Competitive Algorithms, Cooperative Alternatives: Rethinking Evolutionary Logics in AI through Symbiogenesis and Human–Machine Collaboration

Abstract   

This project interrogates how evolutionary algorithms (EA) embed Darwinian logics of competition into knowledge production and labour, and proposes cooperative alternatives inspired by symbiogenesis and human–machine co-flourishing. Focusing on NASA’s SPIKE telescope scheduling system and human data annotation, the study reveals how “fitness” functions encode hidden assumptions about competition, value, and visibility. Drawing on Lynn Margulis’s symbiogenetic theory and labour studies on collective and union practices, I develop alternative algorithmic architectures that foreground epistemic diversity, equity, and cooperation. The project employs Ruth Levitas’s Utopia as Method to excavate assumptions, critique their implications, and construct speculative alternatives. This dissertation is deliberately scoped to two core cases—SPIKE and annotation labour—analysed through existing documentation and recordedpersonal experience, ensuring feasibility within the MSc timeframe. This matters because as algorithms increasingly govern resource allocation—from telescope time to the prioritisation of research proposals—understanding their embedded values becomes essential for democratic knowledge production and equitable labour relations.

Rationale

This research examines how evolutionary algorithms shape knowledge production and human–machine labour relations through a philosophical investigation of their embedded logics. Framing this as a wicked problem confronting contemporary societies, the core question driving this work is: How do evolutionary logics of competition apply in algorithms at the times of AI? I explore this through multiple cases: the SPIKE scheduling system, which showcases how such logics structure algorithmic decision-making in astronomy, and the field of data labelling labour, which demonstrates how these logics render invisible and dispel the very human actors and component entities that sustain AI’s evolution. The project also asks: what alternatives might symbiotic logics offer? This matters for four interconnected reasons:

1. Philosophical Significance: Why Evolutionary Algorithms as Worldviews Matter
   The core question driving this work is:How do evolutionary logics of competition apply in algorithms in the age of AI? And what alternatives might symbiotic logics offer?Evolutionary algorithms are not mere technical tools; they are carriers of worldviews. By embedding competitive Darwinian logics into code, they canonise particular scientific theories and render them resistant to critique. While in the humanities and social sciences theories are constantly debated, once theories are embedded in scientific and computational infrastructures they often present themselves as “objective science.” Once canonised in algorithms, such theories become difficult to question, spreading their embedded problems throughout society while claiming objectivity (Haraway, 2015; Daston & Galison, 2010). This veneer of neutrality conceals assumptions and diffuses problems across technology, society, and everyday life. A philosophical approach makes these hidden assumptions visible again, re-opening them to scrutiny.
2. Societal Urgency: Knowledge, Labour, and Inequality in the Age of AI
   The urgency of this project lies in how evolutionary algorithms shape both knowledge production and human labour. My two years of experience as a data annotator—requiring high skill but offering minimal wages and frequent exposure to harmful content—highlight the precarious working conditions of the “unequal cooperation” between humans and AI. Workers oftentrade off themselves, consenting to exposure to violent or erotic material in order to secure more cases, revealing how AI’s evolution is built upon human sacrifice. These conditions make the human–machine relationship particularly visible and ground abstract philosophical concerns in tangible labour realities. At the same time, algorithms like SPIKE determine what parts of the cosmos humanity observes, acting as epistemic gatekeepers. Beyond astronomy, evolutionary logics systematically privilege some actors while marginalising others across fields, often along lines of geography and institutional prestige. As the AI boom accelerates, critique has largely focused on data bias and auditing practices—that is, the “end-product” layer of the problem. Yet the algorithmic layer itself—the foundational logics of competition—remains under-examined. This project directly addresses that gap.

The Core Insight

Algorithms are not neutral tools but embed specific worldviews. When we use evolutionary algorithms to schedule telescopes or train AI systems, we are not simply optimising efficiency. We are encoding assumptions about competition, value, and visibility—what deserves to be seen, and who deserves to work. This research makes those hidden choices visible and asks: what if we designed algorithms based on logics of cooperation rather than logics of competition? What if we incorporated a collective “union logic” into selection processes? Can we begin to imagine symbiotic, utopian futures of AI algorithms and human–AI ensembles?

Literature Review

This research builds on four main bodies of scholarship: technical literature on technical studies on evolutionary algorithms, critical algorithm studies, human–AI labour relations, and symbiogenesis.

1. Evolutionary Algorithms as Translations of Biology in Astronomical Practice and Beyond: Translation of Darwinian Theory into AI Computational Practice

Evolutionary algorithms (EA) form a family of optimisation techniques inspired by Darwinian natural selection, including Genetic Algorithms (GA), Evolution Strategies (ES), and Genetic Programming (GP) (Mitchell, 1998; De Jong, 2006). They operate through iterative cycles of selection, mutation, and reproduction, where “fitter” solutions survive to the next generation. These are not neutral metaphors: they are active computational translations of biological theories that privilege certain visions of evolution. As such, EA embed specific interpretations of evolution into computational infrastructures.

In astronomy, EA have been deployed for tasks ranging from telescope scheduling to galaxy classification and signal detection. SPIKE, developed in the 1980s for Hubble and later adapted for the James Webb Space Telescope, exemplifies one of the earliest applications of EA to the allocation of scarce scientific resources. The system treats observation proposals as competing organisms in an evolutionary landscape, with a fitness function determining which proposals best optimise telescope usage. Each proposal becomes an “organism” defined by parameters whose setting remains opaque. Technical documentation reveals that fitness is calculated based on factors including target visibility windows, instrument configuration changes, and scientific priority scores. Yet who sets these parameters—and according to which values—remains unclear, raising critical questions that this research addresses.

2. Critical Studies of Algorithmic Knowledge Production
   Critical algorithm studies have illuminated how technical systems encode social assumptions and reproduce power relations (Winner, 2017; Benjamin, 2019). Haraway’s (1988) notion ofsituated knowledgesreminds us that algorithms, like any knowledge apparatus, are shaped by positionality and perspective; thus, the knowledge they produce is always partial. In the case of SPIKE, the underlying fitness functions do more than optimise—they quietly curate a vision of what scientific value ought to be, under the guise of an objective “god trick” (Haraway, 1988), even though they remain essentially partial.

Barad’s (2007) work on the philosophy of quantum physics further illustrates how measurement and apparatuses enact the world. Algorithms such as SPIKE do not simply optimise—they enact visions of scientific value, shaping what phenomena become visible and which remain obscured. Yet most critiques of AI bias centre on data disparities, output disparities, or algorithmic audits (Blodgett et al., 2020), rather than the deeper architectural logics. While there is substantial work on algorithmic auditing—evaluating fairness, transparency, and justice—these approaches largely examine algorithms after they are built (the “end product”), rather than questioning the foundational logics embedded during design. This project addresses precisely that unexamined gap.

These insights also resonate with long-standing debates in the history and philosophy of science. Daston and Galison (2010) trace the emergence of objectivity alongside new scientific instruments, showing how ideals of neutrality co-evolve with technologies of vision. Astronomy, mediated through telescopes and algorithms, exemplifies this entanglement: the telescope does not simply extend human sight, it shapes what we come to know as the cosmos. Bruno Latour’s laboratory ethnographies echo this perspective, demonstrating how scientific facts emerge through messy negotiations between humans and machines, numbers and narratives. Following this line, SPIKE is neither invisible nor impartial. It is a quiet actor in the production of cosmic truth, a gatekeeper of the stars. To study how it allocates telescope time is to study how our image of the universe is filtered, structured, and governed by code.

3. Human–AI Labour Relations and the Given Situation of Algorithmic Evolution

The emerging literature on data work and AI labour (Gray & Suri, 2019; Irani, 2015) provides crucial context for understanding how algorithms evolve through human contributions to data annotation, often under exploitative conditions. My own experience as a data annotator illustrates how highly skilled cognitive labour is rendered invisible, echoing Haraway’s (1985) figure of the “women in the integrated circuit.” This reveals a profound historical shift: from twentieth-century scientific labelling—where scientists labelled in order to prove expertise and gain self-affirmation through the scientific practice of “trained judgement” (Daston & Galison, 2010)—to contemporary data work, where precarious or low-status labourers remain invisible despite performing similar tasks. This transformation highlights how the same activity—labelling—carries entirely different meanings for agency and subjectivity depending on socioeconomic position. Crucially, annotation labour is governed by its own global logic of competition: workers are selected on the basis of lowest cost and highest output quality, mirroring the selection mechanisms of evolutionary algorithms themselves. Thus, the competition that drives algorithmic scheduling also structures the digital labour market, binding human sacrifice into the very evolutionary logics of AI.

Current forms of human–machine cooperation are often highly unequal. Data annotation exemplifies this imbalance: humans work under precarious and poorly paid conditions to sustain the evolution of AI, while companies reap disproportionate benefits. Such dynamics resonate with broader critiques of platform and digital capitalism. Examining how the evolutionary logic of AI advances at the expense of human wellbeing provides a concrete entry point for philosophical analysis. If SPIKE is the gatekeeper of the stars, annotation labour is the toll paid at the gate: silent sacrifices that fuel algorithmic evolution.

4. Symbiogenesis and Alternative Evolutionary Frameworks

Against the grain of competitive machinery, Margulis’s symbiogenetic theory offers a provocative alternative and a cooperative counterpoint to Darwinian competition, serving as a resource for imagining cooperation between different biological organisms and heterogeneous entities. Evolution, she argued, was not only a struggle for dominance but also a history of unlikely alliances—bacteria fusing, collaborating, and creating entirely new forms of life. This spirit of reciprocity, though rooted in microbiology, holds imaginative potential for rethinking the architectures of algorithmic allocation.

My supervisor, Dr Hackl, also suggested that I read The Swarm, a work of fiction about organismal cooperation in the deep ocean. This recommendation aligns with EFI’s ethos of taking speculative fiction seriously as a mode of knowledge production. Such a perspective opens possibilities for exploring how “union logics” might be brought into algorithmic selection processes, thereby creating alternative frameworks that resist zero-sum competition. Importantly, this idea of unions and cooperative logics is not merely speculative: in the digital labour market, unionisation is typically absent and corporations often actively suppress collective organising, even as workers invent creative forms of resistance (). Situating speculative “union logics” alongside these real-world struggles provides a powerful counterpoint to both the competitive evolutionary logics embedded in EA and the unequal human–AI labour relations outlined above.

Taken together, these four strands of literature illuminate both the competitive logics already embedded in evolutionary algorithms and the possibilities for imagining alternatives. Technical studies of EA demonstrate how Darwinian frameworks are translated directly into computational infrastructures, while critical scholarship reveals how such systems enact partial visions of value and truth under the guise of neutrality. The literature on human–AI labour exposes how algorithmic evolution depends upon precarious and often invisible human contributions, raising urgent ethical and political concerns. Finally, Margulis’s symbiogenetic theory and speculative perspectives offer conceptual resources for rethinking algorithmic architectures around reciprocity, plurality, and cooperation. Building on these bodies of work, this project positions itself philosophically: to excavate the hidden assumptions encoded in EA, to critique their social and epistemic consequences, and to construct speculative alternatives that foreground justice and collective flourishing.

Case Studies: Making Evolutionary Logic Visible
To understand the stakes of algorithmic mediation across different domains, I examine multiple cases under the philosophical umbrella of evolutionary logic in algorithms.
The first case (SPIKE) serves as a “hard case” of evolutionary algorithms: technically, it is a direct application of EA, while philosophically it carries the significance of epistemic gatekeeping. The second case (annotation labour) functions as a “soft case”: it is not an EA technique in itself, but it is intimately connected to the “evolution” of AI and to the human substrate of algorithmic evolution. It enables us to ask: when algorithms are advanced through such evolutionary processes, what kind of evolutionary logic governs the co-labour of humans and machines within this dynamic of algorithmic evolution?

1. Primary Case: How SPIKE Shapes Cosmic Knowledge
   The Hubble and James Webb Space Telescopes cost hundreds of thousands of dollars per minute to operate, yet their observation schedules are determined by an algorithm calledSPIKE, jointly developed by STScI and NASA in the 1980s, which continues to be used today. SPIKE employsGenetic Algorithms (GA), a type of evolutionary algorithm, to optimise telescope time, treating different observation proposals as genome-like entities that undergo crossover, mutation, and iterative selection until an “optimal” schedule emerges.

When astronomers worldwide submit observation proposals to use Hubble or Webb, they enter a complex algorithmic competition they may not fully understand. SPIKE receives thousands of proposals annually, each requesting specific observation windows, instruments, and exposure times. The system treats this as a massive optimisation problem: how to pack the maximum “scientific value” into limited telescope time whilst respecting technical constraints.

The evolutionary algorithm works by:

1. Creating an initial population of possible schedules
2. Evaluating each schedule’s “fitness” based on multiple criteria
3. Selecting high-fitness schedules to “reproduce” with mutations
4. Iterating until an optimal schedule emerges

Research proposals are treated as organisms, translated into fitness functions through parameters whose setting remains opaque. While I acknowledge, as discussed in supervision, that fully understanding SPIKE’s operation presents challenges—the data may not be publicly accessible—this case remains valuable as it represents an extreme instance of epistemic gatekeeping in knowledge production. SPIKE is not a neutral scheduler but a gatekeeper of cosmic knowledge: by deciding which parts of the universe deserve to be seen, it vividly encodes Darwinian competition logics into the infrastructures that govern our epistemic futures.

Potential Algorithmic Biases in SPIKE

Based on how evolutionary algorithms typically function and SPIKE’s documented emphasis on efficiency, several concerns merit investigation:

• Efficiency optimisation: If the algorithm prioritises observations requiring minimal telescope movement and instrument changes, this could disadvantage innovative research requiring unique configurations
• Possible institutional effects: Systems that consider past performance might create feedback loops advantaging established institutions
• Geographic considerations: Time zones and ground station locations could create systematic access inequalities for certain regions

These potential biases warrant systematic investigation through data analysis.

Hypothetical Impacts of the Potential Algorithmic Biases in SPIKE

If these biases exist, a proposal to study an unusual celestial phenomenon requiring complex instrument configurations from a smaller university in the Global South could face multiple algorithmic disadvantages before any human evaluates its scientific merit. The algorithm’s efficiency optimisation could become a de facto gatekeeper determining whose questions about the universe get answered.

2. Complementary Case: Annotation Labour and the Human Cost of Algorithmic Evolution
   Alongside astronomical scheduling, this project turns to the hidden but essential work of data annotation. Modern AI systems, from natural language processing to reinforcement learning and neuroevolution, rely on vast amounts of labelled data to evolve. Annotation labour is therefore not a direct application of evolutionary algorithms, but it constitutes thehidden substrate of algorithmic evolution: the indispensable human condition that allows evolutionary and learning-based AI to function. In this sense, annotators are not external to the system but are incorporated into itsselection environment, co-determining how algorithms evolve.

Drawing from my own two years of annotation experience, the work requires high levels of skill and concentration—passing language tests, undergoing extensive training, and demonstrating precise judgement—yet is compensated with minimal wages. Economic precarity often forces annotators to accept tasks involving violent or pornographic material in order to access more cases, exposing them to long-term psychological risk. Unlike twentieth-century scientists who engaged in labelling as a way to establish expertise and affirm their proud and trained judgement (Daston & Galison, 2010), contemporary annotators perform similar cognitive labour under conditions of invisibility. This dynamic echoes Haraway’s (1985) figure of the “women in the integrated circuit”: highly skilled workers whose contributions remain structurally undervalued.

of current human–machine cooperation. The apparent “evolution” of AI is built upon hidden human sacrifice, making annotation labour an exemplary case of unequal cooperation. Just as SPIKE acts as a gatekeeper of cosmic knowledge, annotation labour represents the hidden toll of algorithmic evolution: unseen sacrifices that sustain machine “progress” and corporate profit-making. This inequality is further reinforced by the structure of the digital labour market itself: the fierce competition and chronic over-supply of annotators on global platforms drive wages down to rock-bottom levels, enabling corporations to select ever cheaper and “better,” “fitter” workers while undermining collective organising power (Irani, 2015; Gray & Suri, 2019; Wood, Graham, Lehdonvirta, & Hjorth, 2019). In this sense, annotation does not merely illustrate invisibility and exploitation but also demonstrates how the evolutionary logics of AI are entangled with the competitive architectures of the gig economy. Studying this case is therefore not only a matter of labour critique but also a philosophical inquiry into whether this co-labour between humans and machines embodies competition, exploitation, or the possibility of genuine cooperation.

Taken together, these two cases are mutually reinforcing. SPIKE demonstrates how evolutionary algorithms act as epistemic gatekeepers, structuring what counts as legitimate cosmic knowledge, while annotation labour exposes the hidden human costs that sustain algorithmic evolution, revealing unequal forms of human–machine cooperation. Read side by side, they foreground the central philosophical question of this dissertation: whether the evolutionary logics embedded in AI inevitably reproduce competitive exploitation, or whether they might be reimagined through cooperative and symbiotic alternatives.

3. What Gets Lost in Competitive Logic

Across all cases, when competitive optimisation becomes the sole criterion, we lose:

• Serendipitous discoveries from “inefficient” observations or unconventional approaches
• Diverse perspectives from researchers and workers outside dominant institutions and privileged working environment
• Unconventional approaches and possibilities that don’t fit standard and competitive templates

• Equitable access to humanity’s shared scientific instruments like resources and opportunities

• Human wellbeing in the service of algorithmic evolution

This is not just about fairness, but about the kinds of knowledge and society we produce. When algorithms embed competitive logic into infrastructure, they shape not just who can participate but what questions can be asked, what knowledge can be generated or canonised  and what futures become possible.

Methodology

This research employs Ruth Levitas’s “Utopia as Method” as an organising framework, drawing on her triadic approach of excavation (archaeology), critique (ontology), and construction (architecture) to examine and reimagine algorithmic systems and human-AI relationships. Positioned as a philosophical project, it integrates theoretical analysis with empirical investigation through accessible methods, ensuring both critical depth and practical feasibility.

Phase 1: Excavation / Archaeology – Uncovering the Evolutionary Logics in AI

In the archaeological mode, I examine the historical and infrastructural foundations of evolutionary algorithms, focusing on how embedded logics shape scientific possibilities and human-AI relationships. This involves:

• Document Analysis: Conducting close reading of publicly available technical documents, papers, and guides to trace how concepts like ‘fitness’, ‘optimisation’, and ‘efficiency’ are operationalised.
• Expert Interviews: I will probably conduct interviews if time permitted with:
  • Computer scientists working in astronomy (contacts from Edinburgh astronomy summer school)
  • Astronomers using SPIKE
  • Other data annotators and gig workers
• Accessible Data Collection: Compiling available data on telescope allocations, worker conditions and situations, and algorithmic decision-making patterns.
• Timeline Construction: Situating algorithmic evolution within broader technological and social developments.

This phase aims to render visible the historical sedimentation of values within EA’s architecture and type of given human-AI co-labour relationships and to contextualise their emergence within techno-scientific rationalities.

Phase 2: Critique / Ontology – Reading Algorithmic Assumptions
In this phase, the project interrogates the underlying assumptions about knowledge, value, and ethics encoded in the design of evolutionary algorithms (EA):

• Philosophical Analysis: Examining how competitive logic shapes possibilities for knowledge and labour.
• Discourse Analysis: Conducting critical readings of algorithmic documentation and annotators’ labour records, with particular attention to how optimisation is framed and how computation and human–machine relationships are presented as neutral rather than value-laden.
• Textual Mining: Using digital-humanities tools such as Voyant Tools to explore rhetorical frequencies and associations—for example, how terms like “efficiency,” “scientific merit,” or “competition” appear and cluster within official texts and relevant literature.
• Comparative Analysis: Identifying patterns across different manifestations of evolutionary algorithms.
• Power Mapping: Tracing how algorithmic competition creates and maintains inequalities.

This project employs philosophical textual close reading of technical and labour documents to bridge the gap between philosophical and technical layers, with occasional support from digital humanities tools, to identify linguistic patterns. In the SPIKE case, I will examine how terms like fitness, efficiency, and optimisation are defined, showing how certain goals (e.g. minimising telescope slews) are framed as fit, efficient, or optimal in preference to others. In the annotation case, I will triangulate my own auto-ethnographic experience with publicly available task templates and ethnographic accounts, analysing how instructions frame annotation as modular “micro-tasks” while obscuring its skilled and affective dimensions. These close readings will be interpreted through critical theories in feminist STS and philosophy of technology: Haraway’s concepts of “situated knowledges” and the critique of the “god trick” expose the partiality behind neutrality; Daston and Galison’s account of the historical contingency of objectivity to situate fitness as codified authority; and Winner’s and Latour’s insistence that scientific and technical systems act politically. In this way, the method identifies how value-laden choices are encoded in apparatuses and institutions, and highlights which kinds of subjects are sustained and privileged, shaping which forms of knowledge and subjectivity are enabled to flourish while others are marginalised.

The aim is not to discredit algorithmic developments, but to examine how particular ontologies of science and subjects or actors are sustained, flourished and privileged under the current framework, often with subtle yet material consequences for access and representation.
Phase 3: Construction / Architecture – Speculative Reconfiguration

The final phase explores how algorithmic systems and arrangements might be otherwise. Drawing on symbiogenetic theory (Margulis, 1998) as an alternative evolutionary framework, I will outline speculative design directions and imaginative configurations that resist zero-sum competition:

• Principle Translation: Reimagining algorithmic values through cooperative and symbiotic rather than competitive metaphors, emphasising reciprocity, diversity, and interdependence.
• Alternative Metrics: Proposing new “fitness” criteria:
  – Epistemic diversity rather than efficiency
  – Geographic inclusion rather than institutional prestige
  – Methodological novelty rather than standardisation
  – Collective benefit rather than individual optimisation
  – Worker wellbeing alongside system performance
• Prototype Framing: The Endosymbiotic Scheduler as Counter-Model:
  Instead of developing a technical prototype, I will sketch a conceptual and speculative model of an “endosymbiotic scheduler.” This counter-model challenges the Darwinian assumption that only the “fittest” deserve to survive, which currently structures both knowledge production and labour distribution.

• In astronomical scheduling, it asks what would happen if proposals were selected not through competitive optimisation alone but by identifying complementarities—projects that enrich each other through methodological diversity, geographic distribution, or shared data infrastructures. Instead of filtering out less efficient or standardised proposals, the algorithm could foster synergies, multiplying epistemic value through cooperation.
• In annotation labour, it becomes a metaphor for resisting the atomising logics of the gig economy. Rather than allocating tasks solely to the cheapest and fastest annotators, it would embed cooperative metrics—worker wellbeing, skill-sharing, and sustainability—into the selection environment. This vision resonates with empirical experiments in cooperative digital labour, such as worker co-operatives in the digital economy (de Peuter, de Verteuil, & Machaka, 2022) and Irani and Silberman’s (2013) influential Turkopticon, demonstrating how annotators and crowdworkers have resisted fragmentation and created forms of collective agency. These practices offer inspiration for imagining algorithmic architectures that embed cooperation not only metaphorically but organisationally. Alongside this, I will envision a possible human–AI ensemble based on mutual benefit within the dynamics of co-labour.
• Planetary Market as Structural Context:
  This counter-model must also be situated within the planetary political economy of digital work. As Graham and Ferrari (2022) argue, platformised labour markets operate at a global scale, producing chronic oversupply, wage suppression, and algorithmic control across borders. Workers in the Global South are differentially positioned within these circuits of extraction, yet the same infrastructures can enable transnational solidarities. Embedding symbiogenetic principles into algorithmic architectures therefore requires attending to planetary asymmetries in how digital work is distributed, priced, and valued, while designing for cross-border forms of cooperation.

By bridging these domains, the endosymbiotic scheduler functions as a speculative and experimental design for algorithmic systems that evolve towards cooperation, reciprocity, plurality, and justice—across both cosmic knowledge and human labour.

This methodology balances theoretical critique with practical investigation. Levitas’s framework provides a systematic way to move from diagnosis through critique to the construction of alternatives—essential for research that seeks not just to identify problems but to imagine different futures. By integrating speculative theory with real-world practices of digital worker cooperation, the project foregrounds epistemological inquiry over technical mastery. The goal is to reflectively redesign: how might algorithms participate in shaping futures that are more just, inclusive, and generative?

Project Report Format: 8,000-word Analytical Essay with Comics Supplement

The final project will adopt a multimodal format that combines rigorous textual analysis with visual storytelling. This modified approach draws inspiration from Nick Sousanis’ comics-based research Unflattening (Harvard University Press, 2015) but remains feasible within the MSc timeframe.
This choice rests on four considerations. First, my personal proficiency with comics as an expressive medium makes the format both practical and advantageous; an earlier comics project of mine even attracted publisher interest. Second, the format aligns with the Edinburgh Futures Institute’s emphasis on interdisciplinarity and creativity, integrating textual scholarship with visual narrative in ways that reflect EFI’s core values. Third, the complex subject matter of astronomical observation and algorithmic knowledge production—exemplified by NASA’s SPIKE scheduler—benefits from visual representation, which can make abstract theories tangible and engaging. Finally, adopting a multimodal structure addresses word-count constraints: rather than a 15,000-word traditional dissertation, an 8,000-word essay complemented by a concise research comic allows for both analytical depth and accessible communication.

In sum, a semi-comics-format dissertation can leverage my strengths as a researcher-artist and also provide an ideal multimodal vehicle to convey my research effectively, creatively, and meaningfully.

Here are some excerpts from Sousanis’ work, Unflattening, the first comics-based research submitted as a doctoral dissertation at Teachers College, Columbia University.

https://www.youtube.com/watch?v=Ln7J10yn9iA

Following discussions with my supervisor Dr Hackl on feasibility, I therefore propose the following structure:

Primary Component: 8,000-word Philosophical Analytical Essay

• Frames the research question and its significance
• Provides theoretical grounding
• Presents detailed case study analysis
• Develops alternative frameworks
• Maintains the intellectual depth expected of MSc research

Supplementary Component: Research Comic (Not Full Graphic Novel)

• Illustrates key concepts of algorithmic evolution and cooperation
• Makes abstract ideas tangible through visual narrative
• Leverages astronomy’s visual richness to enhance accessibility
• Addresses the “tangibility deficit” in previous work
• Remains manageable within the project timeline

The visual component will be integrated with, rather than separate from, the essay, creating a cohesive multimodal argument that combines rigorous philosophical analysis with creative expression.

Proposed Structure of the Dissertation, Timeline and Ethics

Proposed Structure of the Dissertation

Chapter 1. Introduction

• General background and significance of the project
• Research problem and questions
• Rationale and philosophical positioning
• Organisation of the thesis

Chapter 2. Literature Review

• Evolutionary algorithms and their translation of Darwinian theory into computation
• Critical studies of algorithmic knowledge production
• Human–AI labour relations and annotation as hidden substrate of algorithmic evolution
• Symbiogenesis as alternative evolutionary framework

Chapter 3. Conceptual Framework and Methodology

• Ruth Levitas’sUtopia as Method (excavation, critique, construction)
• Philosophical analysis combined with case study approach
• Desk research based and expert interviews if time and resource permitted
• Incorporation of comics as multimodal supplement (research comics as method)
• Ethical considerations

Chapter 4. Case Studies: Making Evolutionary Logic Visible

• SPIKE scheduling system: evolutionary logic and epistemic gatekeeping
• Annotation labour: unequal cooperation and hidden human cost of algorithmic evolution
• Reflective discussion: deepening the excavation and critique dimensions of Levitas’s method, synthesising the implications of the cases for epistemic and labour futures

Chapter 5. Construction: Speculative Alternatives – Symbiotic Architecture

• Symbiogenetic theory as cooperative counter-logic: translating symbiogenetic principles into algorithmic design
• Alternative metrics (epistemic diversity, inclusion, wellbeing)
• “Endosymbiotic scheduler” as conceptual model
• Comics illustrations of speculative architectures
• Discussion of Implications: positioning cooperative logics within wider debates in philosophy of technology and AI governance

Chapter 6. Conclusion: Opening Alternative Futures

• Summarising and Restating findings
• Theoretical, philosophical, critical and constructive contributions
• Implications for science, labour, and AI governance
• Limitations and directions for future research

Appendix (if needed): Selected comics panels, interview materials (if any), technical notes

This dissertation adopts a topic-based (Dudley-Evans, 1999) and case study hybrid structure rather than the conventional IMRAD format. This choice reflects the philosophical orientation of the project: it is not designed to test hypotheses or report empirical results, but to advance critical argumentation, illustrate key dynamics through case studies, and construct speculative alternatives. In this sense, it follows the typical pattern of topic-based dissertations, which organise the argument around themes and concepts rather than around sequential stages of data collection and results.

The element of discussion is therefore not confined to a single chapter but distributed throughout the dissertation. In Chapter Four, reflective discussion links the SPIKE and annotation labour cases to broader questions of epistemic and labour futures. In Chapter Five, a further discussion of implications situates the speculative alternatives within wider debates in philosophy of technology and AI governance. Finally, Chapter Six consolidates these discussions in the concluding synthesis. This distributed approach ensures that critical reflection is integrated at every stage of the analysis rather than appended as a final step.

Visual supplements in the form of short research comics will be interwoven throughout the dissertation. In the case study chapter, comics will be used to visualise algorithmic processes and provide narrative transitions between SPIKE and annotation labour. In the speculative chapter, comics will illustrate the conceptual model of an “endosymbiotic scheduler,” making abstract alternatives more tangible. This integration follows the EFI’s ethos of multimodal and creative research, ensuring that the visual elements function as methodological and rhetorical devices rather than as mere illustrations.

Organisation of the Thesis
This dissertation is organised into six chapters. Chapter One introduces the project, setting out the background, research questions, rationale, and philosophical positioning. Chapter Two reviews the literature, examining evolutionary algorithms, critical algorithm studies, human–AI labour relations, and symbiogenesis as an alternative framework. Chapter Three establishes the conceptual framework and methodology, outlining Levitas’s Utopia as Method, the case study approach, and the integration of comics as a multimodal supplement. Chapter Four presents the core case studies—SPIKE and annotation labour—each followed by reflective discussion linking to questions of epistemic and labour futures. Chapter Five develops constructive alternatives, translating symbiogenetic principles into algorithmic architectures and proposing the conceptual model of an “endosymbiotic scheduler,” with visual supplements used to illustrate speculative designs. Finally, Chapter Six concludes by synthesising findings, contributions, and implications for knowledge, labour, and AI governance.

Visual supplements in the form of short research comics will be interwoven in Chapters Four and Five to support narrative transitions and illustrate speculative architectures.

Project Provisional Timeline

The dissertation is scheduled for completion within the next two weeks, with drafting and integration structured by chapter:

• Week 1 (24–31 August):
  • Drafting of the Literature Review (Chapter 2)
  • Drafting Methodology (Chapter 3)
  • Drafting the SPIKE case study (Chapter 4), alongside preliminary comics sketches
  • Completion of the annotation labour case study and reflective discussion, alongside preliminary comics sketches (Chapter 4)
• Week 2 (1–7 September):
  • Drafting speculative alternatives and implications, alongside preliminary comics sketches (Chapter 5)
  • Drafting the concluding synthesis (Chapter 6).
  • Finalising the introduction (Chapter 1) and the abstract
  • Visual supplements will be finalised in parallel. The final days will be dedicated to editing, integration, and polishing.
  • Incorporating supervisor feedback where available
• Week 3 (8–14 September) (buffer):
  • Completion of all comics supplements
  • Completion of editing, integration, and polishing (wording + visual-text coherence).
  • Incorporating Feedback
  • Submission

Scope guardrails (to ensure feasibility)

• No new data collection; interviews optional and post-submission ensured by the approval of EC later submission.
• SPIKE analysis relies on public documentation/secondary sources; avoid dependency on restricted data.
• Comics supplement capped at 4–6 panels integrated with text (not a standalone graphic novel).

Definition of Completion (Checklist)

• 8,000-word essay complete; comics panels embedded; British spelling; citations consistent; ethics respected; interviews for both case studies only conducted if needed and conditions permitted.

Ethical Considerations
This research draws primarily on publicly available documentation and institutional archives, which minimises concerns regarding personal data or consent. It is guided by the following ethical commitments:

1. Responsible Critique– While interrogating the epistemic assumptions embedded in algorithmic design, the research acknowledges the extraordinary scientific achievements such systems have enabled.
2. Constructive Framing– Critique is framed constructively, always accompanied by speculative proposals for more inclusive and pluralistic algorithmic futures. The aim is to imagine better pathways, not to condemn individuals.
3. Transparency– All interpretive procedures, sources, and theoretical frameworks will be clearly cited to ensure analytical accountability.
4. Engaged Dialogue– Should time permit interviews with astronomers, scheduling professionals, computing experts, and annotators, full ethical clearance will be sought, with attention to consent, confidentiality, and reflexivity.

At its current stage, the project is desk-based and poses minimal ethical concerns. If interviews proceed, informed consent, the option of anonymity, and the right to review quotations will be guaranteed. Reflections on annotation work will remain at a general level, in order to respect NDA obligations, while still contributing valuable insights into industry practices.

Expected Outcomes

This dissertation will generate three interlinked contributions:

Theoretical Contribution: A Feminist-Philosophical Framework for Analysing Algorithmic Evolution as Worldview
The project will develop a systematic critical framework for examining how evolutionary algorithms operate not as neutral tools but as infrastructures that encode Darwinian logics of competition. By tracing how these logics are canonised into technical systems, the research contributes to broader debates in feminist STS and philosophy of technology. In particular, it connects algorithmic visibility regimes with colonial orders of seeing, showing how infrastructures such as SPIKE naturalise epistemic hierarchies under the guise of neutrality. In doing so, the dissertation intervenes in ongoing critiques of instrumental rationality in AI design and foregrounds how alternative logics might be imagined.

Critical Intervention: Making Labour and Power Relations Visible
By juxtaposing the case of astronomical scheduling (SPIKE) with the hidden labour of data annotation, the research illuminates both epistemic and material stakes of algorithmic infrastructures. It demonstrates how evolutionary algorithmsand the evolution of algorithms create hierarchies—whether between research proposals competing for telescope time, or between workers competing for annotation tasks—revealing the unequal cooperation on which algorithmic “evolution” depends. This intervention highlights how optimisation and efficiency logics reproduce exclusions not only in the production of cosmic knowledge but also in global digital labour systems, echoing coloniality in data relations. For debates on AI governance, this research highlights that algorithmic neutrality cannot be assumed: even technical parameters such as fitness functions and efficiency metrics embed contestable values. By exposing these hidden logics of competition, the project offers a conceptual resource for policymakers and system designers to move beyond narrow fairness audits and reconsider the architectural assumptions underlying algorithmic systems.

Constructive Alternatives: Speculative Cooperative Architectures Grounded in Symbiogenesis and Collective Labour Practices
Moving beyond critique, the dissertation develops speculative alternatives rooted in Lynn Margulis’s theory of symbiogenesis and in real-world experiments of collective organising in the digital economy. Symbiogenesis is in itself a feminist intervention into the Darwinian evolutionary paradigm, and this project seeks to translate that intervention into the scholarship of algorithms while situating it within critiques of coloniality in data and infrastructures. These alternatives include:

• Design principles for cooperative rather than competitive algorithms
• Alternative metrics prioritising epistemic diversity, inclusion, wellbeing, and worker sustainability over narrow efficiency
• Conceptual models for “endosymbiotic systems” that foreground mutual benefit and co-flourishing between humans and machines
• Inspiration drawn from worker co-operatives and initiatives such asTurkopticon, which demonstrate how annotators and crowdworkers have resisted fragmentation and created forms of collective agency

The visual supplement, in the form of a short research comic, will serve as both a methodological and rhetorical device: visualising the entanglements between algorithms, labour, astronomy, and feminist critique. By making speculative designs and hidden processes tangible, it provides an alternative epistemic lens through which to engage the stakes of algorithmic mediation.

Collectively, these outcomes demonstrate how feminist technoscience and feminist STS, together with the philosophy of science and technology, speculative methodologies such as utopia as method, selective use of digital humanities tools, and insights from digital labour organising, can expose the coloniality embedded in algorithmic infrastructures while simultaneously proposing cooperative, symbiotic alternatives. The dissertation thus seeks not only to critique but also to reimagine the political, ethical, and ontological futures of AI.

Broader Impact

The research contributes to urgent conversations about AI governance, labour rights, and epistemic justice. By examining how evolutionary logic shapes both cosmic observation and human work, it provides insights applicable to:

• Algorithm design in scientific infrastructure
• Platform labour and gig economy regulation
• AI ethics and governance frameworks
• Alternative approaches to optimisation and resource allocation

Conclusion: Opening Alternative Futures
Ultimately, as algorithms increasingly govern our world—deciding and shaping what we see in space, how we work, and what knowledge is produced and counts as legitimate—questioning their evolutionary logics is not optional but imperative. This is not only a technical issue but also a political and ethical one. This project excavates the competitive assumptions embedded in algorithmic infrastructures, critiques their consequences for knowledge, labour and colonial orders of vision, and constructs cooperative alternatives grounded in symbiogenesis and feminist critique. By doing so, it advances feminist technoscience perspectives and situates its intervention within wider critiques of the coloniality of global data infrastructures. Consequently, it contributes not only to academic debates in philosophy of technology, feminist STS, digital humanities and critical data studies, but also to urgent public conversations about how we might reimagine algorithms in ways that foreground epistemic justice, equitable labour, and symbiotic cooperation.

References

Barad, K. (2007). Meeting the universe halfway: Quantum physics and the entanglement of matter and meaning. Duke University Press.

Benjamin, R. (2019). Race after technology: Abolitionist tools for the new Jim Code. Polity.

Blodgett, S. L., Barocas, S., Daumé, H., III, & Wallach, H. (2020). Language (technology) is power: A critical survey of “bias” in NLP. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.

Daston, L., & Galison, P. (2010). Objectivity. Zone Books.

De Jong, K. A. (2006). Evolutionary computation: A unified approach. MIT Press.

Dudley-Evans, T. (1999). The dissertation: A case of neglect? In P. Thompson (Ed.), Issues in EAP writing research and instruction (pp. 28–36). University of Reading, Centre for Applied Language Studies.

de Peuter, G., de Verteuil, G., & Machaka, S. (2022). Co-operatives, work, and the digital economy: A knowledge synthesis report. Social Sciences and Humanities Research Council of Canada, Knowledge Synthesis program, Skills and Work in the Digital Economy.

Gray, M. L., & Suri, S. (2019). Ghost work: How to stop Silicon Valley from building a new global underclass. Houghton Mifflin Harcourt.

Graham, M., & Ferrari, F. (Eds.). (2022). Digital work in the planetary market. MIT Press.

Haraway, D. J. (2015). Simians, cyborgs, and women: The reinvention of nature. Routledge.

Haraway, D. (1988). Situated knowledges: The science question in feminism and the privilege of partial perspective. Feminist Studies, 14(3), 575-599.

Irani, L. (2015). The cultural work of microwork. New Media & Society, 17(5), 720-739.

Irani, L. C., & Silberman, M. S. (2013). Turkopticon: Interrupting worker invisibility in Amazon Mechanical Turk. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 611–620. https://doi.org/10.1145/2470654.2470742

Latour, B., & Woolgar, S. (1979). Laboratory life: The construction of scientific facts. Sage.

Levitas, R. (2013). Utopia as method: The imaginary reconstitution of society. Palgrave Macmillan.

Margulis, L., & Sagan, D. (2023). Microcosmos: Four billion years of evolution from our microbial ancestors. University of California Press.

Mitchell, M. (1998). An introduction to genetic algorithms. MIT Press.

Sousanis, N. (2015). Unflattening. Harvard University Press.

Schätzing, F. (2006). The swarm (S.-A. Spencer, Trans.). HarperCollins. (Original work published 2004)

Winner, L. (2017). Do artifacts have politics? In The whale and the reactor (pp. 19-39). University of Chicago Press.

Wood, A. J., Graham, M., Lehdonvirta, V., & Hjorth, I. (2019). Networked but commodified: The (dis)embeddedness of digital labour in the gig economy. Sociology, 53(5), 931–950. https://doi.org/10.1177/0038038519828906

Word Count: 6,333 words (excluding title and references)

Aug 30, 2025