The Brooklyn Rail

On September 26, the Gallery Climate Coalition (GCC) announced the Art Fair Co-Commitment Statement, an alliance of “13 organisations representing more than 40 art fairs … committed to a consensus on new standards of operating.” Fairs like Art Basel, Frieze, Liste, and TEFAF will halve their carbon emissions by 2030 and aim for near-zero waste practices. The announcement comes with specific commitments and GCC assists with a seventy-two-page Art Fair Toolkit for Environmental Responsibility.

Art fairs’ environmental impacts typically focus on air travel and freight, but GCC’s Toolkit presents a bigger picture: single-use waste, diesel generators in temporary venues, and the costs of climate control. Websites might be considered, too, as became clear during a panel on September 28 with Michael Connor and Bri Griffin from Rhizome in conversation with Alex Nathanson and Benedetta Piantella about their project with Tega Brain, Solar Protocol, presented as part of the exhibition Group Hug at Onassis ONX.

Solar Protocol is a social software and hardware project that runs websites through a network of solar-powered servers around the world. The information of a website is stored on each server; when you search for that website through a browser, the server with the most sunlight (and closest to you) sends out the information that you then access. Though I have simplified, their website explains the startlingly straightforward process of how Solar Protocol works, with instructions for hosting a website on Solar Protocol, assembling your own server, and so much more. They produced a descriptive zine that presents how anyone, for less than 1,500 dollars, could manage their own solar powered media. The zine describes their efforts as engaging “the aesthetics of the small, the immediate, and that which can be stored, but not always accessed all of the time. It asks us to think about power locally and rest in equal measure.” If a person can do it, then why not a larger organization?

Rhizome did just that. Their interest in addressing the carbon impact of digital technologies and creating another node on the Solar Protocol network required research and support. This was not an overnight affair, which is why organizations like art fairs with goals to achieve in the next five years may wish to consider such steps now. Rhizome’s Lead Developer Mark Beasley established a node in California with a climate impact portal that presents the energy usage, as well as clear descriptions of their decision making process and long term goals. Everyone knows that measuring carbon emissions for digital projects is “complex and imperfect,” as the Rhizome page reiterates many times because of the distracting disputes that such quantifying practices can become from the overall endeavor. Perfect can be the enemy of the good, as the saying goes, and environmental efforts often learn how to improve through practicing what is possible, which then reveals next steps. Rhizome’s mission to support and maintain born-digital work—ranging from art as websites, video games, browsers, but also including a video platform and exhibition sites—requires many servers and softwares, which also poses a fiscal burden on the non-profit organization. As Connor explained during the panel, adopting Solar Protocol served the organization’s ethos and revealed ways to reduce monthly server bills.

Solar Protocol reimagines infrastructure and makes visible the energy demands and connections enabling access to a website. As Nathanson and Piantella said, energy intermittence in such a network occurs at multiple levels: the geopolitical, like Chinese internet policy; and the personal, as, for example, if a relationship ends so that the main caretaker of a server loses access to the roof where hardware is kept, making the server go down for a period. When we think of underwater cables, satellites, and data centers, we may not remember the mundane lives of those who manage this hardware and software, but that social element must be reinserted into our imaginaries for the kind of overall change that these ecological efforts present.

In 2020, Piantella, Nathanson, Tega Brain, and Keita Ohshiro published a white paper—“Solar-Powered Server: Designing for a More Energy Positive Internet”—with revisions to their practices occurring through working with the community of volunteers undergirding the network. This is as it should be. The system’s global scale and multiple parties ensures resilience. Even if some servers are down, a website’s information is still available on the others.

All this reminded me the importance arts have in showcasing new practices of making, maintenance, and engagement. Privacy and security are often invoked around why data can’t be shared, but that doesn’t apply always and everywhere, and the arts can present how, where, and when it can be. These small cases can make big changes as social, software, and hardware infrastructures meet and alter each other.

As Griffin asked: “Can computers exist in a way that doesn’t harm the environment or the people interacting with it?” Enormous image files, every second of a sound or video has an impact. Rhizome’s Natural Cycles open call invited submissions to reimagine the expectations of a website that would then be hosted on Solar Protocol’s node. Rather than the gross (in)efficiency of structural choices that UX design typically implements for the purpose of limiting future changes, this call required rethinking how, what, and when information is provided. As an initial foray, Rhizome partnered with Gray Area Festival on their tenth anniversary for a workshop, titled Life in 256KB, that introduced the notion of tiny data as a limitation and an opportunity, which attendees then experienced when given “a data budget of 256kb each to make a solar-friendly website.” People could choose to collectivize, and so combine their overall data budget, or work independently within the constraint.

Participants could choose.

Choosing how to manage data—as explored by the Life in 256KB workshop—brings me to the questions of agency, and its myriad theories, that have been an ongoing part of my research as relates to “artificial intelligence”—a term I continue to dislike for its broad application to the point of confusion. Machinic agency, autonomous machines, and other such terms around generative AI reiterate a problem within the overall discourse of agency. Agency is a term that means broadly similar things, but significant distinctions appear across disciplines like computer science, psychology, sociology, law, with rampant disputes in philosophy on its meaning. Without going into academic detail about each of these, agency is generally used as the ability to make something happen. This common conception already presumes a great many things, like collating autonomy and agency, where autonomy is understood as the freedom to choose (self-government), and agency is having the necessary resources and abilities to enact that choice. People need autonomy to enact agency but agency can be limited by degrees of autonomy.

The work of Georgica Pettus currently part of Incidental Container at Public Works Administration (PWA) launches into some of this confusion as regards the advent of Generative AI. In the video work Coincidence of Wants (2022), Pettus describes AI as being distinct from other softwares not due to “a structural difference but a theatrics of autonomy.” Those theatrics matter by enabling the illusion that automation equals autonomy. When we don’t understand the operations that lead to an activity, or know that our involvement was not sufficiently causal for the outcome, it is hard not to presume some kind of independently willed action.

Pettus’s work also invokes a theatrics of authority (often associated with autonomy), which dissipates as the unmoving camera allow viewers to absorb details of her presentation. She is in a gray suit seated in a conference room, but the room is small, fluorescent-lit, and without windows, all of which suggest the mindless contexts of middle-management, that is a cog in bureaucratic wheel, with reduced agency. The chairs are mismatched. Her suit is ill-fitting. No glossy tech demo here! Her clear speech retains the monotone rhythm of a voice over, and her generally affectless expressions and comportment imply an unprejudiced, balanced perspective, with presumptions therefore of autonomy due to lack of bias; but the even tone in her speech also sounds programmatic or ‘mindless,’ thereby undermining the individuality we link to autonomy and so associate with agency. When she compares AI to Adam Smith’s invisible hand to explain that we have “a tendency to offload our agency despite our continued involvement” as a way to resist responsibility, but then points out that neither of these two ghostly figures can “accept responsibility in a meaningful way either,” the insidious and confusing ways we attribute and assess agency become even more apparent, to which I will return in the conclusion.

Pettus’s work is situated within a highly conceptual exhibition by Jason Isolini that cross-sects the global politics of art storage, the largely ignored economics of digital art storage (and our general data hoarding), galleries as temporary storage, the consumerism contributing to the growth of self-storage units, the circulation of objects and people through these contexts, the documentation of artworks, freight labels for transfer and frequently necessary for insurance, the question of an artwork’s being seen as part of its validation (as well as the artist’s), the documentation of having seen an artwork, and the circulation of that documentation. It is worth noting that an artist will make choices across these moments based on different values and considerations. Trade offs occur.

Incidental Container iterates upon an earlier exhibition produced in a CubeSmart storage unit for the duration of a month-long free trial period. There, works by nine artists including Jake Brush, Molly Soda, and Sarah Friend were shown on equipment using a fifteen-day free return policy. At that halfway point, Isolini produced a 360-degree virtual tour of the installation, which code he then turned into translucent curtains, which hung in the storage unit for the duration of the show. Those are now available as documentation artifacts sewn into tote bags—another temporary container. At PWA, Isolini has created a semblance of the storage unit with video documentation of those artworks from the inaccessible gallery within the storage unit streaming through a free thirty-day trial Dropbox account on monitors likewise on free-trial basis. The gallery is located down the stairs of the 50th street 1 train station—transience all around.

This human variability gets ignored in most projections of global climate impact, since “integrated assessment models” (IAMs) adopt neoclassical economic theory, as described in an excellent piece in Triple Canopy, with its presumption of rational agents who:

choose the cheapest options and maximize the sum of welfare. Perfect information is fed into their brains. Their foresight is equally perfect; they do not hesitate or make mistakes…. Not flesh or blood, this figure is the utility-maximizing subject, whose discernment is unclouded by any particular allegiances, interests, outside influences, or other idiosyncrasies. … No classes tear it into halves. No relations of race, gender, or core and periphery mark it with tensions or fissures.

Economic models oversimplify our agency in attempts to predict the impact of huge things like climate, energy, or AI. Timothy Morton calls “things that you can study and think about and compute, but that are not so easy to see directly: hyperobjects.” Maybe rethinking agency can help.

Hurricane Helene slammed those hyperobjects together when over two feet of rain hit Spruce Pine, North Carolina where two mines produce resources for most of the world’s solar panels and semiconductors—“ultra pure Quartz for crucibles that all global semiconductor production relies on” as one geologist put it—with consequences for global economics. In The New Yorker, reviewing Ed Conway’s book Material World: The Six Raw Materials That Shape Modern Civilization (2023), Elizabeth Kolbert described the region and its significance. So even if I had convinced you earlier in this article to make your own solar-powered server, you might now suddenly feel defeated by this news.

There is very little sense of agency regarding these systems.

The sense of futility stems in part, I believe, from the all or nothing attitude towards agency.

The philosopher Jennifer Corns posits “distinct agentive forms” that helpfully broaden how we can conceive of ourselves as agents. In future articles, I will go into the details of the notion of minimal agency from which Corns derives her argument (for the purposes of this explanation, I am focusing, as she does, on the forms of agency enjoyed by humans); key here is recognizing that agentive forms “are overlapping and dynamically related, but the systems, capacities, and norms which partly constitute these different agentive forms are nonetheless usefully distinguished,” with at least three extant by virtue of being biopsychosocial beings.

Posthumanist theories, science and technology studies, and social constructivism have argued for the way in which one is never isolated and independent, but a being influenced by and influencing an environment. So what kind of agency is possible regarding the climate crisis or generative AI? Corns usefully emphasizes the plurality of agentive forms within any identifiable agent. Your social agency may supersede your biological agency for some period of time if you think that you really need to talk to someone at an event, but the body can overwhelm social agency so that you leave the party banter for the bathroom. We recognize upbringing as influencing one’s potential. The list goes on. As Corns indicates, “[o]ur agentive forms likely proliferate. So, for some instances, it seems to me that we have financial, aesthetic, familial, and creative agency.” She elaborated on this in a webinar to explain that “Whenever we can identify a range of activities, an environment in which the activities are done, and a standard of success for succeeding in that range of activities, we have identified a form of agency.”

It gets more complicated. We know that biological hunger can alter psychological states, which can impact social interactions. We then excuse such mindsets and cranky behavior by attributing it to hunger; this represents an attitude open to Corns’s multiple agentive forms, since we don’t denigrate your social agency for things we attribute to biological agency (although repeated behavior of this kind might lead me to judge you as someone lacking in the ability to regulate your agentive forms since your social agency inadequately plans for biological needs leading your biological agency to derail social norms). Corns’s argument makes explicit this model of multiple and distinct agentive forms in which we are already participating every time we assess the complexity of agencies presented in our encounters with each other. As she further explained in conversation with me: “These many forms are functionally integrated and regulated by a range of processes. Conscious decision making is one type of process through which one form is prioritized over another, but there are others.”

Agentive norms indicate the range of conditions under which an agent constitutes its integrity as such, so with agentive forms that means a set of norms for biological, psychological, social, familial, etc. behaviors. My biological agentive form might have norms that only require eating every six-ish hours, where a diabetic might have norms necessitating food more often. My psychological agentive form might find it perfectly normal to spend several days without speaking to anyone, where another might become distressed without human interaction over the course of a day. As adults visiting aging parents, we may struggle as norms of familial agency change across all parties. Each form of agency involves norms that specify when one is succeeding as a friend, a daughter, a thinker, and so on. These standards of success will differ across people, cultures, and times, though there seem to be some loose generalizations possible across each form.

This goes some way to explaining the wildly different responses to generative AI. An artist who uses softwares and enjoys coding may experience it as enhancing their creative agency, with possibilities for broadening their social or financial agency. An illustrator who sees these programs as impacting their employment feels their creative agency devalued by employers preferring the speed of slick renderings in Generative AI, with negative consequences for that illustrator’s social and financial agency.

Identifying distinctions within agency provides a way to understand what agentive forms are being disrupted and clarify why that matters, to whom, under what circumstances. That can also illuminate why some other agential entity might be indifferent to me, if it doesn’t know about or value an agentive form as I do. It can also clarify interpersonal conflicts around differing norms for distinct agentive forms. In our discussions about the (dire) state of things, recognizing distinct agentive forms may reveal problems in the discourse and new ways to pursue the goals we seek. In next month’s column, I will revisit these ideas as they relate to the concept of collaboration in the use of Generative AI and production of generative art.