The EPC

For a brief moment, an eclectic little database of poetry was the largest website on what we now generically call the internet. This site, the Electronic Poetry Center (known generally as the EPC), was founded in 1994 by Loss Pequeño Glazier in collaboration with Charles Bernstein as part of the Poetics Program at the University of Buffalo. Its outsized role in the early internet was a product of the automated population of the EPC’s nascent WWW address with the bibliographic records, electronic correspondences, and archival documents Glazier had been compiling on Gopher, FTP, and other alternate protocols in the years before the widespread adoption of HTTP for internet traffic.1 Hosting an exceptionally heterogeneous array of formats, the EPC retains a formidable repository of the earliest intersections of poetry, poetics, and digital networks.

Now, when I prompt Google to search for “EPC,” it is the forty-fourth hit in my results, well below the top hits for other EPCs: “Evangelical Presbyterian Church,” “Efficient Power Conversion Corporation,” and the Wikipedia page for “Engineering, procurement, and construction.” This is the product of several removes: first, Google’s preference for advertising entities and encyclopedic information; second, the hand-coded and thus SEO-resistant nature of the site; and third, its 2019 relocation from long-time University of Buffalo servers to the University of Pennsylvania, where it reunited with Bernstein, but lost the decades of algorithmic clout it had gained at its previous URL. Today, this tremendous repository of primary documents on poetry and poetics just barely registers above an informational page on the Dewey Decimal Classification Editorial Policy Committee (EPC).

Beyond PageRank, the site is in ruins. Broken links, abandoned projects, deprecated tags, and missing images join with unplayable media, neglected collections, and obsolete formats. The site coaches its users through variable tactics for access: “If you have a broken link to an EPC resource, use the top tabs (e.g. author pages & digital library, etc.), fill in the custom search bar at the foot of this page or, for a known URL, substitute ‘writing.upenn.edu/epc/’ for ‘epc.buffalo.edu/’ in the address bar above.”2 However, its current state of disrepair also dials up an earlier internet, within which the EPC emerged. In his groundbreaking book Digital Poetics: The Making of E-Poetries, Glazier outlines the EPC as a kind of “subject village,” clustered around poetry on the web. Writing in 2001, he notes that this type of site “not only delivers texts but also offers slow connect times, error messages, misgivings, and the megabytes of misinformation that typify a largely unedited textual space.”3 Thirty years after the site was initiated, I find my 5G wireless navigation of its dilapidated links eerily anticipated by conversations on the unique challenges of access, information overload, and interoperability of digital collections that remain on the site. These conversations play out across the pages of R/IFT magazine, the Buffalo Poetics listserv, and the stranger offerings found amongst the Error-404 dead ends that delineate the EPC.

Reading through the continuity of these errors presents a kind of time travel, in which we might deploy the Wayback Machine as a portal into the variantology of the site as a fluid text. Not only are the Internet Archive captures more complete and functional than the currently hosted site, they also track the developments—and abandonments—of the EPC over time. Glimpsing these snapshots drawn from 1997 to 2019, alternative visions of what an “electronic poetry center” might afford can be found within abortive projects and imaginative productions of a site materializing alongside the emergence of Web 1.0. These visions were modeled in response to the all-too-familiar terrain of “the megabytes of misinformation that typify a largely unedited textual space.” They presciently map onto today’s internet, marked by manifold crises on zettabyte scales, driven by the corporate black-box algorithms that Matthew Kirschenbaum has heralded as the agents of a coming “textpocalypse.”4 But they also enact speculative hopes for smaller communities of care, rich poetic dialogue, and experimental flourishing, even within the breakdowns of what Bernstein elegiacally deemed “electronic pies in the poetry skies” upon the dissolution of the Poetics Listserv in 2001.5

In this way, just as the EPC points back to the persistence of error and misinformation from its inception, it simultaneously gestures to better futures for the internet before the mores of always-on social media practices took root. Within these glitches, nested deep in lost file hierarchies, an obsolete imaginary of the network continues to subsist alongside mainstream practices. Here, the little database presents an opportunity to reimagine the internet a little more strangely, where future developments might rediscover older forms of invention made long before the establishment of today’s generic conventions. Unpacking the EPC library offers a moment to speculate on how and why we’ve come to use certain dominant forms and formats over others.

Like the illuminative occlusions of server transfers at Eclipse, the EPC reveals a range of otherwise unseen features in the moment of its breakdown. Given the current disrepair of site links, the only way to fully explore the EPC is to download it. Even as it has displaced ordinary practices of downloading, the streaming internet produces a novel perspective through which to interrogate what it means to gather a collection in the file hierarchies of personal computers. In a reversal of what Peter Lunenfield termed “the secret war between downloading and uploading” in 2011, today’s networks are defined by the continuous upload of user data—if not by choice, then via dataveillance—just as downloading repertoires are increasingly rarified.6 No longer a default option for accessing cultural objects, downloads become necessary only in exceptional circumstances of connectivity, practical use, or personal preservation. These circumstances may include teaching materials in a classroom with spotty WiFi access (to ensure live playback); looking to more directly interact with media (to annotate a PDF or sample an MP3); or hoping to preserve a meaningful image from the ongoing streams of content (the image of a meme or a lover). The broken relocation of the EPC, like a Steam game with corrupted data, requires a full download to properly reparse its content. Using a popular tool like Wget or WebCopy can deliver the full file structure in just over an hour, scraping the entire site’s contents of just under 5 GB of material.7 From there, the site reveals its secrets within file hierarchies not unlike those that structure my research for this manuscript, within which it is now nested. Not unlike, I would dare to presume, the file hierarchies that structure the reader’s own little database adjacent to the reading of this book, which I hope generates new downloads from within its sources, citations, and pages.

The most salient secret this download reveals is an innate relatability to its structure. More than anything else, the EPC reminds me of my backup hard drive, a total wreck that never seems to have what I’m looking for, if I manage to get it to boot at all, but also opens a window into my daily repertoires of use from what now seems to be distant personal eras: folders indexing that year abroad, or segmenting grad school collections across two drives, or recording details from the recently forgotten past of two summers ago. Organizational folders that mean well are disregarded for a smattering of files that were either hastily or accidentally placed out of order (behind the scenes in the EPC, an obsolete “hotlist.html” sits alongside the operant “index.html”). Some project folders are neater than others, the full RIF/T magazine archive (1993–1998), coedited by Glazier and Ken Sherwood, is in pristine condition, situated alongside a folder for “ezines,” preserving an essential archive of the earliest experiments with poetry periodicals on the internet.8 The complete records of the Poetics List are grouped in a single folder through date-stamped .txt files.9 Vast portions of the site are devoted to its newest addition: a “mirror” of Tom Raworth’s complete personal website (1997–2017), saved two years after Raworth had passed and the site went dark, recovered through Internet Archive captures. The only addition to Raworth’s site is a poignant note left by its preservationist, Steve McLaughlin, in a splash page reading: “Tom Raworth thought of this site as ephemeral and that it might disappear when he did.”10 Clicking through the link reveals the full site as though it had never disappeared. The EPC’s longevity is no less a surprise, and perhaps a fuller recovery will yet emerge.

These folders reveal a structural intimacy among the organizational folders of the little databases. The EPC, like Textz, Eclipse, PennSound, and UbuWeb (like every ordinary hard drive), is the product of idiosyncratic nested hierarchies of folders: (mis)named and (dis)ordered according to repertoires of collection, preservation, and transmission made by editors over time. None rely on metadata schemas for archival organization or Content Management Systems (CMS) for content design or embedded algorithmic social features for user engagement or large-scale computational tools for cultural analytics. All, instead, mirror the same practices that preserve digital objects throughout the little databases that make up every personal computer. If every chapter of this study has focused on a single format: the EPC is defined by thousands of HTML files, each written by hand from the very moment the protocol was released. Housed in the nested hierarchies of the EPC, across these HTML pages, a more intimate internet emerges. One built on the quirks of care and glitches of maintenance inscribed when the internet was largely “under construction.”

Longing for the complications of these simpler forms, the hand-coded HTML files of the EPC read like love letters hastily scrawled in Word documents from decades past. Error-prone and excessive, playful and sincere, they encode desires inscribed for uncertain futures. They enfold digital objects in unstable media formats to record meaningful histories for unknown potential uses, guarding against the oblivion of erasure. Hailing from the distant digital history of just a decade or two past, these persistent networks of digital objects continue to tell new stories, subject to ongoing transformations not unlike scratches and dust accumulating on film running through a projector. A contingent media poetics of the little database continues the correspondence initiated by these letters. A way to attend with care, and play, to cultural memories subject to ongoing and unknowable transformations ahead.