In an entertaining talk titled “A Punch Card Ate My Program,” computer scientist and open source software developer Walt Mankowski (2018) demonstrates how layers and networks of invention, design decisions, dependencies, contingencies, and historical puzzles spanning a few hundred years can be revealed by a simple – if perplexing – programming bug. The bug in question comes from COBOL, widely considered a legacy programming language, where legacy means “of, relating to, or being a previous or outdated computer system” (Merriam-Webster).¹ Mankowski’s story of the bug, meanwhile, explains how COBOL inherited and carried forward elements of programming’s punch card past, to the point that a punch card could be said to have broken an order processing application written for a home shopping television network in the mid-1990s. As Mankowski concludes, such cases represent how “decisions made at the time that were perfectly reasonable can kind of come back and bite you.”

The problem of legacy systems can be boiled down to “coping with success” (Bennett, 1995), as past successes only become failures in retrospect. Additionally, there are issues of maintenance and scale: as software evolves remedial action must be taken to ensure its integrity while the system and its underlying code have grown in scale and scope. Importantly, these issues should not merely be understood as technical problems, but also – if not especially – as management problems (idem). As Bernhard Rieder suggests in an interview conducted for this special issue, the concept of legacy highlights a wider condition: “Software-making is […] constantly faced with the ‘legacies’ of previous work: since we necessarily build on top of other things, our space of expression is both widened and structured by what came before.”

At another fundamental level, as the etymology of “legacy” suggests, legacy systems may be understood as material-semiotic problems of appointment and designation, sending and receiving – problems not unlike the broken order processing program that Mankowski was tasked with fixing.² What counts as the source of the bug? Where, how, and when will it surface? Will the crystal windchimes I ordered arrive on schedule?

The computational use of legacy, like other seemingly straightforward terms in computer science and related fields, is thus messier than it may at first seem (as hinted at by the phrase “of, relating to, or being” in its dictionary definition). Even within specific organizations or contexts, there is controversy over what counts as legacy or outdated (Plennert, 2018). However, as Rieder (2020) notes in relation to the fundamental-yet-slippery concept of information, this may also be understood as a productive ambiguity: the term’s utility does not derive from uniformity of application, but rather from a semiotic surplus that allows for varied-yet-limited use in defining a problem space or making sense of a given situation. For media archeologists and Internet historians, the problem space designated by legacy systems is one of a range of tensions that are continually present in our historiography: questions of materiality and immateriality, agency and structure, competing determinisms, collapsed temporalities, and so on. It suggests path dependence, where constellations of material objects and historically located decisions can form barriers that limit future action, but it also recalls the Derridian trace, which Spivak defined as “the mark of the absence of a presence, an always-already absent present” (1997). A legacy system points to, well, a legacy; however, as a manifestation of historical contingency it also points us in the direction of past futures that were never realised.

Contributions to this special issue revolve around these themes of the past’s persistence as well as its felt absence. Each takes the productive ambiguity of legacy systems in different directions, from articles that deal with cultural and political legacies relevant to histories of networks and digital culture, to others that investigate how specific technologies bear the markers of past debates and decisions.

The issue begins with Megan Finn and Quinn Dupont’s exploration of the legacies of Computing Professionals for Social Responsibility (CPSR), an organization founded in 1981 to address limited technological understanding in public debate around military applications of computing. Over time, CPSR prioritised civil liberties, reflecting what the authors call a shift from “closed world” discourse (Edwards, 1997) to “digital utopianism” (Turner, 2006). This change in direction highlights a relative absence within discussions of the ethics of technology today: the early CPSR sought to expose the limits of information technology and thus raise important questions about its reliability, but such a position seems lacking in broad discussions about algorithmic bias, efforts to combat fake news, and other debates. This account of CPSR’s legacy thus may offer inspiration for upgrading how we approach the ethics of technology.

Niels Kerssens draws on the concept of legacy systems to reframe a “lost network event” as a marker of continuity, and thus to complicate typical narratives of “winners” and “losers” in histories of networks (c.f. Russell, 2017). Kerssens delves into the history of Euronet, the Europe-wide internetwork that might be remembered only as another “victim” of the success of TCP/IP. In Kerssens’s reframing, this “lost network event” becomes one manifestation of a larger socio-technical vision of European cooperation. Euronet can be easily forgotten or deemed a failure in comparison to heroic narratives of the rise of ARPAnet and Internet protocols, however this measure of success misses out on how Euronet’s purpose was sustained and pursued elsewhere.

Polina Kolozaridi and Dmitry Muravyov offer another history of a lost network event, that of Tomsk’s Tonet. The authors unravel the distinction between local nets and the global Internet. They do so through by analyzing how stories of the inevitability of a global Internet featured in the development and adoption of a local network. As such, they connect the recent call for locating “missing net histories” (Driscoll & Paloques-Berges, 2017) with the ongoing task of analyzing the construction and significance of broader technological imaginaries (Flichy, 2007).

We turn our attention to the World Wide Web in Rudolf Amman’s history of the “misremediation” of academic text standards in HTML. Although HTML and the web more generally are understood to build on a legacy of hypertext as envisioned by Ted Nelson and others, Ammann contends that this perception overshadows an important lineage of academic textual production. His account serves as a reminder that decisions and debates that shaped the web were not limited to idealists versus commercial actors or other simple oppositions, instead revealing how specific professional and technological backgrounds of diverse participants informed a relatively ad-hoc development of HTML.

Rounding out the original articles in this special issue is Meg Leta Jones’s history of the HTTP cookie. As Jones notes, the cookie forms a strange case study in the web’s history, given its existence at both the back-end and interface of the web, both as infrastructure and lightning rod. Combining a history of the cookie’s original design with one of the regulatory frameworks and commercial activities brought to bear on it, Jones reveals the cookie’s “legacy of controversy” as both subject and object within various discourses around online privacy, commercialization and control.

We conclude with an interview with Bernhard Rieder, whose forthcoming book Engines of Order: A Mechanology of Algorithmic Techniques (2020) provides a strong theoretical contribution to software studies as well as a historical analysis of dominant technologies and techniques within current practices of information ordering. As Rieder explains in the interview, software-making is thoroughly shaped, but also expanded in scope and scale, by past legacies. This dynamic highlights another ambivalence of legacy systems as explored throughout the special issue.

Taken together, the various contributions to this issue provide evidence of a rich terrain for engaging with seemingly failed, abandoned and forgotten objects, discourses, and practices in Internet histories. They reveal that doing so requires attention to symbolic and material legacies and lineages, and remind us that these histories must be appreciated to fully understand the present – including what is missing from it.

Disclosure statement

No potential conflict of interest was reported by the authors.