized environment organized in accordance with open access? After the cost of a book’s first copy is covered, an unlimited supply of subsequent copies, provided directly from an electronic repository, would be virtually free. The publisher’s function as a gatekeeper would cease to exist because there would be no more gates—and therefore an end to “the university world’s thralldom to the prestige hierarchy of the established publishing venues.” Acquisitions editors would be superfluous. Already they do little more than weeding out inferior texts and “reading around on company time,” Baldwin claims. Improved search engines would handle the selection process, taking readers right to the works they want, which would be available on the global bulletin board. Peer review could therefore be eliminated, along with unnecessary apparatuses such as professionally designed layout, indexes, dust jackets, blurbs, and sales catalogs. Most bookstores would disappear; libraries would be reduced to storehouses of old-fashioned volumes; and virtually all cultural intermediaries—book reviewers, literary agents, advertisers—would be eliminated, because their functions would be replaced by the all-powerful search engines putting readers in direct communication with texts in the all-encompassing cloud.

A future in which all books, digitally unified, “become one massive tome,” as Baldwin puts it, sounds like the dystopias of Jorge Luis Borges’s “The Book of Sand” and “The Library of Babel.” In fact, it seems so extreme that Baldwin may dangle it in front of the reader as a provocation, intended merely to expose flaws in the current system. But he fails to consider a main objection to his argument: Is a “global bulletin board” even feasible, and if so, could improved search engines compensate for the elimination of editors, reviewers, librarians, and others who dominate the current process of selection? He does not address the problems of eliminating firewalls in countries like China, nor does he deal with the inadequacies of existing search engines. He barely mentions artificial intelligence, which could flood cyberspace with more misinformation than could ever be overcome by search engines of the future.

Cyberspace may have room for an infinite number of texts, but we must find a way to preserve them. Digitized works are fragile. They are made up of minuscule ones and zeroes that decompose, and they require metadata that is equally vulnerable and that can become obsolete, leaving them lost in the cloud. Present techniques of surveillance and migrating texts from one format to another are expensive and incapable of preserving a universal digital library. Baldwin’s assertion that “everything can be saved” is dubious. My advice to anyone who wants to preserve a digital work is: print it on paper.

Baldwin may have underestimated the staying power of the printed codex. More books are published each year than the year before; publishers are making more money; the number of independent bookstores is increasing; libraries are thriving; and despite the omnipresence of digital devices, readers continue to enjoy holding a volume in their hands and turning pages. The paratextual elements of the printed book—its page design and packaging—continue to be effective, even if only a minority of readers care about its aesthetic aspects, such as the smell and feel of print on paper. Talented professionals continue to produce and disseminate this valuable product, and they provide a public service while making money from it.

Baldwin is correct: the more open access, the better. And he rightly warns us that we live in a real world of vested interests, where the better can be sacrificed to illusions about the best. But he puts too much faith in the self-correcting capacity of “digitality”: “Throw all content against the wall and see what sticks.” I would suggest implementing open access, while covering the wall with shelves and filling them with solid, attractive, and durable books.