Art In the Age of Obsolescence
Rescuing an Artwork from Crumbling Technologies
Six months ago, I wrote some thoughts on the paradigm shift in how museums think about the balance between preserving an artwork and sharing it with the public. My conclusion was this: time-based media art installations do not truly exist until they are installed and, thus, these works must be exhibited — or “exercised” — with a certain degree of regularity. This is a concept first championed in the conservation field over a decade ago, by Pip Laurenson of Tate. Lovers, by Japanese media artist Teiji Furuhashi (1960–1995), is an excellent example of this. What follows is the story of how our team rescued this important example of early-1990s Japanese media art from a crumbling foundation of obsolete technologies (MS-DOS and LaserDisc, for starters) and ensured that it will live on so that generations long into the future are able to discover and enjoy it.
In museums, where the work of the conservator is often dictated by what is scheduled for exhibition, it is unusual to have the opportunity to conduct in-depth research or conservation treatments on collection works that are not slated to appear in the galleries. There is, however, a rich tradition at The Museum of Modern Art of offsetting this trend through collaborations with academics and researchers. Through this, we are often able to build small-scale research projects that give students incredible real-world experience — and afford museum conservators the sort of research we wish we had more time for. About a year ago, I realized that Lovers was a perfect case study for a course I was teaching at New York University, called Handling Complex Media. The artwork is composed of a veritable cocktail of technologies and media formats: 35mm slides, analog video, robotics, software — you name it. So we pulled Lovers from storage, along with its two-inch-thick folder of documentation, and began the work of understanding just what we had.
What does it look like when a museum pulls a time-based media installation artwork from storage?
Lovers had not been exhibited since being acquired by MoMA in 1998, and when it was first shown in 1995 (prior to acquisition), the Museum hadn’t yet established its internationally renowned program of media conservation. Therefore, much of our initial work was about discovery. What do we have? How do we connect all the parts? How do we even turn it on? From storage, we withdrew a panoply of materials…
Out came LaserDiscs, 35mm slides, speakers, wires, accessories, slide projectors, an eight-foot-tall metal tower containing video projectors with robotics to control which direction they are pointing, two flight cases full of behind-the-scenes control hardware and software, and a hefty folder containing documentation, manuals, installation specifications, and correspondence with the artist and his studio. Our art handlers carefully delivered all of this material to a small viewing room at MoMA’s art storage facility in Queens. Although Lovers calls for a 32 x 32' space for proper installation, this was the best we could do for a basic assessment. After two days of combing through manuals and carefully wiring the various components to one another, we were ready to power on the artwork for the first time in decades.
The power button is pressed. Hard drives whir and click. Several beeps and clicks later, the artwork comes to life. The metal tower full of video projectors sputters as its robotics begin spinning the video projectors around on their axes. The computers spew cascading lines of inscrutable text and code onto their cathode-ray-tube screens…
With the artwork completely wired, and (to the best of our knowledge) running properly, we were now ready for our group of NYU Moving Image Archiving and Preservation (MIAP) students to begin their research project. The class was tasked with understanding and documenting the following: What is the anatomy of the artwork? How does it work? What condition are its various components in? What components are at risk of failure? Where can we source backups and/or replacements for the exact components used by the artist, and if exact replacements are not available, which components have significant aesthetic impact on the work beyond mere behind-the-scenes utility?
Four weeks later, the students presented their findings and a comprehensive 33-page report to MoMA conservators and curators. As the work was not slated for exhibition, all was then packed up and sent back to storage. Thankfully, however, our research revived interest in exhibiting Furuhashi’s work, and we were soon tasked with assessing the feasibility of preparing Lovers for exhibition…in four months. Preparing such a technically complex installation for exhibition means being confident not only in our ability to install the work properly according to the artist’s instructions, but also in its ability to run for 10 hours a day, seven days a week. We knew that we absolutely had to replace any technical components that were at risk of failure and could then not be replaced, and thanks to the students’ in-depth research and analysis we already knew exactly which of these components were at risk of failure.
The original LCD video projectors and the behind-the-scenes control hardware needed to be replaced due to their instability and rarity. This meant a full-on re-implementation of the original control and timing hardware and software would be necessary. Additionally, the NYU MIAP students’ research had revealed several gaps in the installation documentation. There were many unknowns regarding the parameters for successful installation, questions we knew we could only answer by working with Shiro and Yoko Takatani, Kyoto-based members of the Dumb Type artist collective and performance group, of which Furuhashi had been a pivotal member. Due to his battle with AIDS, Furuhashi was frequently hospitalized during the creation of Lovers, and Shiro Takatani was responsible for much of the artwork’s technical execution. His input would be critical in our efforts.
After our initial conversations with Takatani, and consultations with a few engineers, we assessed that it would be feasible to, in essence, “stabilize” Lovers in time for the exhibition. Our aim was to replace the at-risk components, translating the work to more stable technologies, while prioritizing two essential tenets of conservation — minimal intervention and reversibility.
Our adventure began with the video projectors. My colleague Kate Lewis carefully studied and documented the original projectors — their brightness, throw distance, and fundamental method of projection (in this case, LCD). Armed with this information, she selected a few seemingly suitable replacements that met these parameters as closely as possible. Fundamentally, though, selecting the appropriate replacement came down to seeing. We conducted projector “shootouts,” during which each option was compared side-by-side by MoMA conservators, curators, audiovisual technicians, and Shiro Takatani.
Replacing the video projectors also meant considering the hardware that cradled and mounted the original video projectors to the robotics of the tower. Working with W. Thomas Porter, a custom fabricator, we designed and built new mounts that fit the new video projectors, but completely matched the look and feel of the originals, thus not introducing any material inconsistencies (other than, of course, the drastically smaller LCD video projectors). Kate and Thomas meticulously documented their work, saving and labeling all of the original hardware as they removed it. Original hardware was packed with its corresponding projector for long-term storage.
Meanwhile, I began working with engineers Mark Hellar and Ranjit Bhatnagar to reverse engineer, understand, and document exactly how the original custom software and IBM-compatible PCs controlled the robotics, video playback, and interactivity. Before we could do this, though, we needed to engage in a bit of preventative conservation: we needed to ensure that, if these PCs were to fail before we were able to complete our reverse engineering, we would have a backup plan. We did so by producing disk images of their hard drives — essentially a complete recording of their contents. This not only provided us with a preservation master of the original hard drives, but also gave us a safe copy to work from, so that we could closely inspect the original software without needing to touch the original computers.
On the first PC — which controlled the robotics for the slide projectors and the interactive video of Furuhashi — we found nothing that aided in our understanding of how the work functioned. On the second PC, however, we were astonished to find a text file containing timing and position instructions for the remaining four robotic motors. This file was, in essence, the equivalent of a musical “score” for those four motors that told us what the motors were doing, and in what sequence. What it did not tell us, however, was how the software and PC hardware actually accomplished this.
At first glance, the hardware connecting the PCs to the robotics was completely incomprehensible. How did it work? There was only one way to find out. We needed use analytical and diagnostic tools to reverse engineer exactly what the PCs were doing.
Using a Saleae logic analyzer, we were able to record the signals the original PCs were sending to the motors. These precise patterns of electronic pulses were our evidence of how the computers acted out the timing metadata, or “score,” that we had found in the brains of the computer.
With a custom-written 48-line Python script, we were able to gather our data in a way that was completely automated and far more accurate than we could have hoped for. We hooked the Saleae up the original control equipment and began gathering data.
Now we had the score, and we knew how to perform it, but that still was not enough; all of this documentation was very scientific and precise, but it didn’t tell us how the work felt. Furthermore, there was still the PC that did not contain plainly readable metadata, and only an impenetrable binary file. How could we reverse engineer the robotics and behavior of the slide projectors and interactive video of Furuhashi? Observation and careful documentation was the answer. I proceeded to spend hours upon hours running the original system, carefully watching and listening to the robotics, while also capturing video and audio documentation. In the end, it was this direct observation that allowed us to reverse engineer the basic algorithm.
Armed with all of this evidence, we set about devising a new system to faithfully reproduce the timing and behavior of the original control hardware. This began by designing a system using an OpenFrameworks program to control the timing, interactivity, and video playback. This program would talk to an Arduino microcontroller in order to send commands to the motors. As we worked to design and construct this new hardware and software, MoMA’s exhibition team was hard at work designing and constructing the gallery space according to Furuhashi’s specifications.
Once this space was built, we migrated our little software and hardware development lab into the gallery. We had reached a critical juncture: does it work? Because the robotics run in synchronization with the videos of the figures walking and running, if anything is even remotely inaccurate, it would be immediately apparent due to the mismatch between what the projected figure was doing and what the motor was doing. With bated breath, we connected everything yet again, and pressed the start button.
At first we were blissful. The motors were whirring with activity, the figures were walking around the room and simply “looked” right. That is, for first 10 seconds or so. Slowly, the motion of the figures fell more and more out of sync with the motion of the robotics. Consulting our documentation, we realized that there was an incredibly subtle but crucial detail we had overlooked: we had oversimplified the elegant motions of the robotics. And so, back to the drawing board. Many calculations (and some trips to the mathematics stack exchange) later, we had the algebra that would give us an accurate reproduction of the original motion:
x = 0.269704 + (342.283/y)
We ran our new code through some visualizations and analysis, and were pleased to find that our new approach would bring us to within .0002 seconds of accuracy compared to the original. A few hours of coding later, Ranjit had implemented this new approach.
Now came the ultimate moment of truth: if we tested the newly improved software and the motors still weren’t right, it would mean our entire approach was somehow fundamentally flawed. Had we misinterpreted the data? Was there a bug in our software? It was now mere days before Takatani would arrive to review the installation. With all of this weight on our shoulders, we loaded the new software, held our breath, and pressed the start button one more time.
Immediately we knew we had nailed it. The motion of the figures was dead-on, and maintained synchronization for hours. High fives were exchanged. Now it was time to simply let the artwork to run continuously for hours on end, to simulate the stress it would encounter being run for the course of the exhibition. All was well in the world, our work nearly finished. Thankfully, the artwork continued to run successfully for days, with only one major surgery required to repair some aging and brittle electrical wires.
With all of our scientific and analytical work complete, it was now time to turn our attention to the “soft skills” and in-depth understanding of the artwork’s look and feel that we could only learn from Shiro Takatani. Our conservation files from the 1990s contained no documentation of the proper light levels in the space, no examples of proper projection alignment, and no record of the correct configuration of the eight channels of sound.
Not only was Shiro responsible for the technical execution (and, in essence, the art direction) of Lovers, he had also been responsible for supervising its proper installation and exhibition up until its ultimate acquisition in 1998. Now, exhibiting the work for the first time at MoMA in 20 years — and now with a staff of media conservators — we were ready to engage in the transfer of knowledge from the individual to the institution.
Once we had completely documented all of Shiro’s knowledge regarding alignment, lighting, and sound, he told us it was time to move on to the refinement and correction of the motion and timing of the robotics. With puzzled looks on our faces, we reminded him of our quantitative proof that we had reproduced the timing and motion of the original control software within a completely imperceptible .0002-second margin of error. Smiling, and ever patient, Takatani, who had stewarded this work for years, explained that the timing of Lovers had been reviewed and refined nearly every time the work was installed. He suggested, therefore, that although we had perfectly reproduced the behavior, timing, and motion of the final snapshot of the artwork as it existed when it was collected, it was now time to continue its active life, and carefully refine the motion as Furuhashi would have wished. After much discussion among conservators and curators, and knowing that all changes we made would be finely documented and completely reversible, we agreed.
Shiro and I sat together in the installation for days, carefully studying each moment where one of the projected figures began to walk, came to a stop, or suddenly turned to run in the opposite direction. Taking extensive notes as we observed, we went through round after round of refinement, until eventually Shiro turned to me, smiled, and nodded. We had done it. This time it was not high fives that were exchanged, it was hugs.
Had Teiji Furuhashi not tragically passed, well before his prime, in the early 1990s due to AIDS, he would have been 56 years old.
On the final evening of his old friends Shiro and Yoko’s visit, we threw a dinner in his, and their, honor. Although congratulations had already been exchanged for a successful conservation campaign, it wasn’t until this evening that I knew we had truly succeeded. It was then that Yoko raised her glass and explained that, for decades, keeping Teiji’s legacy alive through Lovers had been their responsibility, and that now they were passing this torch to MoMA. These words spoke far louder and clearer than any kind of technical analysis or evidentiary data produced during our work.
Despite the great effort exerted by my colleagues and myself, work has in many ways only begun. Just as the original equipment that controlled Lovers had aged, obsolesced, and become unusable, so will our newly restored solutions. The field of conservation is continually evolving, not merely technologically, but philosophically and ethically. The day may come when our work here seems somehow wrong or misguided, so it is our job as responsible conservators to ensure that we produce the requisite documentation, ensuring that our work is truly reversible. That way Lovers can live on under the stewardship of another generation of curators and conservators, who will leave their own trace on the palimpsest that is the history and evolution of a time-based media artwork.
This post is dedicated to the memory of Teiji Furuhashi (1960–1995). Teiji Furuhashi: Lovers is on view at The Museum of Modern Art through April 16.