I’ve been meaning to read Wendy Hui Kyong Chun for some time now. Updating to Remain the Same is on my to-read list, but I recently ran across a reference to Programmed Visions: Software and Memory in Rogers (2017), which I wrote about previously, and thought I would give it a quick read beforehand.

Programmed Visions is a unique mix of computing history, media studies and philosophy that analyzes the ways in which software has been reified or made into a thing. I’ve begun thinking about using software studies as a framework for researching the construction and operation of web archives, and Chun lays a useful theoretical foundation that could be useful for critiquing the very idea of software, and investigating its performative nature.

Programmed Visions contains a set of historical case studies that it draws on as sites for understanding computing. She looks at early modes of computing involving human computers (ENIAC) which served as a prototype for what she calls “bureaucracies of computing” and the psychology of command and control that is built into the performance of computing. Other case studies involving the Memex, the Mother of All Demos, and John von Neumann’s use of biological models of memory as metaphors for computer memory in the EDVAC are described in great detail, and connected together in quite a compelling way. The book is grounded in history but often has a poetic quality that is difficult to summarize. On the meta level Chun’s use of historical texts is quite thorough and its a nice example of how research can be conducted in this area.

There are two primary things I will take away from Programmed Visions. The first is how software, the very idea of source code, is itself achieved through metaphor, where computing is a metaphor for metaphor itself. Using higher level computer programming languages gives software the appearance of commanding the computer, however the source code is deeply entangled with the hardware itself, the source code is interpreted and compiled by yet more software, which are ultimately reduced to fluctuations in voltages circuitry. The source code and software cannot be extracted from this performance of computing. This separation of software from hardware is an illusion that was achieved in the early days of computing. Any analysis of software must include the computing infrastructures that make the metaphor possible. Chun chooses an interesting passage from Dijkstra (1970) to highlight the role that source code plays:

In the remaining part of this section I shall restrict myself to programs written for a sequential machine and I shall explore some of the consequences of our duty to use our understanding of a program to make assertions about the ensuing computations. It is my (unproven) claim that the ease and reliability with which we can do this depends critically upon the simplicity of the relation between the two, in particular upon the nature of sequencing control. In vague terms we may state the desirability that the structure of the program text reflects the structure of the computation. Or, in other terms, “What can we do to shorten the conceptual gap between the static program text (spread out in”text space”) and the corresponding computations (evolving in time)? (p. 21)

Here Dijkstra is talking about the relationship between text (source code) and a performance in time by the computing machinery. It is interesting to think not only about how the gap can be reduced, but also how the text and the performance can fall out of alignment. Of course bugs are the obvious way that things can get misaligned: I instructed the computer to do X but it did Y. But as readers of source code we have expectations about what code is doing, and then there is the resulting complex computational performance. The two are one, and its only our mental models of computing that allow us to see a thing called software. Programmed Visions explores the genealogy of those models.

The other striking thing about Programmed Visions is what Chun says about memory. Von Neumann popularizes the idea of computer memory using work by McCulloch that relates the nervous system to voltages through the analogy of neural nets. On a practical level, what this metaphor allowed was for instructions that were previously on cards, or in the movements of computer programmers wiring circuits, are moved into the machine itself. The key point Chun makes here is the idea that Von Neumann use of biological metaphors for computing allows him to conflate memory and storage. It is important that this biological metaphor, the memory organ, was science fiction – there was no known memory organ at the time.

The discussion is interesting because it connects with ideas about memory going back to Hume and forward to Bowker (2005). Memories can be used to make predictions, but cannot be used to fully reconstruct the past. Memory is a process of deletion, but always creates the need for more:

If our machines’ memories are more permanent, if they enable a permanence that we seem to lack, it is because hey are constantly refreshed–rewritten–so that their ephemerality endures, so that they may “store” the programs that seem to drive them … This is to say that if memory is to approximate something so long lasting as storage, it can do so only through constant repetition, a repetition that, as Jacques Derrida notes, is indissociable from destruction (or in Bush’s terminology, forgetting). (p. 170)

In the ellided section above Chun references Kirschenbaum (2008) to stress that she does not mean to imply that software is immaterial. Instead Chun describes computer memory as undead, neither alive nor dead but somewhere in between. The circuits need to be continually electrically performed for the memory to be sustained and alive. The requirement to keep the bits moving, reminds me of Kevin Kelly’s idea of movage, and anticipates (I think?) Chun (2016). This (somewhat humorous) description of the computer memory as undead reminded me of the state that archived web content is in. For example when viewing content in the Wayback machine it’s not uncommon to run across some links failing, missing resources, lack of interactivity (search) that was once there. Also, it’s possible to slip around in time as pages are traversed that have been storedat different times. How is this the same and different from traditional archives of paper, where context is lost as well?

So I was surprised in the concluding chapter when Chun actually talks about the Internet Archive’s Wayback Machine (IWM) on pp 170-171. I guess I shouldn’t have been surprised, but the leap from Von Neumann’s first articulation of modern computer architecture forwards to a world with a massively distributed Internet and World Wide Web was a surprise:

The IWM is necessary because the Internet, which is in so many ways about memory, has, as Ernst (2013) argues, no memory–at least not without the intervention of something like the IWM. Other media do not have a memory, but they do age and their degeneratoin is not linked to their regeneration. As well, this crisis is brought about because of this blinding belief in digital media as cultural memory. This belief, paradoxically, threatens to spread this lack of memory everywhere and plunge us negatively into a way-wayback machine: the so-called “digital dark age.” The IWM thus fixes the Internet by offering us a “machine” that lets us control our movement between past and future by regenerating the Internet at a grand scale. The Internet Wayback Machine is appropriate in more ways than one: because web pages link to, rather than embed, images, which can be located anywhere, and because link locations always change, the IWM preserves only a skeleton of a page, filled with broken–rendered–links and images. The IWM, that is, only backs up certain data types. These “saved” are not quite dead, but not quite alive either, for their proper commemoration requires greater effort. These gaps not only visualize the fact that our constant regenerations affect what is regenerated, but also the fact that these gaps–the irreversibility of this causal programmable logic– are what open the World Wide Web as archive to a future that is not simply stored upgrades of the past. (p. 171-172)

I think some things have improved somewhat since Chun wrote those words, but her essential observation remains true: the technology that furnishes the Wayback Machine is oriented around a document based web, where representations of web resources are stored at particular points in time and played back at other points in time. The software infrastructures that generated those web representations are not part of the archive, and so the archive is essentially in an undead state–seemingly alive, but undynamic and inert. It’s interesting to think about how traditional archives have similar characteristics though: the paper documents that lack adequate provenance, or media artifacts that can be digitized but no longer played. We live with the undead in other forms of media as well.

One of my committee members recently asked for my opinion on why people often take the position that since content is digital we can now keep it all. The presumption being that we keep all data online or in near or offline storage and then rely on some kind of search to find it. I think Chun hits on part of the reason this might be when she highlights how memory has been conflated with storage. For some the idea that some data is stored is equivalent to having been remembered as well. But it’s actually in the exercise of the data, its use, or being accessed that memory is activated. This position that everything can be remembered because it is digital has its economical problems, but it is an interesting little philosophical conundrum, that will be important to keep in the back of my mind as I continue to read about memory and archives.

References