Scale
Caroline Sinders
-
Could Libretaxi ever entirely replace Uber? Should it? Thinking through building big tech alternatives, I’m interested in seeing where our alternatives could fail at scale. By alternatives, I’m looking at data unions, open source software, cooperatives, art projects, collectives, and public infrastructure tech. By building these alternatives to function as something ‘better’, can equity (within these alternatives) scale? What do ethical, data structures, and tech cooperatives need; how would they need to be built;how would they function; and how or where would their ‘ethics’ break as alternatives ‘scale up’ to replace big tech?
This research is in response to unpacking ‘ethical technology’. I recognize that aiming to design for ‘ethical’ technology is inherently a positive idea, but what is ethical in practice? I’m interested in the space of where intentions do not align with actuality or impact, and the kinds of harms that creates. From there, by situating the realities and lessons from pre-existing projects, this can be used to leverage, influence and inform technology practices, software, and product design.
Within my Excavations residency, I’ve been exploring: what does ‘ethical’ mean structurally for data cooperatives and non-profit tech companies? How would they be built, how would they function and how or where would their ‘ethics’ break when the company or software scales up and down?
-
On Scale: Can Scale Ever be Equitable and Refuse it's Current Identity Bestowed On it By the US's Silicon Valley?"Design can play a role- it's not just about shaping materials, but investigating and guiding what happens before that moment and from that point onwards. It can be a tool to limit and heal the damage caused by its needs, rather than a mechanism to create new desires and immediately frustrate them." - Formafantasma from Ore Streams, Visual Essay
My residency focused on tech alternatives. When building big tech alternatives, when do or could these alternatives fail and what is a part of that failure? Is it scale, governance, lack of funding, maintenance, etc (or all of the above- it could and probably is, all of the above). 'Alternatives' here refers to data unions, open source software, browser or search engine alternatives (using open source or a more 'open' ethos), text message platform alternatives (such as Signal), cooperatives, art projects, collectives, and public infrastructure tech. Often the discourse around these alternatives is positivist (which is not entirely a bad thing) and these alternatives are painted as something 'better' than the current big tech paradigm. This essay, and my project, is not here to defend big tech (not at all), but I believe that criticism and analysis of these alternatives is paramount to make them more equitable, usable, and useful. I don't think we can dismantle the master's house with the master's tools. I do believe we need to build alternatives, ones that center care and communities and focus on harm reduction. With that stated, I'm interested in exploring this question: can equity (within these alternatives) scale? What do ethical, data structures and tech cooperatives need; how would they need to be built; how would they function; and how or where would their 'ethics' break as alternatives 'scale up' to replace big tech?
By conducting this research, I hope use cases and findings can be used to inform current alternatives, from research, building, designing, community management and maintenance.
My Excavations residency builds on on-going research that I've been exploring through a 2019 Mozilla Foundation fellowship, and recent speculative design and research work conducted for Consumer Reports on the future of data unions and data cooperatives. My Mozilla Foundation fellowship allowed me to build alternative systems and tools for data training and labeling, called TRK (which stands for Technically Responsible Knowledge) and is a part of the Feminist Data Set project. TRK focused on the systemic issues around labor, invisibility, and creation of maintaining data structures and training data algorithmic models. I interviewed research labs, artists, startups who use Mechanical Turk style services, and microservice workers across Crowdflowr, Fiverr, and Mechanical Turk. I even became a Mechanical Turker myself.
Critical design and making are necessary research tools. To me, Feminist Dat Set is a critical design and process driven making project, interrogating every step of the machine learning pipeline from start to finish using intersectional feminism as an investigatory framework. Often the tools needed to make Feminist Data Set don't exist. For example, what is a feminist data training platform? Answering this question created the open source tool TRK. But now I want to scaffold on that idea, and look at scale; what would 'break' TRK and make the tool not feminist or unethical? From my research, I believe that's scale, capitalism, and governance. Scale can be the amount of people using that technology, art piece, or something created and maintained by the cooperative; or money made by that idea; or the amount of employees; or all of the above. Can technology, services and labor that are invisible but within technology (like Mechanical Turk) ever be a form of public infrastructure, and be inherently 'equitable' and what would that system need to exist, and how would it be designed? How do we preserve equity when building our alternatives to big technology and not recreate the harms of big tech? This is what I'm researching.
For six months during COVID19, I set out to explore these ideas for my Excavations residency. I'll be honest here, I was overly ambitious. COVID19 has brought a particular kind of gaping burnout and anxiety, where projects I could have completed in a pre-pandemic time take twice as long. With that stated, and out of the way, it's additionally important to acknowledge that this is a topic I'm still researching, and exploring, even if technically the residency has ended. My research question has been gnawing at me since 2019 wherein these questions popped up from my work within my Mozilla Foundation Senior Fellowship and work surfaced in my grant from the Sloan Foundation and Ford Foundation's Digital Infrastructure grant. I thought my Excavations residency would create some clear paths forward, and in a way, be tied up in a neat but comprehensive essay and website. But, like all good research questions, it created more questions than answers.
In reality, the outcome of the residency is this: before I get to mapping or thinking about a public infrastructure data cooperative or a data alternative, based in a cooperative or non profit, or a more equitable tech alternative, building a robust foundation on scale, maintenance and sustainability is key, along with more knowledge on the frictions facing public institutions that are knowledge centers, like libraries and open source communities.
Why Libraries?
Libraries and librarians have been referenced as institutions to model on and after within the Wikimedia Community (an educational site, and a community along with a nonprofit), and frequently referenced models and experts to emulate to improve technology from technology researchers and think tanks, like New Public and Harvard Kennedy's Shorenstein Center on Media, Politics and Public Policy. But public infrastructure, especially libraries, are not linchpins to our future successes, they face their own issues, struggles and frictions. Specifically, my interest is also in the vocational awe that North America, in particular, anoints to librarians and libraries, without necessarily recognizing the pain points and problems librarians and libraries face. Vocational awe also plagues open source maintainers, civil society and technology critics as well. Some of the same issues that plague libraries and librarians also plague current open source projects, however, I would argue that libraries are even more stretched thin. Librarians and libraries have expectations to extend emotional labor, safety and support, such as Narcon training to librarians (which is a net-good) but without necessarily creating infrastructure or safety or more support for librarians and libraries. This vocational awe seeps into technology criticism and public infrastructure criticism, and it's this vocational awe I believe can exist as a blanket covering deeper systemic issues, and will still create more problems as opposed to solutions. I'm thinking of this example just as an analogy and related example within my question→ if we aim to build public infrastructure alternatives to technology (and we need that), how can we avoid replicating or creating harms (that we also see in big tech but also new harms) and is scale a part of this issue? A thousand libraries or librarians may not solve inequality within technology, and I believe it's important to recognize pain points along with benefits when we build alternatives (so we can mitigate, fix, adjust those pain points).
As an artist, I'm interested in taking this provocation to it's specific conclusions. If experts are saying or desiring public infrastructure, and for more knowledge creators and knowledge maintainers, like librarians, to be involved in digital communities, then I want to see the feasibility of that and it's frictions through a lens of harm reduction, product building and community building through research, world building and making.
What does this look like if it's built?
Balancing Public Infrastructure with Volunteer Labor
If we look at public infrastructure alternatives, or cooperatives or non-profits, one must also look at the Wikimedia Foundation and the Wikimedia community and related projects. Wikimedia is something to behold, with English language Wikipedia being the fifth most visited website in the world, Wikipedia is written in over 300 languages, and Wikimedia Commons has around 68 million images, photographs, videos and music files. Those examples are just a small slice of the Wikimedia projects: a list of Wikimedia projects can be found here. One could argue that the Wikimedia projects are a net benefit to the world (and I'd wholeheartedly agree!), but they are not without problems. The Wikimedia Community has it's own drawbacks- the community are volunteers, often facing burnout. The community is toxic, often being overwhelmingly male and English speaking. There is a beauty to the communities, in how new language wikis are growing, but one could argue scale is an issue here- there isn't the tools or support or staff (volunteer or of the Foundation) to handle harassment. (As a side note, I used to work at the Wikimedia Community on the Anti-Harassment team).
But, is our future one built purely on volunteer labor- where those, already with the most privilege, succeed? There is plenty of research digging into the drawbacks of open source volunteering models, that I won't dig into right now, but I want to acknowledge it. Effectively, volunteer labor is not necessarily sustainable nor does it create equity driven interactions or communities. Open source, similarly, while creating transparency, does not immediately create healthy community engagements. Tl;dr: not every website can or should be Wikimedia because I don't think our future for public infrastructure or better technology includes volunteer labor running websites. I think that can be a part of it, but it can't be the solution.
Scale Via Numbers
Can cooperatives, non profits or public infrastructure alternatives to tech tools ever 'scale' up and remain equitable? Where do 'ethics' break, where does toxicity start, and what are solutions within that?
I thought I'd be looking at scaling numbers of clients or users, and I am, but I realized through this residency, that's a later part of the problem but not one to be starting with. To start, it's important to look at the "design" of the project itself- how it's volunteers or staff are structured, how they engage with the community, what community or safety protocols they have (like codes of conduct), what the project actually does, and how it accomplishes those things, and so on and so forth. All of that matters, and all of that is designed, even if it's a protocol or way of working that is not necessarily documented or inherently structured. A lack of structure of a 'way of doing something' that isn't documented but known by the community is still a form of structure and design.
Background Research
While in residency, I read and reread Alex Hanna's and Tina M. Park's paper, Against Scale: Provocations and Resistances to Scale Thinking.
I found it to be inspiring, grounding and foundational. Below are a few points and paragraphs I've been ruminating over: From Alex Hanna's paper on scale:
What is scalability: "Scalability is achieved when a system is able to expand without rethinking basic elements [26]. The system is designed in such a way that can accommodate new inputs without changing its fundamental framework. To create such a system, thus, requires an understanding of what those core elements or attributes of the system are - of what aspects can remain unchanged and what elements can be changed to accommodate a growing system. Improving efficiency also means eliminating excess and removing non-essential elements to ensure the system runs smoothly. To engage in scale thinking, then, is to try and reduce a complex process or interaction to its most elemental and simplistic exchange of input to output."
Scale and Process: "By centering scalability in the design and development process, scale thinking revolves around three key tenets: (1) Scalability is a morally good quality of a system; (2) Quantification is a necessary part of designing scalable systems; and (3) Scalability is achieved by identifying and manipulating quantifiable, core elements or attributes of a system"
Some Things Don't (Shouldn't) Scale: "In an essay titled "Startup = Growth," Y Combinator founder Paul Graham discusses how the distinguishing feature of startups is its ability to grow, stating "not every newly founded company is a startup" and most of those are in service. The startup is an instantiation of scale thinking which requires immediate growth. "A barbershop doesn't scale," he quips. In scale thinking, this feature of any given system is assumed to be the most relevant for success and longevity. As Werner Vogels, chief technology officer of Amazon, wrote, "scalability cannot be an after-thought. It requires applications and platforms to be designed with scaling in mind..."
Can 10,000 libraries actually save social networks?
Yes, back to libraries. Prior to starting this specific research fellowship, I saw and am seeing a lot of conversations around public infrastructure and libraries but without a lot of concrete examples of what that would look like, what it would accomplish, and how that would affect the actual librarians
During this residency, I spoke to Siân Evans, a librarian with MICA. Our conversation yielded the following points:
Public libraries serve so many functions (as meeting center, knowledge center, community center and even social workers, impromptu EMTs) that they are stretched super thin. Evans' friend works in a public library, regularly receives Narcan training. Librarians are often overworked, and are at the behest of public funding
10,000 Librarians For Social Networks Can Paint Over the Issues with Public Infrastructure Alternatives.
Thinking of librarians as a metaphor often ignores the fact that public infrastructure has specific, smaller budgets and massive constraints (eg libraries in particular). It's not as simple as we say it is (there's always complications)
Who owns the infrastructure or knowledge?
Evans, in our interview, talked about working with Occupy Wallstreet and being a part of their library. There were issues at the every end of deciding where this library went, because then it came down to who physically 'held' the library, and that was a form of ownership. With public infrastructure, which institution is 'housing' it is also the owner of it. If it's a public institution, then which city, state etc is funding it. Sustainability is still an issue- because what happens if funding shifts or disappears?
What I've Learned
What Makes Some Cooperatives Work is Inherently a Part of Their Design That Fights Traditional, Silicon Valley Definitions of Scale
Mutual aid groups, like BedStuy Strong, functioned so well because it only functioned for BedStuy. Someone could have started a Cobble Hill Strong, using some suggestions from BedStuy Strong, but BedStuy Strong couldn't have survived by helping ALL of Brooklyn. Thus, the design of these alternatives is to avoid scale at all costs, and that's what creates the success or the functionality of these initiatives.
Local is a Form of a Scale- But it's Not Scalability
Related to the above, there can and should be a shift in thinking around scale. Local and constrained size is a form of a scale. By redefining scale, I think we can better understand the alternatives we need to build- ones that center humanity, harm reduction, harm mitigation and care. From Hanna's and Park's paper "mutual aid networks, by their nature, are not intended to "scale". While scale thinking emphasizes abstraction and modularity, mutual aid networks encourage concretization and connection. Mutual aid is intended to operate as a mode of radical collective care in which individuals in the network have their direct material needs met, regardless of considerations of those receiving aid falling into a set of datatified categories, such as "deserving or undeserving." While scale thinking encourages top-down coordination, mutual aid considers building skills for "collaboration, participation, and decision making."
Cooperatives Can Grow Too Much to Scale Up Even When They Want To Grow
During my residency, I had a really lovely conversation with Caroline Woodward of Open Collective on the design of their website, their intentions and how they are growing. For example, Open Collective is a tech platform that enables a network of 600+ nonprofits and aligned entities to support 7000+ groups to legally raise and spend $30M+ each year. So one of these 600+ could technically sponsor me or you or anyone without any of us becoming a traditional nonprofit. Anyone can sign up and be on open collective regardless of non profit status (eg. you don't need one).
Design Matters for Cooperatives, Even with Scale/Anti-Scale
As I mentioned earlier, I'm using design here loosely but also specifically. Design can be the actual work structure, the tech the project is using, how the tech is built or manifested, the 'rules' or norms for how the community interacts with one another and the project itself. Design can also be a specificity and a lack- a lack of documented structures or norms is still a form of design.
While researching during this residency and discussing it with by Woolward and Evans, the notion of this expansive definition of design came up: how things are communicated digitally matters but the 'structural' and community design matters, as well. We can think of design as having multiple meanings, such as being the actual 'structural' design of a community, of the hierarchy, workers, trust and safety, admin, etc to then the actual design of a website. All of these forms of design must be defined, structured, and created. For example, having a code of conduct doesn't just determine safety, that code of conduct has to be implemented and staff has to be trained, and structures have to be created to intact reports and respond. But, having a code of conduct is a good step in the right direction towards community safety. Another example is creating clear spaces and structures for cooperative feedback, this is a form of governance design. This kind of governance and communal feedback has to be designed, and then implemented. Regardless of size, design matters; design matters even if the cooperative or collection or project is small and anti-scale.
Scaling Anti-Scale and Value setting
Caroline Woodward and I had a really interesting conversation about collective buy-in for a community, thinking aloud through how could a community of 7,000 actually function like a cooperative, including 7,000 voices and choices and weigh in and points on decision making. Perhaps, in the context of Open Collective, how can the idea of Open Collective shift towards cooperative thinking, like the benefits of ownership and stewardship, as opposed to viewing Open Collective as a kind of Patreon alternative?
What does ownership and stewardship entail and then look like at the scale of 7000 people? This is something that Open Collective is thinking about. From this prompt, I wonder-- could there be different kinds of touchpoints or requirements for people within the cooperative. An immediate example that comes to my mind is my teachers union. I just joined a teaching union here in the UK. By joining the teachers union, we recently had a vote for striking (to stroke or note strike). The first point of involvement was to vote, and then the second was: would you agree to strike. Without one having to join the union leadership group (which you could), there are basic but escalating ways to be involved. After voting, there was a survey. Our union asked us: do you want to feel out the survey? You don't have to, it's very helpful if you do, and so it becomes these different levels all the way up. Perhaps, with a cooperative it's worth thinking through the most basic thing you need for people to do, and then it's breaking that down into different kinds of hard and soft requirements. Eg with the union, it's better if I vote but I won't get kicked out if I don't vote.
This reminds me of something I learned when researching "Responsible Design for Digital Communities" in how BedStuy Strong designed different ways for community members to interact with each other. From my report, "[BedStuy Strong uses] Slack, Google Voice and Airtable to help organize deliveries. Samantha Garfield, a volunteer with BedStuy Strong gives an example of how the group tries to meet and organize across their nearly 3,000 members. "The other day we posted an all-hands meeting where we were looking to get broad based buy-in on a set of guiding principles for our mutual aid group, and we were struggling with how we can we do that effectively because there might be hundreds of people on the call, and the goal is not just to ask them to thumbs-up, thumbs-down, but to genuinely invite hand-raisers, [all] different levels of possible engagement. So we had a conversation going where we were, A, asking people to Stack in the Zoom chat, B, asking them to weigh in via a straw poll on Slido, and then C, concurrently posting a Q&A on Slido and having people upvote or downvote questions. So there were sort of levels to how people could voice their dissent based on how extreme the dissent was," outlines Garfield on the workflow..."
Collective Trust and Safety: Thinking Through Collective Actions
A potential downside to collective/cooperatives is who responds to harassment and harassment migration (something I've learned from my Digital Infrastructure grant work).
Even with codes of conduct or trust and safety teams, we have to ask what are the protocols and structures for those teams and how were they trained? For example, without training, groups can have inconsistent responses to harassment, become unsure how to respond to harassment, or respond inappropriately.
For a platform, even if there's pushback from their communities like within Twitter or Wikimedia, a platform can listen to users, but they don't necessarily have to. They can make different kinds of executive decisions about policy. At first, this doesn't sound beneficial to users. But what if a lot of the most vocal users are super toxic and from a particular region (let's say Western Europe, North America) and are overwhelmingly male? Making nuanced policy decisions to protect marginalized groups is good- that's a kind of executive decision a platform can make. But if this push for equity has to come from the community- will it? Or will pushes for equity be drowned out?
How does this relate to Wikimedia? Wikimedia, the communities, are overwhelmingly white, male, and from the Global North. The nonprofit that helps 'power' Wikimedia does not have a lot of control, partially by design (but it could if it wanted to). Wikimedia functions by having a lot of buy-in or approval from it's communities- often meaning initiatives have to be proposed or supported by or coming from the community. I think that means a lot of nuance around harassment and toxicity is much slower moving than other platforms. Some open source communities have a bit of an allergy towards nuanced forms of equity and social justice. So, this is where some kind of decision being made is helpful, especially if it results in a net-positive, or creating more generally equity across all groups. Thinking of toxicity, toxicity mitigation and scale- this kind of mitigation works only if a smaller group is in charge- so how would that function for a cooperative? I don't have an answer for this.
Next Steps:
This topic still needs to be researched and I think it's a fruitful one. There's a lot to be analyzed and looked at within this project. My goal is to start creating more specific and nuanced case studies, one being feminist data set, but others being working cooperatives and collectives like Open Collective, an open source library/community, data cooperatives, mutual aid groups, etc, and applying questions of scale, refusal, community design and structural design (of the tool or app, and of the community) to each use case.
Coming out of the research, I want to create a potential card game or design artifact- I'm imagining this as an anti-monopoly. My rationale is I'm interested in a way to get people to think about the constraints of this issue--- scalability is a very real constraint. Not everything can and should be scaled. What we give up when things scale is the ability to have human connection and an easier way for a human to solve a problem. What I mean by that is to imagine the customer service, or for a cooperative, people equally weighing in or having meaningful ways to weigh in. What is a cooperative with 200,000 people? Maybe that can't be- maybe that has to be a variety of cooperatives, divided up by reason, location, domain, expertise (similar to unions) that can be a part of bigger networks Maybe it's recognizing, at times, the power of decentralization while holding onto a cooperative structure (e.g not everyone can be or needs to be in the same 'group' but they can be in similar kinds of groups). The game would play with and explore these dynamics.