rereading Project Blog

For almost twenty year, the Calibre project has provided people with a way to manage their e-books, building personal libraries, transferring them on and off e-readers, converting books between format, and more. From the perspective of a reader, it's incredibly freeing to be able to build my own personal library, and to know that my books are safe from censorship, DRM, and the kind of retrospective erasure that has only gotten more threatening over the years.

I can hardly emphasize that enough — we are currently living through an era where global fascism is on the rise, has taken deadly root in the United States, and is poised to erase queer identity and lives of color from the Internet and from literature, all under the auspices of getting rid of “woke” and “DEI.” Technology created and enabled this new form of fascism, it's only fitting that open-source technology provides us with a way of resisting, of reading what authoritarians would try to erase.

That relationship, Calibre as a gift to readers, Calibre as a celebration of literature, Calibre as something that can resist fascism, that relationship changed with the announcement of version 8.16.2, including new “AI” features that “allow asking AI questions about any book in your Calibre library.” I've said my piece about AI and fascism, and I continue to believe that AI is primarily funded by and profits fascists, that AI in its current form is based on eugenicist thought, and that the energy consumption required for AI presents an untenable threat to the environment, but none of those truly gets at the heart of why Calibre version 8.16.2 is such a deep betrayal.

Indeed, I don't think one needs to agree with me or anyone else about AI as relates to fascism in order to believe that AI has no place in the reading and writing of books. A book is, first and foremost, a work of art, whether it's a textbook, a novel, a comic book, a cookbook, a zine, a collection of delightfully naughty limericks, a collection of architectural photographs, or any other kind of book. There is human expression that goes into not just the selection of information, or the organization of that information into language, but in the individual word choices, the emergent poetry of mathematical form, the clever bit of voice in a technical manual, and the fun turns of phrase in a novella. AI, whatever one believes it is, is not human, has none of these creative drives and flourishes. When AI consumes and digests language, what it produces is not language, but some pale thing that is merely language-shaped. To consume a book, to reduce it to something merely book-shaped, is a form of technological iconoclasm.

Perhaps worse still, the desire to mechanically reduce a book down to its most basic summaries is a fundamental disrespect to the artistic form. Even should the output produced by an AI chatbot take the form of actual language, the choice to create that chatbot and loose it on artistic expression is to fundamentally fail to understand the core fact that makes a book art, and that makes art a form of labor: the art is not the contents of a book, something that can be extracted neatly from the book's structure. The book is the art, in its whole and its entirety. The desire to reduce a book, to extract from it some essence and discard its inconvenient summation, is similarly iconoclastic.

There are secondary concerns that cannot and should not be ignored, such as that AI models cannot be ethically trained by any technique currently known to human kind — they require vast, nearly unfathomable seas of uncompensated labor, stolen labor. But all these concerns are purely secondary to the mere and plain fact that books are art, and AI is not art.

Software that promises to help you preserve and read your books, but then presents you with an option to reduce those books down to 𝑛-grams and Markov chains is thus a betrayal.

This is perhaps too harsh, though, towards the immense gift that Kovid Goyal gave to the literary world with his twenty years of work on Calibre. Perhaps. After all, I am not Kovid's customer, and he does not owe me any of his time on this Earth. That said, I liken the situation rather loosely to someone who volunteers at a local library. We have no right to expect that they continue to volunteer; they can quit at any time and go along their way. But we do have a right to expect that they won't choose to deface the books under their care. Kovid's gift, immense as it is, does not justify Calibre's current betrayal.

Still, though, I am perhaps too harsh here as well. Surely we, readers, writers, publishers, and everyone else who celebrates and adores books, should not have let this situation come to pass in the first place. Who are we to have let the future of literature rest on how Kovid chooses to spend his life? Couldn't we have chosen a different path long ago, invested as a community in a future where books can flourish outside of DRM's brutal walls, rather than entrusting our literary future to the fluke that Kovid Goyal chose to be so generous with his time?

All of the above is true at once, I posit: Kovid betrayed us with Calibre 8.16.2, and we as a community never should have let that happen, never have placed such a burden on individual generosity. The synthesis of these two truths, then, is nearly self-evident. There must be a new effort, free from AI encumbrances, that is built from the ground-up as a community effort. Something that can outlive the decisions of any one participant, something that does not place any one individual under an untenable load.

The rereading Project aims to do precisely this. Though it started as an attempt to create a long-term archival fork of Calibre, stripped of its AI antifeatures, the rereading Project has already grown over the past week into an effort to create a community with strong and fair governance and stewardship. The initial draft charter for the rereading Project explicitly does not include any form of “benevolent dictator for life,” or BDFL, but puts forward consensus-driven committees and community responsibility. It is my sincere hope that rereading will outlast me, and will surpass me when I falter.

For now, the focus of the rereading Project is three-fold: (a) build a sustainable and AI-free replacement for Calibre, (b) build the foundation for long-term governance needed to ensure sustainability, and © build infrastructure for the Project that does not depend on walled-off and gated subdivisions. Arcalibre, our archival fork of Calibre, is the first step towards addressing the first goal. Our draft charter and investigations into sustainable collaboration platforms are the first steps towards the second goal. This blog, the rereading Community Forums, and the rereading Project Mastodon instance — each accessible via RSS and ActivityPub — are all first steps towards the third goal. We're far from achieving any of these goals, but these first steps are important.

I invite you to join us on this journey, to help build something that respects readers, authors, editors, independent publishers, and all artists who find their home in the medium of electronic books. It's a long road, so let's walk it together.