The Book after the Book after the E-book after the AI-book
Giselle Beiguelman, Nov. 2025
The Book after the Book was updated, in its code and web design, in 2025 in the context of the exhibition Antagonistas at the Museu de Arte Contemporânea da USP (MAC-USP), ten years after the work had been incorporated into the museum’s collection. The rules we gradually established—such as gray-scale gradients on the shelves of the bookcase that fade out the edges, the updating of JavaScripts, and the “Snapshot” and “Context” boxes—function as a kind of new grammar for the site.
As I wrote in Museums of Losses for Clouds of Oblivion (2019), I do not believe in the restoration of websites. Together with Professor Ana Gonçalves Magalhães, responsible for incorporating this net art work into MAC-USP’s collection, we developed a methodology to deal with net art pieces in the museological universe that starts from the premise that preserving net art presupposes the impossibility of recovering it.
Sites like The Book after the Book (1999) require such intense rounds of updates and code reprogramming that we decided to assume that future net art museums will be museums of the unfinished, unrepaired and unrecovered. This strategy may allow us to deal with irreversible losses without relying on the subsequent process of disappearance of the artworks.
In this sense, the current version does not erase its original paleoweb incarnation, which remains in MAC-USP’s collection, but rewrites it with new historical markers: in 1999, the emphasis was on breaking the metaphor of the book in relation to the web; in 2025, I realized that we also need to break nostalgia itself, reformatting everything so it can survive on smartphones, meet accessibility standards, and operate within much stricter online security and surveillance environments.
Responsiveness is perhaps one of the most emblematic changes. In the original version, each page assumed a relatively fixed screen: 800×600, a certain text density, the idea of the “window” as a stable frame. By making the site responsive, the shelves now fold into the phone: columns become vertical blocks, grids collapse, menus reorganize themselves into two lines. I say we because it would have been impossible to carry out this recompilation work without the help of ChatGPT.
It is true that AI had several difficulties understanding my recoding proposals, which sought to preserve the atmosphere of refusal of the canonical printed-page format, of navigation guided by standardized menus on every page, and of the intentional disorder of lines and typefaces. Even so, we arrived at some agreements, taking into account the diversity of screens through which people access the internet today.
On the one hand, this is an obvious democratic gain: the reader can enter the bookshelf from their phone, on the subway, on the street, without needing a vintage desktop or a Netscape emulator. On the other hand, responsiveness destabilizes formal decisions from the turn of the 2000s: spacings conceived as an “architecture of the gaze” are now recalculated by layout algorithms, and the gesture of scrolling the page partially replaces the gesture of “clicking on the interval.”
In the original version, these intervals were pages where a script created a fade that prevented navigation back and forth between pages using the browser’s own controls. That was what broke the metaphor of the browser as a book and suggested an environment in which one actually browses.
The information-curation system, already intrinsic to the shelves of the bookshelf that structures the site, has been preserved. However, each shelf now comes with a synopsis and an archival note, as if they were labels in an exhibition. This puts pressure on the original intention, since the first project presupposed getting lost in Borges’s sandy “intervals” of the links—even though I have kept my menus quite enigmatic and intentionally unstandardized in how they appear.
The links in my “revolving-shelves” have not been updated in the sense of incorporating new entries, but they have been reviewed, and dead links were removed while keeping their original entries. In this sense, the bookcase, which once operated only as a heterotopia, also begins to function as an archaeology of the web, with a map of what has survived over these 25 years and an indication of what no longer exists.
However, it is impossible to update the without running into cybersecurity as a new regime of soft censorship. Redirect scripts that worked in 1999 are now blocked or considered suspicious; pop-up windows, nested iframes and certain embedding practices have simply ceased to be viable. The web that became responsive and “user friendly” is also a web of HTTPS certificates, content security policies and blockers for anything that seems intrusive.
Nevertheless some important pages resisted all the transformations. They are the index and the iconic p.0.
This directly affects creativity and design. Solutions that depended on glitch`es, bugs or “violences” of the interface (windows that multiply, back-and-forward loops that trap the user) have been domesticated by browsers like Chrome to protect users. In a world of phishing and surveillance, security acts as an aesthetic and annoying filter, sanitizing the risks that many net.art works from the 1990s explored as raw material.
By including, for instance, oldweb.today (the only new link in this site’s vast collection) as a gateway to an internet “powered by Netscape,” I introduced a kind of controlled time machine. After all, oldweb.today emulates the past in a sandbox, under the rules of the current web. It is an ingenious and symptomatic solution: the Paleoweb can exist today only as an encapsulated exception, surrounded by layers of contemporary infrastructure that keep it harmless, like the apps that trap us in bubbles.
This technical mediation reorganizes the pact with the reader. They know they are navigating an archive, no longer the risk zone of the 1990s internet.
Taken together, the modifications we implemented behave like a palimpsest over the original writing done in the language of pre-Google, pre-smartphone browsers, with CSS style models that were often a bit crazy and metadata that were very carefully organized.
Carrying out this review was also a trip through time. I was quite impressed by the vitality of sites I loved and have come to appreciate even more—such as those of the duo JODI, still working impeccably since the 1990s—and pleased with the many references I made to Mark Amerika, whose works I continue to regard as key references.
Finally, between the two versions, an important critical space opens up. Revisiting the Paleoweb of The Book after the Book from the present means realizing that it was not only the device of the book that changed, but the very field of forces that defines what is “possible” or “acceptable” to do with language, image and code on the network.