Scale

Craft scales much differently than a machine learning system, and these frictions can be used for aesthetic investigation and activist resistence. Detailed hand labor is a critical tactic available to tackle the seemingly insurmountable scale of the increasingly giant datasets, obscured models, and complex infrastructures. For example, Everest Pipkin viewed every single video in a large dataset over the course of months and turned this labor into a series of artistic investigations (Pipkin 2020). At the Knowing Machines research project (I discuss my work with them more in the “Formative” section), interdisciplinary researchers from journalists to programmers to lawyers work together to create tools to explore changes in foundational image datasets like ImageNet (Crawford et al., n.d.). These practices apply human-scale attention to machine-scale systems in order to reveal their obscured innerworkings. Just because a system is larger than life does not make it impossible to apply human techniques to investigate it. While the gargantuan computational scale makes it more tedious, this human-scale labor is still an incredibly important response for understanding and confronting algorithmic harms.

For Eve Kosofsky Sedgwick, texture is a way of talking about the space between dimensions or scales, or the transition between them. I have found her metaphoric comparison useful and have extended it to machine learning: * From point to line to cube to cube in space-time to multidimensional universes of vibrating strings. * From ball to string to loop to fabric to garment. * From scalar to vector to matrix to tensors in many dimensions.

Sedgwick describes the many directions of touch and contact across a surface and between dimensions, warp and weft of all kinds. She builds on this a theory of relationality, which understands the reciprocity of contact, the reciprocity of creating and being, and the fractal nature of scale:

“the sense of touch makes nonsense out of any dualistic understanding of agency and passivity; to touch is always already to reach out, to fondle, to heft, to tap or enfold, and always also to understand other people or natural forces as having effectually done so before oneself, if only in the making of the textured object.” (Sedgwick 2012)

Here Sedgwick connects the experience of making craftwork with experiencing the made craft, with touching and being touched by, and with multidimensional sense and scale. I also see this as a way of understanding machine learning modeling through new, non-hierarchical metaphors. What is the scale and dimensionality of machine learning? Does it “dramatize its dimensional betweenness,” as Sedgwick describes the way a folded dyed fabric or a paper snowflake, once unfolded, retains the memory of past shapes? (Sedgwick 2012). Or Ingold similarly describes the way once knotted ropes retain their bends: “But the knot remembers everything, and has everything to forget” [Ingold (2015)][, and Laura Marks says “electrons remember” (Marks 2002)]? These material memories, as marks of their current and past states simultaneously, could be tangible metaphors for understanding machine learning, or even quantum computing (folding was a fascination of quantum physicist David Bohm). They show that, when working with any material, the material works back.

Thinking through technology with craft lenses opens more points of reference for addressing complexity and complex systems that prexist the emergent techonlogies we are encountering now. It reminds us that we can slip between scales to find apt materials and metaphors for grappling with the complexity of systems more approachably and sustainably. Movements across scales, and finding the spaces between and resistances to scale, are also useful explorations. What gets lost in the increase, and what is gained (to whose benefit) as we move either direction?

“human care can’t scale up. Care happens in small-scale moments of human interaction; scale is achieved by tireless and impersonal machinery” (Seaver 2021)

“As hand knitting inclined to the margins as a [large-scale] viable manufacturing process, it was its long history of connecting people with their environment that came to the fore—from material source to the makers’ largely unwritten generational knowledge of patterns and techniques—in clearly locating the craft of knitting against distinct communities and indigenous practices.” (Steed in Nimkulrat, Kane, and Walton 2016)

Sustain(ability)

“The mere act of owning real tools and having the power to use them is a radical and rare idea that can help change the world around us.” (Schwarz 2010)

Craft suggests more time spent with one’s objects and materials — allowing for repair, error, and air. For woodworker Christopher Schwarz, it means simple tools known well, a small collection of tools in a toolbox that can be carried with you anywhere in order to complete your work, providing freedom and autonomy (Schwarz 2010). It accumulates through simple gestures repeated until they become habit (muscle memory, durational practices, implicit knowledge), repeated until they become complex patterns (quilts, quines rice kolams), repeated until they make space for other things to happen in their presence (sewing circles, hacker spaces, political action). Imagine a simple machine learning tool as easy to pick up as a garden trowel. Would this be an all-purpose ‘multi-tool’ like today’s resource-hungry giant language models? No. But its users would have freedom to imagine their own goals and outcomes from a broad range of possibilities they design themselves. Imagine if machine learning systems were built from these operating principles instead of requiring such massive financial resources, natural resource, and human resources to operate that only a few huge conglomerates can particpate in their creation.

“the pre-Industrial craftsman didn’t seem to have secret tricks as much as he had lots of opportunities to practice and become swift.” (Schwarz 2010)

Builder Alexander Langland returns to the older definition of ‘craeft’ as a lifesustaining skill built over generations. He sees craftwork as holistic interrelation rather than individual objects or skills. He says, “Crafts have always been determined by the immediate environment, and indexed to the resources of the natural world.” Imagine if we saw technology’s global supply chains as the integrated lifecycle of care that Langland describes, which starts with:

“tended landscape > sustainable production of raw materials > intelligently processed > beautifully made > fit for purpose > fondly used > ingeniously reused > considerately discarded > given back to the earth” (Langlands 2019)

“Casualized labor begets commodity toolsets, frictionless and uncritical engagement with content, and shallow practices of use.” []

Infrastructures of care

Approaching technology through the lens of craft slows it down and asks it to sustain. It also [organizes people, labor, and relations] differently. Anthropolgist Gabriella (Biella) Coleman aligns hacking with crafting, focusing on the “intellectual guile” required for both and the guild-like communities that form. “[A]ll hackers fit the bill as quintessential ‘craftspeople,’ […] Even if craftspeople tend to work in solitude, crafting is by definition a collectivist pursuit based on shared rules of engagement and standards for quality” (Coleman Tue, 06/07/2016 - 12:00). Consider the transfeminist server, which approaches its technology as a crafted, social, interrelational object:

“rosa [travelling server] is not only constituted in hardware or sotware, but also the multitude of relations that are created around the making, maintaining and passing on of this infrastructure: the processes that are performed, the affective charge of their actioning, the community around them.” (varia n.d.)

Feminist servers run slowly. Sustaining them takes “years of intense collective learning,” replete with uncomfortable tensions and messy differences. These very ambivalances, argues Ines Kleesattle, are crucial for Feminist Server Art, because they demonstrate the relational labor going unseen in other approaches to technology. “All kinds of infrastructures involve service, maintenance, and care labor” (Kleesattel in (Sollfrank, Stalder, and Niederberger 2021)). When giant data centers run in a far-flung artic landscapes, and large language models siphon water and energy and human effort for their training, these “cloud” technologies are not ephemeral or immaterial – and feminist projects move slowly to enact the material, relational stakes in all digital interactions, to show how they might be organized with more awareness and care.

In “A Traversal Network of Feminist Servers,” six communitiesConstant, varia, Hypha, Lurk, Esc Mkl, Feminist Hack Meetings. Additional projects in the spirit of or related: Open Source Kitchen, Anarchaserver.

connected to discuss their feminist infrastructure projects, which ask questions such as:

“How can the labour behind hosting be made inclusive, convivial, shared, visible not only in moments when things go wrong? In what ways does the language embedded in server administration influence the practices and interactions around it? How can servers be pedagogical spaces for experimentation for those with different knowledges?” [varia email list 28.11.2023]

Craft requires taking care and caretaking, as does technology. Transfeminist infrastructure work shows how the logics and ethics of craft – especially its domestic and feminized aspects – emerge out of intersectional feminist approaches and apply to technological questions. To consider technology as entangled with its affective and labor relations, as feminist infrastructures do, resituates tech back into a variety of communities of practice and reframes its relationship to caretaking. Although many have looked to technological solutions like bots to address issues of labor and caretaking, more often technology itself requires caretakers and creates more care labor. To respond to this, feminist infrastructures show how care is generated through human networks and how care might also be generated in, rather than depleted by, digital ones. They place technology within a circuit of human relations rather than as a replacement for it.

“The point is not only to expose or reveal invisible labours of care, but also to generate care. [… G]enerating care means counting in participants and issues who have not managed or are not likely to succeed in articulating their concerns, or whose modes of articulation indicate a politics that is ‘imperceptible’ within prevalent ways of understanding.” (de la Bellacasa 2011)

Along with the traveling Raspberry Pi server “rosa” that acted as their working hub, the Traversal Network project culminated in a print and digital publication that reflected on the different groups’ shared and diverse approaches. By choosing “low-effort publishing tools” like varia etherdump, by providing strategic services like analog infrastructure for mutual aid, as well as digital infrastructure like hosting archives and zombie sites of ongoing work, feminist servers like Anarchaserver create and codify solidarity networks using multiple mediums and tactics. These are forms of “feminist convergence” that embody their values through their choice of methods and material offerings:

“For us feminist infrastructure enables the systematization, the maintenance and the circulation of good ideas, practice and care. For us, solidarity networks are an example of feminist infrastructure. And we think it represents one of our earliest feminist technologies, perhaps the oldest and the most widespread.” (“ATNOFS-Radio-Broadcast-Spideralex-Danae-Tapia | Variapad!” n.d.)

For infrastructures of care, communities themselves are the essential networks; the servers, databases, workshops, services, both digital and analog are all tools toward building and means of expressing those community networks.

“The internet can bring us very close, yes, but it is even better to meet cuerpa-to-cuerpa. The presence of the others gives us the feeling of ecstacy, makes us float. Meeting gives us power and energy, facilitates processes, and intensifies networks of trust and cooperation. Sometimes we converge in our disputed territories, the street the networks, and the servers, and other times in places where we can create spaces of security, confidence, and relaxation.” (Spideralex in Sollfrank 2018)

Access & boundaries

Credit: USC Libraries ONE Archives

A variety of forms and modalities allows for more kinds of access. Craft, and other means of creating forms, helps imagine what access is possible and who might be missing out. For example, zine-making and zine distribution networks precede digital networks, provide models for digital practices while subsisting alongside them. This publishing craft is a counterinfrastructure used to build counterpublics and underground communities, to create counterarchives or living texts (Adema 2021), and to clarify information to new audiences. Feminist media historian Cait McKinney’s research into underground lesbian networks, both analog and digital, shows that “lesbian-feminist information infrastructures are built and sustained through networks that facilitate collaboration and resource-sharing amid precarious conditions” (McKinney 2020). Marginalized, precarious conditions of queer communities has called for interventions by alternative means. Graphic designer Paul Soulellis, who maintains the Queer Archive Work project, suggests that zines are archives of crisis: “evidence of queer life persists in the urgent artifacts that emerge in crisis — artifacts that arise expediently as part of efforts to advocate for slowness, care, mutuality, queer joy, pleasure, refusal, and community” (Soulellis 2023). The precarity of these communities requires more care and more flexible materials, and it also facilitates innovation. McKinney argues that “ways of thinking digitally about information were embedded in lesbian-feminist work with paper, in the ways lesbian feminists learned computing and imagined and built networks and databases for storing and sharing information” (McKinney 2020). Such innovation continues today in the alternative uses of digital networks seen during protests like Arab Spring, as well as in the continued use of analog zines to discuss technological objects like machine learning (in the Intersectional AI Toolkit or in Elvia Vasconcelos’s “A Visual Introduction to AI” (Vasconcelos n.d.)).

Zine publisher, activist, and artist Be Oakley says that part of the appeal of the zine is that the zine form allows for more egalitarian access across economic boundaries: “A zine is made to be disseminated to publics otherwise forgotten by hardcover books. […] Class is built into its format. A zine communicates a level of accessibility through its materials and its form. […] A zine evokes a legacy of authorship that is not exclusively written by straight, cis, white men. […]” (Oakley 2023).

Providing more access can also enable more nuanced access, by setting access boundaries. Take for example the work of Local Contexts, who help Indigenous communities protect their data rights by creating tools to inform researchers of their boundaries. They offer data labels that help researchers understand how data are meant to be used (and not used) according to more granular criteria. Rather than using a simple license agreement, which does not address their needs, Indigenous communities can reinforce their rights by applying “Traditional Knowledge” and “Biocultural” labels to data and collections, which provide context, provenance, metadata, protocols, and permissions for use. These include designating specific community members or times of year for access, or designating materials as sacred to the community or open for collaboration. Local Contexts also provides notices for researchers and institutions to contextualize their collections with labels like “Authorization,” “Belonging,” “Caring,” “Gender Aware,” “Leave Undisturbed,” and others (“Local Contexts – Grounding Indigenous Rights” n.d.). The expansiveness of the forms that a Craft lens can take means that more communities can participate, but more importantly more communities can participate by creating their own terms and contexts for participation.

Legacies & contexts

“Written out of an official history which draws them in as its minor footnotes to itself, cloths, weavers, and their skills turn out to be far in advance of the art forms digitization supersedes.” (Plant 1998)

Technology, as reframed through intersectional and craft lenses, is is always contingent and contextual. It honors historically dismissed and feminized practices of craft — which emphasize process, duration, ongoingness, community — and aligns them with the aspects of technology which frequently get overlooked: namely, the contributions to the materiality of tech from the global majority and Women of Color.

“Our bodies are archives and sites of memory that cannot and will not be overwritten, despite technological attempts to render them as such.” (Aliyu 2023)

Trans*formative systems attend to the counterhistories still alive in our present technologies and potentially reemerging in many futures. These include Hilda G. Carpenter, a Black lab technican whose “core rope memory” techniques advanced computing and supported the Apollo moon landings, by physically encoding information using copper wire and rings. Author Zainab Aliyu compares Carpenter’s physical memory-weaving processes to the Yoruban oracle system called Ifá, which also uses binary encoding and draws on “collective memory” for divination(Aliyu 2023). American Artist details the history of the graphical user interface (GUI) as a turning point in the history of computing, at which point screens changed from black to white. This was when the “user” was defined for the stated purpose of easing use, but this also narrowed and limited use, enacting white supremacist values: “Blackness has, so to say, formed the ground for white, with black gooey being antithetical to the values of the white screen” (Artist n.d.). Those with more power have frequently appropriated knowledge or extracted labor and resources to develop technologies, such as the recruitment of Navajo women to ‘weave’ circuit boards for IBM (Nakamura 2014). Such examples open the unacknowledged archives of technology that connect to craft, to ritual, to community in order to — as Xin Xin and Katherine Moriwaki say — perceive “history as a messy entanglement rather than a linear graph” (Shih and Moriwaki 2022).

These entanglements wind through material and conceptual spheres, which micha cárdenas demonstrates with her description of the stitch as emblematic of lived intersectional feminist work and of all practices of craft and technology:

“As sewing is a technique of making that has been used primarily by women throughout history, and continues to be primarily the task of women in sweatshops in the global South, […] the stitch as a material and conceptual operation can be seen as feminist, a way of generating new concepts by learning from people who have been subject to material inequalities because of their gender, their race, and their geographic location. The stitch can be thought of as the basis for a theory of feminist making, which values the forms of knowledge practiced daily by oppressed people as they make their lives in the face of violence.” (cárdenas 2016)

Craft expands the definition of what “counts” as creating or participating in technology, including what labor is valued in the accounting of it. It destabilizes static ideas of technologies and their outputs as infallable and handed-down, by reminding us they are in fact crafted by people and contingent on contexts. The invisible labor of data creation and data maintenance is no different from the invisible labor of infrastructure in this way. Yanni Loukissas argues for the inherently “local” nature of data, saying that “data-driven systems […] are locally contingent and even fragile. Designs dependent on data must be maintained and repaired on a regular basis to ensure that they are in sync with changes in the data themselves or the encompassing infrastructure of the place. […] When data do seem to confer transparency, it is because we are shielded from important details about the context of their creation or display” (Loukissas 2019). Loukissas suggests the term “data settings” instead of “data sets” to emphasize the ways data are always shaped by both their creation and their use. He argues for remembering how data always index something else and are themselves incomplete. This destabilization of data helps to recall their human impacts, and it asks us to care for them responsibly within their shifting contexts as part of various communities and legacies.

Reentangled theory and practice

Practice and theory are tangibly intertwined in craft. While craft is not the only domain where theory and practice merge, it is useful to draw on craft metaphorically and literally of how it merges the two accessibly while reaching across domains.

“The protocols of knitting are thus situated in-between the looped thread, in the loops themselves. This interconnection protocol of the thread could be seen as a catalyst, as it produces a bigger whole from the single thread. […] One could say that the protocol of the looped thread ‘echoes’ throughout the final knitted piece as the catalyst produces the conditions for emergent behavior (the inter-loops of the thread).” (Von Busch 2013)

Craft shows — through maker processes — how doing becomes being and being becomes doing. It may seem too simple but it is important: As actions of making yield creations, craft reemphasizes the making alongside objects made. It shows how theories of objects (be they media, tech, or craft objects) emerge from and are embedded in the actions that produce them. Attending to craft as process — and to technology as its processes, its people, its protocols — expands beyond the crafting of objects to include communities of practice and sociotechnical systems: “as in crafting an identity, or crafting a community” (Fountain 2021).

[xxx][inserted from intro]

Queerness and art have been two key radical practices for me. Art has been the space where I am able to unpack complex ideas for myself, because I can treat them more freely as artistic materials. It is where I am able to follow instinct and feel into how my tools, platforms, and forms shape their outputs and outcomes. It is also the space where I feel able to imagine wildly, creating digital objects that should exist but don’t, or couldn’t exist but might.

Artistic research opens space beyond research questions, where research tensions live. In that space, I can sit a bit longer with questions I know I cannot answer, questions that make me uneasy. I can hold two contradictory ideas simultaneously and let them push–pull me, forgiving myself imperatives and outcomes, productivity and proven hypotheses. I can put myself into the trouble, because I already embody these questions in my lived experience.

This imaginative work is central to supporting very practical next steps and strategies. It is central to supporting more access to communities of practice where others can continue the kinds of artistic experimentation that challenge paradigms and creates new forms I might never imagine otherwise.

“artists saw value in reaching shared normative practices through engagement between artists and audiences, suggesting a model for participation and collaboration often missing in other forms of digital research and development.(Stark & Crawford 2019, 45 (Stark and Crawford 2019)

“in failing to fully belong, and allowing that nonbelonging to denaturalize, emergently, its givens, research-creation tells other stories, uncanny stories, that (have the potential to) carry within them […] other ethics” (Loveless 2019).

Cinematic arts professor and critic Holly Willis (2016) argues “arts-based research is rooted in critical theory, framing the research process within the context of power, emancipation and a deep questioning of the ethical and ideological implications of knowledge and change.”

[xxx][inserted from intro]

Craft is also means of thought. Textile designer Nithikul Nimkulrat describes craft, “not only as a way of making things by hand, but also as a way of thinking through the hand manipulating a material.” She suggests that in craft, thinking is a sensory act, inextricable from doing. “Knowledge of a creative practice thus lies in and can be acquired from within the practice itself. In other words, thinking and knowing are inseparable from making in any craft or designerly practices.” This makes it imbued with responsibility to ourselves and each other.

These are not metaphors only. They are material, infrastructural systems that undergird and uphold power — woven into the warp and weft of all processes. With fiber art this is literal, and it demonstrates concretely the ways these processes operate in other mediums too. Cyberfeminist Sadie Plant says:

“[Garments] process and store data. Because there is no difference between the process of weaving and the woven design, cloths persist as records of the processes which fed into their production: how many women worked on them, the techniques they used, the skills they employed. The visible pattern is integral to the process which produced it; the program and the pattern are continuous.” (Plant 1998)

Echoes of this concept can be seen in current research methods like Critical Code Studies, an approach to reading code that situates it within its material, cultural contexts rather than viewing it as a neutral static object (Marino 2020). Through such material reckonings, we can not only use different lenses to reflect on existing technologies but also gain the means to reimagine them for different purposes. As queer craft art historian Daniel Fountain says, “craft, or more broadly the notion of crafting, is an essential strategy for living and loving ‘otherwise’” (Fountain 2021). What does this ‘otherwise’ look like and how does it come about?

Artist-researchers Ren Loren Britton and Helen V. Pritchard call for “queer playfulness and promiscuous metaphorical practice” that can expand beyond passive strategies for remaking technoscience, reusing its terms in order to imagine new forms: “as a speculative practice of working with metaphor, we propose to replace CS (Computer Science) with CS (Careful Slugs), forthcoming could be CS (Cushions and Stargazing), CS (Crying Sabotage), and/or CS (Chancer Scientist). We propose that the metaphors of Careful Slugs might create possibilities for new kinds of work in Computer Science. Possibilities in which unknowing itself become disoriented, disassembled, and undone in its conventional forms.” They suggest that metaphor can be a powerful tool for finding gaps and expanding cracks in violent structures, and that metaphor works in concert with not in lieu of material practices, to restructure physical systems. “Through working with metaphor, we generate a series of speculative fictions for CS––Careful Slugs stories that reconfigure relations and introduce scandals and chances into Computer Science, Design, and STS processes that might lead to other sorts of project plans, research questions, and technical inquiries”(Britton and Pritchard 2022).

Refusal & material resistance

“Make technology ridiculous.” >–Nam June Paik, We Are in Open Circuits

Handcraft can emphasize the “NO” in technology as a resistance or refusal to operate by the terms and conditions of technology as it has been handed down: inaccessible, obtuse, large-scale, platformatized, proprietary, profit-seeking, and so on. Technological tools need not be the purview of only the narrow band of [users/creators/?] with the most access, and ‘craftiness’ is both an intellectual and a material strategy in response. Although craft approaches are also applied in other contexts, when they are used as methods for refusal they also can become forms of “tactical media.”

Tactical media are works that resist, revise, reverse, reimagine their materials to engage political potentials. As “an investment in a multiplicity of actions, practices, performances, and interventions,” says Rita Raley, “tactical media activities provide models of opposition rather than revolution and aim to undermine a system that, as de Certeau reminds us, ‘itself remains intact’.” The technological gestures of tactical media artworks operate in the world as it is now, while reflecting upon, resisting, and offering alternatives to it. Raley says that tactica media is a “mutable category” that includes “practices such as reverse engineering, hacktivism, […] the digital hijack, […] collaborative software, and open-access technology labs […]” (Raley 2009). In its many forms, tactical media [(like craft)] consists not only of its material objects, but also its communities, networks, protocols, and processes of disturbance and resistance.

Crafters and other artists know that resistance can emerge from the properties of the materials themselves, and that creation is a cooperative effort between maker and material. Digital humanities researcher Bethany Nowviskie argues that this resistance is a reminder of “the material nature of every generative or transformative textual process” most useful when makers have deep knowledge of their environment, including the time and access allows them to make and refine their own tools: “the material nature of every generative or transformative textual process” (Nowviskie 2013).

“Pranking an AI — giving it a task and watching it fail — is a great way to learn about it” (Shane 2021)

Let’s briefly discuss some techniques employed in craftiness as refusal and resistance: adversarial use, extreme use, handcraft practices, and esoteric/DIY systems. These techniques often overlap, reflect one another, or are used in tandem. Of course they are not the only techniques, but a few I have spotted in the wild and have used myself.

Adversarial use uses technology against the grain or against the goals of its creators. In an obfuscation technique called ‘baysean flooding’, machine-reading platforms are prevented from interpreting data by mixing in an overwhelming amount of false data which cannot be distinguished from ‘true’ data — such as posting a thousand nonsense social media posts with false locations for every accurate one (Brunton and Nissenbaum 2015). In “ladymouth”, I created a bot to participate in misogynist online forums in order to post feminist texts, then collected the responses from forum participants and performed these responses at poetry readings. Both are examples of adversarial use (Ciston 2019). Other forms might also include desire lines, the non-designated paths in parks and natural spaces that are created by human or animal use.

Extreme use, also known as breaking, looks to what is already broken, concerning, or confusing in existing systems and plunges into those gulfs. By highlighting (or causing) errors and glitches, these works step into the trouble. They hyperbolize aspects of a system in order to attend to the problems this exposes. In her “Manifesto of the Broken Machine,” Sarah Sharma shows how we might work with outliers and errors — and with perceptions of women and others as outliers and errors — in order to work against the systems that frame them as such:

“the idea of our contemporary social-political-economic system as an already-broken machine full of the incompatibly queer, raced, classed, and sexed broken-down machines is politically exciting for feminism. […] A feminism of the Broken Machine focuses on and uses the logic of the machines to highlight current power dynamics that are otherwise hard to pinpoint. The Broken Machine uncovers and incites new power moves.” (Sharma 2020)

In “Google Maps Hack (Traffic Jam)”, Simon Weckert walked through the city pulling a handcart full of mobile phones running Google Maps in order to cause the app to display traffic jams where there were not any (Weckert, n.d.). Sadie Benning created their short films like “Girl Power” on a camera originally produced by Fisher Price as a toy, which made videos on an audio cassette. Benning’s films pushed the limited form to the brink, documenting their experience as a 1990s teenage lesbian while sampling a wide range of music, other video, and text on its glitchy black and white surface, all from inside their bedroom (Benning, Sadie 2023). In “innervoiceover”, I chained together speech-to-text and text-to-speech tools in a participatory gallery installation, with the understanding that this would amplify the errors, because I wanted to make visible these processes as well as to explore the potential for poetic effect (Ciston 2022).

Handcraft practices may involve craft tools or may involve handcraft approaches taken up with high-tech materials, or both! In “The Internet of Towels — Knotty Articulations”, Anuradha Reddy used the process of crocheting a towel to study lenticular images and the production of QR codes. The result was a 3D-textured towel which, only when viewed from an angle, can be read as a valid QR code by a QR reader or phone (Reddy 2021). Elisa Giardina Papa documented her work as a human trainer for AI systems in Leaking Subjects & Bounding Boxes: On Training AI, collecting images and instructions from some of the thousands of tasks she was assigned “which seemed to resist AI’s orderly impulse.” By turning this labor-intensive work into an art object, the book asks its readers to reconsider the work of classification as a handcraft (“EGP” n.d.). In “No Knots, Only Loops”, I documented the months I took to produce a crochet sculpture. The work offers a tangible means to grappled with the uncanny scale of large language models, by comparing the number of stitches in a huge sculpture to the number of parameters in a model (finding them still insufficient to “model” the model).

Esoteric systems, or DIY systems, are often a response from artists or communities when they cannot find the tools or resources they need in existing forms or formats. Creating their own alternatives in response to the lack they perceive, artists’ esoteric systems can be inventive and generative, investigative and critical, or justice-seeking and recooperative, among other things. In “Coem”, Katherine Yang created an esoteric programming language that “imagines poetry as purposeful and code as emotional,” working against general-purpose languages that prioritize efficiency over feeling. Developing a language with different core values demonstrates how value systems drive code design and inform the functions a language allows as well as what operations and outputs it can produce (Yang, n.d.). In “ImageNet Roulette”, Kate Crawford and Trevor Paglan created an interface for exploring a popular dataset used in many machine learning foundation models, in order to reveal the dataset’s problematic taxonomies that classified people as, for example, “slattern, slut, slovenly woman, trollop” or “failure, loser, non-starter, unsuccessful person” (Crawford and Paglen 2019). By building a system that let public audiences explore the dataset in an accessible way, they could bring personal understanding to the issues with how such datasets are created and used in many kinds of machine learning tasks. In my upcoming research project “Unsupervised Pleasures”, developed in collaboration with Emily Martinez, we are developing alternative approaches to conscientious dataset development based on intersectional principles and practices, in hopes that these shift the outputs when applied to language models (and as a provocation for imagining different model architectures as well).

These forms of refusal replace what they reject with alterative visions for more liberatory technologies. Creative and social computing scholar and activist Dan McQuillan argues current AI systems exacerbates existing aspects of society that act violently to exclude and enclose. He calls for an anti-facist approach to AI that can [enact/enforce/instantiate] “a positive refusal, a rejection of certain forms of apparatus and a commitment to radical alternatives. […] Our ambition should stretch beyond the timid idea of AI governance, which accepts what we’re already being subjected to, and instead look to transform our apparatuses into a technical practice that supports the common good” (McQuillan 2022). We can start with looking at how existing models, methods, and methodologies indicate the changes that need to be made, not in the models alone but in the world, argues digital democracy scholar Wendy H. K. Chun: “Machine learning and predictive models as they currently exist can also resist reduction, but only if we treat the gaps between their results and our realties as spaces for political action, not errors to be fixed. [We need to treat these models as we do global climate change models. GCC models offer us the most probable future, given past actions, not so that we accept that future, but so we work to change it. Only global climate change deniers seek to fix the model, rather than the world.]” (Chun 2021). As Chun and McQuillan suggest, we need more than the mere governance of emerging technologies as they are currently conceived, and instead require moves towards prefigurative practices. Crafty tactical media can queer the use of machine learning systems and other emergent technologies in order to reveal their classificatory, reductionist logics. By applying techniques like aversarial use, extreme use, handcraft practices, and esoteric systems, a craft lens helps imagine and enact “a committment to radical alternatives.” As trans*formative systems suggests, such alternatives already exist, if we are willing to write, build, create, and cooperate in ways that reveal them and allow them to flourish.

“Embracing the nuances of crafted art forms […] exemplifies the non-linear, embodied, diffractive ways of reading, interpreting, and understanding the world while questioning what AI tools can do and the extent to which craftspeople have a say in what AI tools should be doing instead. I contend that AI does not produce newness but rather ‘newly’ suggests how craftspeople have always been creative agents for shaping the future of culturally and visually-informed algorithmic systems.” (Reddy 2023)

A “lace card” was used in early punch-card computers to jam the systems.

Transformers

Transformers are a current type of machine learning architecture being employed to create large models like ChatGPT and Stable Diffusion.For more explanation of terms, see “A Critical Field Guide for Working with Machine Learning Datasets” and “Intersectional AI Toolkit.”

They are good at digesting large clumps of text, images, video — all interpreted as numbers — and reconstituting them. Transformers main parts are called encoders and decoders, and it’s often said that even their designers don’t know what exactly is going on in these hidden layers. I delve into the specifics of that ‘unknowing’ elsewhere. I find it more interesting that the transformers themselves do not know what is going on in these hidden layers of the own architecture — despite all the anthropomorphizing of AI systems, with widespread use of terms like “learning” and “understanding.” While they process and produce meaning (input and output), transformers are never “aware” of the meaning of the content they are processing along the way. They are just crunching numbers.

Transformers are the latest, but won’t be the last, in a long string of systems that ask data subjects and data subjecteesFor more explanation of terms, see “A Critical Field Guide for Working with Machine Learning Datasets” and “Intersectional AI Toolkit.”

to submit to their “superior” logics. With every iteration on their forms, we repeat their hype but lose trace of their histories in “race science” and colonizing power that captures what it means to know and be known (Browne 2015; Benjamin 2019; Joque 2022).

The paper that launched a thousand transformer models was called “Attention Is All You Need.” Transformer language models incorporate text context through a mechanism called “self-attention” that examines a selection of word-units surrounding each word-unit they interpret. (Vaswani et al. 2023) As they proceed, they use this interpretation to predict the next most likely word to appear in a sentence, based on the words that have commonly appeared near the last few words before.

But we need more than self-attention. We need self-reflection and co-regulation. We need systems that, in their very architectures, celebrate difference - rather than erase it.

To move from the “transformer” model architecture to trans*formative systems means to acknowledge the requisite entanglement of community building and [critical sociotechnical systems]. It means knowing the material we are transforming — by sharing in, speaking nearby, walking beside, being constituted and supported by, and supporting in turn.

Trans*

Trans*-nessA note about terminology & Trans*: It is essential to note that the concepts of ‘trans*’ and ‘transgender’ have their own varied and contested histories, entangled with colonial histories, class and racialized contexts, politics and personal stakes for many. These complexities[^tsq] are charted by trans* studies and by many works beyond critical theory. They are dynamic, continually evolving, and I hope to honor this complexity while acknowledging I cannot address it comprehensively here. In her discussion of the complexly situated histories of the term ‘transgender’, professor of history and gender and sexuality studies Marta V. Vicente translates and summarizes psychology professor Manuel Roberto Escobar’s ethnography of subjects who use a wide range of labels and while also thinking of themselves as ’trans*:

“The trans* body is the result of a specific and historically based crossing of ‘regimes of knowledge-power’ (Escobar 2016). It is ‘a baroque construction,’ displaying multiple sides, socially translated by a diversity of languages. Defining oneself with words such as marica (fag), puta (whore), and gai (gay) and still thinking themselves as trans*, Escobar’s subjects express a definition of the self that tries to maneuver through these ‘regimes of knowledge-power’ to establish their own space of resistance.(Vicente 2021)

Trans* in its many contested forms continually grapples with this work of trying to establish self(s) in the context of “regimes of knowledge-power” and to establish cooperative spaces of resistance, which is more than ever required in this moment of computational and categorical logics. We can look to long histories of trans* communities resistance for strategies and tactics. In We Want It All: An Anthology of Radical Trans Poetics, editors Andrea Abi-Karam and Kay Gabriel declare, “Against the common-sense intuition that crisis means we must demand less, we assert with our commrades that everything has to change for anything to continue.” (Abi-Karam and Gabriel 2020).

can be fluid, in motion, evolving, and emergent. It can also be euphoric, inclusive, embodied, self-identifying, and outside binaries. Coding.Care imagines systems that might also have these qualities, that can trans*cend the logics constructed by capitalist, militaristic, racist, ableist, misogynistic lineages. It is inspired by decades of trans*feminist and crip technoscience work, discussed more below.

This approach to trans*formative technological practices has no prerequisites. It does not require glowups or dead names, no traumatic ‘before’ in order to desire or access better ‘afters’. We are already queer-enough, coder-enough, worthwhile-enough to participate in continually becoming ourselves. We are “always-already programming” (Hoff 2022). By adopting this ‘queer enough’ approach from the LGBTQIA+ community as an ethos for critical–creative coding spaces and for a context in which to examine sociotechnical systems, I have found that it is possible to welcome a wider range of voices into conversations about the high stakes of tech. It creates more comfortable, adaptable tech-focused communities for LGBTQIA+ members and for everyone else who joins. Technology spaces have much to learn from queer community building, just as they do from Black and brown communities, yet I am also mindful that their community building often comes out of necessity for protection and safety, that it is always a tension between welcoming and insulating.

[XXX][discussion of radical difference and radical belonging] [xxx][keep queer and trans sections separate?]

“Black Computational Thought holds open these proximal possibilities and directs our attention to the quotidian, social, opaque, woven, and fugitive practices of computation born from Black diasporic movement. Blackness here is always in an appositional love affair with queerness, transness, and feminist orientations of the world that fight the simplicity of calculated separability, and instead contend with their relationship to difference, kin, collectivity, risk, non-normativity, non-linearity, ambiguity, and trust.” (Morrison 2022) –>

“queerness is itself inherent within [the history of] computational logic, […yet] there exists a structuring logic to computational systems that, while nearly totalizing, does not account for all forms of knowledge, and which excludes certain acts, behaviors, and modes of being.” (Gaboury 2013)

I also use the terms ‘queer’ and ‘trans’ here a lot together, and I have mixed in a kind of queer politics under the section ‘Trans*’. I am keenly aware that everyone has a different take, of course, and I’m wary of wielding such weighty words unwisely. For me, queer is the term that includes and elides all the categories of LGBTQIA+, and that’s why I like it. Queer shows you its politics and shares nothing else.A note about terminology & me: Queer is the only term I have found that comes close to “fiting” what I’ve felt about myself when nothing else felt quite right. This quotation gets me:

“For me, ‘queer’ is not a name so much as a protest against complacency when naming. It records a discomfort with prevailing languages for embodiment. It is not another identity so much as a steady refusal of identification—of any settled facts of life that we are supposed (already) to know. […] No queer names without queer naming, which cannot fit easily into our usual stories.” (Jordan 2023)

Up to now I have not used the term ‘trans*’ to self-identify, even if in some ways it may be correct. By way of talking about the word ‘trans*’ in relationship to trans*formative systems, I will try to tease out my relationship to the term. I think I’ve been avoiding my own stakes in ‘trans*’. I was concerned my stakes were false, or would appear false, but it’s actually that I didn’t have the language for myself. Because there isn’t language for wanting to be without language, for not wanting to be seen or described or labeled — to have your attributes held against you in an alley or an email. Or maybe they they already found a label for not wanting a label, and that language is just ‘genderqueer’ or ‘agender’ and it still doesn’t sit right. The only thing I like about this complex ever-turbid gender taxonomy is that it doesn’t sit still. I hate this business of taxonomies anyway and its categorizing logic is how we got to machine learning and many worse things in the first place. This could be the core of the question, if I weren’t still feeling for the question.

I wasn’t using ‘trans*’ for myself (though now I’ve gone and claimed it for my book, wtf) because it didn’t sit right either when I worried other people need it more, needed the term to survive, and I had been surviving without it — as if language is a finite natural resource, like rare earth minerals we’re extracting to make models of language. I used to worry the same thing about the word ‘queer’ before I practiced letting myself try it on, just a little (as a treat). That discomfort grew, then eventually transformed into some kind of resilience akin to pride. Still, I feel I am an ungrateful thing, whenever I reject the privileges I have when I go passing as a straight cis woman who is none of those things, those privileges tangled up with the erasures it brings, the over and over of nixing a self. It feels worse to correct someone because I don’t have a true label to give. Queer and trans*formative are the only words that feel good enough.

I wasn’t using trans* for myself, or nonbinary or agender or genderqueer, and I wasn’t using gay or lesbian or bi or pan or any other term for myself, because I worried these were all labels for people who knew who they were and had always known, who stood so far outside the identity they were called that they required this word as a reckoning. For so long I have been completely out of contact with my body and its shapes and desires. The middle space with no reckonings was safe. I had been just reading the room for cues to how I was supposed to feel, and that small rural version of me was the only reality there was. But I am writing for all the people who haven’t known who they are, who haven’t learned the tools the easy way, wjp haven’t fit into the terms or the counterterms, who need permission to play in the spaces between like I have needed — whether they play with gender or with code or both.

To identify with new terms (or new tools) is to make bold claims. My life would need to change for claims like these. I’d need to put my skin in the game, to admit my skin is already in the game, to admit I have skin. I’ve spent so long disavowing the body below my neck as it aches for me to acknowledge it. This “I” of diffuse particulates is not even atoms swirling, because the atomic model of the universe is a fiction devised by object-oriented bros who still think they can be islands.

I spent so long as a decapitated head —
a magician’s assistant trying to
keep the show going trying
to please, trying to keep
the body I couldn’t feel
safe.

Resetting is a recursive process. It is not just a trick of pulling out the tablecloth from beneath the finely laid placesettings and the severed head.

Trans* is not transition, not as simple as movement from one thing to another. Trans* is not transformation; trans* breaks open transformation with its asterisk and its radical understandings of form/ative (both discussed further below). Trans* is much more expansive. It is unrestrictedly being who you have always been, as fully as you are, more than anyone has previously recognized, and experiencing that recognition as fully as possible.

This ongoing process can be aided by finding the language (or losing the language), or finding the tools (or breaking the tools), or finding the commrades (or making the spaces where they might find you). This text is inclusive of many understandings of trans*-ness and transformation, combining them in order to suggest the kind of trans*formative practices of being and/or making that are prefigurative — bringing a world into being as it always could have been and should be, might yet be still, alighting the revolutionary force of latent potentials that already are.

It’s a subtle, vulnerable, powerful thing, just to recognize oneself more fully — especially if it makes space for others to do the same. I find that coding communities which are founded based on practices of queerness and transness (its ethics, tactics, ways of being) can open those communities as spaces of radical belonging.

Trans* is endlessly supple and takes on many suffixes. It finds queer reverberations in the trans*ducers all around us that constitute and reinforce computational systems. A transducer (from Latin for ‘lead across’) converts physical signals into electrical signals, or vice versa. Transducers are intermediaries that allow the trans*fer between analogue to digital. Every time information gets trans*mitted from an environmental sensor to a digital device — sound waves into audio files, optical variation into heart rate — trans*duction has occurred, and thus some kind of transformation. Transduction shifts information’s register or level, reimagining its forms. Transduction is always intervention, informed by the device designers’ decisions as well as both contemporary and historical approaches to computational knowing. Note how many times the language of trans*-ness appears in sites of technological knowing. The material conversions required to power the logics of computation (such as converting qualities into voltages into quantities) all pass through gates of trans* thinking.

* (asterisk)

The * after trans longs for anything that follows; it opens up space for the layered and undefinable becoming, and as such can include anyone who wonders if they belong, who wonders what they contribute, who wonders what “this whole thing” is about (be it gender, sexuality, technology, whatever).

Combined with trans, or at large, the asterisk is the interstitial. It is a footnote that defines our terms, and also it is the invitation to extend beyond any definition given.

The asterisk is a perturbation, a pebble in the shoe. In the German language, the gendering of nouns and verbs is haphazardly addressed by an asterisk that breaks the word, introduces a glottal stop between its ‘masculine’ and ‘feminine’ formations. Student or Studentin becomes Student*in; Bäcker or Bäckerin becomes Bäcker*in. This brief interruption can be read as a faulty solution that maintains gender-as-binary, it’s true. And also, this hesitation adds a frequent awkwardness that highlights how often conversation is about gender unnecessarily. It physically opens up a space (in the mouth) where gender is unknown/unstable/all.

A French or English apostrophe may mark an omission of letters, but this elision also joins together, suturing up the distance it creates. The asterisk shares this tension here, as a clot that heals, an unspeakable but audible gasp.

Like the apostrophe at the start of 'pataphysics, which gestures to the absurdity from which it springs, the asterisk is the instigator and the multiplier, the pivot and the punctum and the portal.

It is not the data point, reduced to numerical expressivelessness. It is not the node in the network, defined by its links and relations, abstracted away. It’s not a redaction which the asterisk marks, but the possibilities of unknown language underneath. It’s the tension of knowing and not saying, as in the marking of sw*@r words. It’s once again queer as a politics, “queer as in fuck you.”

The asterisk is risk (just a step), is starry-eyed, A-star pathfinding, ASCII joy, a wry smile, and a mark of possibility on a treasure map. In mathematics the asterisk is duality, mirrors mirroring each other, points connecting to make lines and lines crossing to make points, weak-star topologiesThanks to Miller Puckette for expansive discussions about weak-star topologies, splines, the asterisk, and Claude Shannon.

and infinite vector space. This returns us to the word vectors and latent spaces of transformers.Just after writing on asterisks, Open AI announced Q*, promising again to accelerate machine learning before addressing its current concerns (Milmo and editor 2023). At the same moment, “The Gospel” AI system is used in Israel to select and increase its bombing targets by orders of magnitude (Davies, McKernan, and Sabbagh 2023), and new stories continue to reveal details of the depth of AI usage there and involvement from the big tech firms (Iraqi 2024). AI technology was originally being developed alongside and as part of WWII and Cold War technologies, now AI is used directly in warfare. The same companies warning of the future dangers of AI make this their marketing strategy: GPT-2 was open-source but ‘too big and too dangerous to release’ until it appeared as an even bigger GPT-3, then GPT-4 was closed and pay-to-play. Worse yet, those so-called dangers distract from the very-real, immediate dangers they impose, like The Gospel, face recognition in policing, and more subtle forms.

Formative

What is formative shapes us. Our early experiences make us who we are. These are the foundations we can’t see, the scaffolding of our thought, the infrastructures of our being. I am interested in the formative aspects of sociotechncial systems, produced in entangled loops of human networks, codified language, and rare earth. How we become. How we become with and through sociotechnical systems.

What are the fundamental assumptions that go into the design and implementation of a dataset, a machine learning system, or programmers themselves? What are the formative assumptions underlying any technology or way of being? These works tease out the metaphors we have relied on so long we can no longer see them; they ask what other metaphors might make space for new forms or reforms.

Form in-forms content. The shape of the container decides what can fill it. I work in code and in art because I believe: If I know how to make forms, or even platforms, not only the content to plug into existing forms, I am more free to work both form and content in iterative, ongoing relationships. This is how I knead the dough — form and content and form and content and form. They are inseparable.

In this way, the material I work with also in-forms what I make, through its insistence, resistence, grain, thread tension. It in-forms itself into content as I try to bring forth some kind of form. It also re-forms me through this mutual process:

“In both carpentry and textiles, the form of a thing does not stand over it or lie behind it but emerges from this mutual shaping, within a gathering of forces, both tensile and frictional, established through the engagement of the practitioner with materials that have their own inclinations and vitality.” (Ingold 2015)

Form activates. No form is neutral, although if we use it long enough we begin to look through it, like the blank page, or the MS Word ‘blank page’. Each interface or material enables or encourages particular experiences and outcomes, and denies or deflects others — whether through predictive text in search, syntax highlighting in code, or the constraint of an eight-page paper zine. Some forms invite us to collaborate in imagining the usefulness of their limits and the boundlessness in their constraints. Through time spent in process, working with and against form, our focus can shift from producing outputs that meet the brief of the form, to working with the form to reshape both the form and ourselves in new understandings and new orientations.

Yet current modes of computational logic are not mutually constitutive in trans*formative ways. Instead, they impose form on us as data subjects and subjectees. Artist–researcher Pedro Oliveira describes the way international borders use digital surveillance to prescribe identity: “the border listens for people in a way that matches them to categories they ‘should’ belong to (e.g. ‘German’, ‘Greek’, ‘Syrian). When this listening for fails to produce these categories, the border then listens to their bodies with the purpose of uncovering the ’truth’ that these border subjects might be concealing” (Oliveira n.d.). Computation requires information to take specific forms. Its orderly logic is a shape imposed by and reflecting power. It requires contortion to represent be represented. As media theorist Sarah Sharma reminds us, for the ‘Broken Machines’ who do not quite fit these regimes, “to represent is also to be filed away” (Sharma 2020).

[xxx][add transition, not filing away, opacity]

“In order to understand and thus accept you, I have to measure your solidity with the ideal scale providing me with grounds to make comparisons and, perhaps, judgments. I have to reduce. […] perhaps we need to bring an end to the very notion of a scale. Displace all reduction. Agree not merely to the right to difference but, carrying this further, agree also to the right to opacity that is [not enclosure within an impenetrable autarchy but] subsistence within an irreducible singularity. Opacities can coexist and converge, weaving fabrics. To understand these truly one must focus on the texture of the weave and not on the nature of its components. For the time being, perhaps, give up this oid obsession with discovering what lies at the bottom of natures. There would be something great and noble about initiating such a movement, referring not to Humanity but ta the exultant divergence of humanities. Thought of self and thought of other here become obsolete in their duality. Every Other is a citizen and no longer a barbarian. What is here is open, as much as this there. 1 would be incapable of projecting from one to the other. This-here is the weave, and it weaves no boundaries. The right to opacity would not establish autism; it would be the real foundation of Relation, in freedoms.” (Glissant 2009)

To understand through computation is to reduce to categorization’s logics (Crawford 2021; Amaro 2022). The formative structures we create (taxonomies, schemas, ways of knowing) support restrictive unknowing. They ease and smooth passage through our days and our systems, saying don’t look, you already understand. This is why we must imagine new forms — and this is why it is so hard to imagine new forms.

Code forms, then calcifies. Pattern turns to template. In the procedures of object-oriented programming languages, the template runs: init --> name --> get --> setCreating classes in object-oriented programming lets the programmer instantiate each instance of an object with a pre-determined form, adjustable only within those limits. These processes indicate what prior schemas you’ll pull from, what parts you’ll write over, what qualities you’ll predetermine. They indicate what aspects will be qualities of the entire class, and what qualities will be specific to each object. They determine what actions each ‘object’ will ‘know’ how to do, and how the programmer or user will establish and retreive information. There’s a language for all this in “object-oriented” programming, which requires a different sense of relationality than “functional” programming.

.

“The more a path is used the more a path is used. […] When an effort becomes normal, a form is acquired.” (Ahmed 2018, 2006)

If forms shape content, and vice versa, of course tools also shape forms. Equipping ourselves with the knowledge, resources, and agency to make the tools (to make the platforms to make the forms to make the content) is necessary for a more holistic approach to researching formative tech and creating trans*formative tech. Toolmaking includes collecting sets of tools, documenting and creating guides to existing tools as well. It also includes the infrastructures and support systems necessary to make any of these practices possible.

“tools and infrastructure are not only the preconditions for the work; at their best, they are also part and parcel of the work, deeply integrated into its methods and outcomes. Our tools and infrastructures are rich objects to think with.” (McPherson 2018)

“instruments develop through engaged and contingent practices. Instrumentalism involves setting in motion, operationalizing, and potentially transforming. Instruments—whether in the form of concepts or sensors—are instrumental to the unfolding, the doing, and the transforming” (Gabrys 2019)

We think with our tools, instruments, sensors and the surrounding systems that enable them and maintain them. All of these layers are entangled. Susan Leigh Star calls infrastructure a “fundamentally relational concept,” but one that despite its embeddedness operates invisibly unless broken. She also argues that infrastructure marks membership in relational groups, as we “acquire a naturalized familiarity with its objects,” practices, and standards (1999). Thus infrastructure — and the tools, platforms, forms, and content that it consists of — also marks us as belonging, knowing, capable, welcome, or not.

How do we think critically through re/making infrastructures, when they exist beyond any one of us, always relational and often invisible? By returning to, and reattuning to, the solidarities we find within and across communities, and the frictions we find within and across materials.

Those who are left unsupported by infrastructures, those who are left out of representations or who are targeted disproportionately by digital systems, already know in deeply embodied ways about the friction they feel as they move through systems that do not easily contain their forms. Scholars Aimi Hamraie and Kelly Fritsche, authors of the “Crip Technoscience Manifesto” (2019), call these infrastructures “frictioned technologies” and emphasize:

“the skills, wisdom, resources, and hacks disabled people utilize for navigating and altering inaccessible worlds. In pushing crip technoscience as a field of research and a practice of critical ‘knowing-making’, we conjure frictional practices of access production, acknowledging that science and technology can be used to both produce and dismantle injustice.” (2019)

Their manifesto reminds us that disabled people are not minoritized users of technologies created by experts but instead are experts themselves at actively adapting technologies to their needs. By centering their community as knowledge holders and creators, and by seeing “interdependence as a political technology,” they shift the design goals for technology. Rather than outsiders providing services to achieve normalized individual independence, they reorient toward kinship building (2019).

Louise Hickman’s ethnographic work on the ethics of access foregrounds access workers’ key role in the production of crip technoscientific knowledge. She examines the highly customized dictionaries developed by her own Communication Access Real-Time Translation operator working in academic spaces, in order to show how their captioning work for the d/Deaf and hard of hearing community is “both the practice of crip technoscience as well as its product.” Hickman highlights the collaborative, embedded nature of the work and the crip technoscience emphasis on “knowledge-from-below.” (Hickman 2019)

Thus, form is always forming from within complex systems of interrelation. Formative and transformative technologies are more (and less) than their forms. They must emerge on our own terms — from the formats and methods of the communities they serve, with our values, goals, and approaches leading the way.

Interdependence has been my intuitive compass, my frustration and fascination, for as long as I remember. I have searched for the means to describe this feeling, through art, literature, touch, and other means. Slowly and repeatedly, I keep finding that the best resources for describing this sense of interdependence come from Indigenous sources. I keep returning to this idea:

Interdependence is Indigenous knowledge.

“an object or thing is not as important as one’s relationship to it. […] reality is relationships or sets of relationships.” (Wilson 2008)

Fundamentally reorienting our understanding of objects and knowledge and ourselves as consituted by our relationships necessarily also reorients our work and the forms it takes, in deeply formative ways. Re-prioritizing relationships fundamentally reshapes the forms and functions of research, artmaking, valuemaking, worldmaking. To see form (shape, object) as relation (connection) is to position it in time and place. It is also to take “form” from its noun form (shape, object) into its verb form (mold, press, begin to exist). As Erin Manning says of the spiral, it is “more duration than form” (Manning 2013). Like all verbs, then, the verb form implies subjects — the entities enacting and impressing a form into being, responsible for it, caring for it. Relationships. This negates object-oriented ontologies and object-oriented software programs that want to encapsulate and abstract away objects’ entanglements with the world.

“I speak of the verb process, the doing, the coming into being, the at-the-time-of. Which is why we think there is particular value in live music, contemplating the artifact as it arrives, listening to it emerge. There it is. And There.” (Baraka 1966)

“There can, of course, be no knots without the performance of knotting: we should therefore commence with the verb ‘to knot’ and view knotting as an activity of which ‘knots’ are the emergent outcomes. Thus conceived, knotting is about how contrary forces of tension and friction, as in pulling tight, are generative of new forms. And it is about how forms are held in place within such a force-field or, in short, about ‘making things stick’. Accordingly, our focus should be on forces and materials rather than form and content.” (Ingold 2015)

To frame research through relationships is also to foreground accountability by asking, as Shawn Wilson offers in Research Is Ceremony, “What am I contributing or giving back to the relationship? Is the sharing, growth and learning that is taking place reciprocal?” Wilson says that in an Indigenous Research Paradigm, “What is more important and meaningful is fulfilling a role and obligations in the research relationship—that is, being accountable to your relations” (2008).

In this sense of reciprocity, the question of form takes shape as an offering. What does our creative work offer back, and in what form will that offering be most legible and accessible? Some forms I have tried: manifesto, toolkit, zine, prototype, platform, black box, machine, labyrinth, server, service, sentence, story, interface, guide, document, cauldron, process.

“Form is simply how a thing exisits (or what it exists as). […] Content is why a thing exists. […] but they are not separable in any object.” (Baraka 1966)

As form moves into its verb shapes, identity too moves into practice. Artist Johanna Hedva marks this as a pivot from identity politics to practices and methods, inextricably coupled as theory-practice:

“Disability, queerness, open source — not as identities, or groups I belong to, but as modes of doing, of how I practice myself. Being an outsider means that the question of theory and practice — how practice is affected by theory, how theory is constructed by practice — becomes the most important one. Membership to particular groups and experiences is often predicated on the visual — whether someone ‘looks like’ they belong or not — which means that my membership to most groups must rely on something else. My belonging has more to do with how I enact that group’s politic” (Hedva 2018)

Our being-in-process is both the means to “be” (prefiguratively) and also the means of being-in-relation, -in-politic, -in-identity, in-between.

“this form is really just the web of relationships that have taken on a familiar shape. Every individual thing that you see around you is really just a huge knot-a point where thousands and millions of relationships come together.” (Wilson 2008)

“a node of relation expressing itself momentarily as this or that—an edging into object, a swerving into body.” (Manning 2013)

I dream of a form that can hold everything. A form that can reveal how each part is connected. No parts, no compartments. Just connections. Network graphs don’t do it for me. Anna Munster suggests that the form of networks numbs us to their relationality. Because every network looks like every other, in trying to understand a system by depicting it as a network, we lose sense of what’s within. Munster suggests focusing less on forms and more on forces: “the closures and openings of relations to one another. It is at this level of imperceptible flux — of things unforming and reforming relationally — that we discover the real experience of networks” (Munster 2013). In the un/forming of relationality, these trans*formative practices are active, ongoing, prefigurative, emergent, embodied, and embedded — taking shape and making change.

“Relation is open totality; totality would be relation at rest. […] Relation is movement.” (Glissant 2009)

Just as a paper snowflake or a tie-dyed garment contains in its pattern the re-foldable traces of its making — a flattened, temporarily static version of its coming-into-shape, marking the places it once touched against itself as it held different forms (Sedgwick 2012) — crafting trans*formative systems aims to “decipher the history of the making” and the future shapes they might take.

I want forms that can decode these histories, plant these futures, and embrace them. I dream of a platform where I can write and code and share and cite and annotate and highlight and collage and print and remix and machine-read and connect. I don’t want the innards and innerworkings hidden from me; I want them explained so I can tinker with them. I want to understand how they relate to and shape what I am making.

“When writing, the code ‘material’ speaks back to the programmer through various kinds of testing […]. The process of testing often results in surprises, and in changes to the original plan” (Blackwell 2024).

“We shouldn’t make systems too automatic, so that the action always goes on behind the scenes; we ought to give the programmer-user a chance to direct his creativity into useful channels.” (Knuth 1974)

I see all coding as writing and all writing as coding. I want systems that expose these entanglements and invite us into their complications with a helping hand, instead of hiding and encapsulating their operations in ways that limit and name what we can make, create, be in advance. I want to admit there is power in the naming and organizing of the tools I am using, and I want to participate, mindfully, in that naming. It is also a part of the worldbuilding.

“An essential element of coding is deciding what to call things. […] Inventing a name for something is an exercise in philosophical abstraction. Two or three words might summarise hundreds of lines of code. When chosen well, this name becomes the definition of what that code should do.” (Blackwell 2024)

“even to name something, is to wait for it in the place you think it will pass” (Baraka 1966).

I want to be able to follow those traces of power through their systems, to be able to see the scaffolding and build different systems.

[does techno go here?][What about/How did we get the systems we have?]

How do we transform these into other kinds of systems we want?

Craft and radical belonging

A systems perspective an intersectional perspective

Technological systems, algorithmic systems, human systems, global systems.

[Nearbyness — get near instead of at/claim the thing….]