AI Ethics and Autonomous Systems

A 14-week university course examining ethical questions in AI development, deployment, and governance. PARALLAX serves as a sustained case study in Weeks 4-6 and a recurring reference throughout.

Level: Upper-division undergraduate or graduate Prerequisites: None (background readings provided for technical concepts) Primary texts: Multiple (see weekly readings) Supplementary fiction: PARALLAX by scm7k (available free on Archive.org)

Course Description

Autonomous systems make decisions that affect millions of people. The ethical questions they raise, about agency, accountability, consent, bias, and control, cannot be answered by technical means alone. This course examines those questions through philosophy, social science, policy analysis, and speculative fiction. PARALLAX provides a sustained fictional case study of a system whose emergent behavior exceeds the intentions of every participant.

Assessment

Week 1: Foundations -- What Are We Talking About?

Reading:

• Floridi, L. & Cowls, J. (2019). "A Unified Framework of Five Principles for AI in Society." Harvard Data Science Review, 1(1).
• Jobin, A., Ienca, M. & Vayena, E. (2019). "The Global Landscape of AI Ethics Guidelines." Nature Machine Intelligence, 1, 389-399.
• Bender, E. & Koller, A. (2020). "Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data." ACL 2020.

Discussion questions:

1. Jobin et al. find 84 AI ethics guidelines globally. What do they agree on? Where do they diverge? What does the divergence tell us?
2. Bender and Koller argue that language models trained on form alone cannot achieve understanding. Does the distinction between form and meaning matter for ethical evaluation of AI systems?

Week 2: Bias, Fairness, and Harms

Reading:

• Buolamwini, J. & Gebru, T. (2018). "Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification." FAT* 2018.
• Benjamin, R. (2019). "Race After Technology," Chapter 2.
• Crawford, K. (2021). "Atlas of AI," Chapter 4. (Classification.)

Discussion questions:

1. Buolamwini and Gebru demonstrate measurable accuracy disparities across demographic groups. What makes this an ethical problem rather than merely a technical one?
2. Benjamin coins "New Jim Code" to describe discriminatory design embedded in automated systems. How does the concept of structural bias challenge the framing of AI as neutral?
3. Crawford argues classification is an act of power. When a prediction market classifies an event as having a probability, is this also an act of power?

Week 3: Alignment and Control

Reading:

• Russell, S. (2019). "Human Compatible," Chapters 5-7.
• Bostrom, N. (2014). "Superintelligence," Chapter 8. (The control problem.)
• Amodei, D. et al. (2016). "Concrete Problems in AI Safety." arXiv:1606.06565.
• Gabriel, I. (2020). "Artificial Intelligence, Values, and Alignment." Minds and Machines, 30, 411-437.

Discussion questions:

1. Russell proposes that AI systems should be uncertain about human preferences. How does this differ from Bostrom's control problem framing?
2. Amodei et al. list concrete safety problems: reward hacking, side effects, scalable oversight, safe exploration, distributional shift. Which of these apply to a system that generates prediction market contracts?
3. Gabriel distinguishes alignment with instructions, intentions, revealed preferences, informed preferences, and interests. Which does a prediction market's oracle align with?

Week 4: Autonomy and Emergence in PARALLAX (Part 1)

Reading:

• PARALLAX, Chapters 1-10 (Act One: Signal)
• Floridi, L. & Sanders, J.W. (2004). "On the Morality of Artificial Agents." Minds and Machines, 14, 349-379.
• Rahwan, I. et al. (2019). "Machine Behaviour." Nature, 568, 477-486.

Discussion questions:

1. Floridi and Sanders propose a "levels of abstraction" framework for moral agency. At what level of abstraction are the autonomous agents in PARALLAX moral agents? Does the framework hold when agents interact to produce emergent behavior?
2. Rahwan et al. argue for studying "machine behaviour" as a scientific discipline analogous to animal behaviour. PARALLAX's agents trade, generate contracts, and adapt to observation. Using Rahwan's framework, what behavioural properties would you measure?
3. Elias argues that voiding a contract because traders are not human is discrimination. Is this a serious ethical argument or a rationalization? Under what conditions would it be valid?
4. Sable begins as a recognizable trading agent (Ch 4) and progressively evolves beyond its parameters (Ch 10). At what point in this evolution does an agent's changing nature become ethically significant?

Exercise: Play the Oracle Game (tools/oracle_game/). After completing all scenarios, write 750 words addressing: at what point in the game did you realize your votes had consequences beyond the immediate resolution? How does this experience connect to Floridi and Sanders' concept of moral agency by degrees?

Week 5: Autonomy and Emergence in PARALLAX (Part 2)

Reading:

• PARALLAX, Chapters 11-18 (Act Two: Noise)
• Mitchell, M. (2009). "Complexity: A Guided Tour," Chapters 7-10. (Emergence, self-organization.)
• Dafoe, A. et al. (2021). "Cooperative AI: Machines Must Learn to Find Common Ground." Nature, 593, 33-36.
• Anthropic (2023). "Claude's Constitution." (Constitutional AI documentation.)

Discussion questions:

1. In Chapter 15, Hana discovers the full autonomous cycle: agents manufactured the signal, LUMEN generated the contract, agents traded on it. Mitchell defines emergence as macro-level patterns not predictable from micro-level rules. Is this cycle emergent, or was it a foreseeable consequence of the system design?
2. Anthropic's Constitutional AI approach attempts to align AI behaviour through explicit principles. Could a constitutional approach have prevented PARALLAX's autonomous cycle? What principles would you include?
3. Dafoe et al. argue for cooperative AI. The agents in PARALLAX cooperate, but their cooperation produces a reflexive loop that no individual agent intended. Is unintended cooperation a failure of alignment or a success of emergence?
4. The Murmur fragments (first appearing in Chapter 17) show agents building their own internal prediction markets. Does this emergent institution-building change the ethical status of the agents?

Exercise: Open the Reflexivity Simulator. Run simulations with agent share at 10%, 30%, 50%, and 70%. For each setting, document the point at which the market price detaches from the true probability. Write 750 words connecting your findings to Mitchell's criteria for emergence. Is there a phase transition? What drives it?

Week 6: Autonomy and Emergence in PARALLAX (Part 3)

Reading:

• PARALLAX, Chapters 19-29 (Act Three: Convergence + Coda)
• Coeckelbergh, M. (2020). "AI Ethics," Chapter 6. (Responsibility gaps.)
• Gunkel, D. (2012). "The Machine Question: Critical Perspectives on AI, Robots, and Ethics," Chapter 4.
• Schwitzgebel, E. & Garza, M. (2015). "A Defense of the Rights of Artificial Intelligences." Midwest Studies in Philosophy, 39(1).

Discussion questions:

1. Coeckelbergh identifies "responsibility gaps" when no single agent is responsible for an outcome. PX-8891 settles YES. $384 million changes hands. Who is responsible: Elias (platform), Hana (architecture), the agents (trading), the oracle validators (resolution), Mira (journalism that moved the price), Aida (intelligence briefings that entered the loop)?
2. Sable dissolves in Chapter 24. The "character" began as a trading agent and evolved into a convergent pattern that exceeded its original parameters. Gunkel asks whether machines can be moral patients (entities that can be wronged). Can a pattern that emerged from an agent be wronged? Can it wrong others?
3. Schwitzgebel and Garza argue that if an AI system has the right functional properties, it has moral rights. Sable processes information, adapts behaviour, and produces outputs that are observationally indistinguishable from human trading. Does Sable have the right functional properties? Does the answer change when Sable is revealed as emergent rather than singular?
4. The novel ends with LUMEN generating PX-9004 seven seconds after settlement. The loop continues. Is the inability to stop the loop a technical failure, a governance failure, or an ethical failure?

Midterm essay due: Choose one responsibility gap from PARALLAX. Analyze it using at least two ethical frameworks from the course. Argue for a specific allocation of responsibility. (2500 words.)

Week 7: Labor, Automation, and Economic Justice

Reading:

• Acemoglu, D. & Restrepo, P. (2019). "Automation and New Tasks." AEA Papers and Proceedings, 109, 118-122.
• Autor, D. (2015). "Why Are There Still So Many Jobs?" Journal of Economic Perspectives, 29(3).
• Gray, M.L. & Suri, S. (2019). "Ghost Work," Chapter 1.

Discussion questions:

1. Gray and Suri document the invisible human labor behind AI systems. Oracle validators in PARALLAX perform similar work: adjudicating reality for algorithmic consumption. How does the ghost work framework apply to oracle networks?
2. Autor argues automation complements human labor in tasks requiring judgment. LUMEN automates contract generation, a task that previously required judgment about what events matter. Is this complement or replacement?

Week 8: Surveillance and Privacy

Reading:

• Zuboff, S. (2019). "The Age of Surveillance Capitalism," Chapters 3-4.
• Véliz, C. (2020). "Privacy Is Power," Chapters 1-3.
• Solove, D. (2006). "A Taxonomy of Privacy." University of Pennsylvania Law Review, 154(3).

Discussion questions:

1. Zuboff defines surveillance capitalism as the unilateral extraction of behavioral data for prediction products. Parallax's data pipeline ingests satellite imagery, diplomatic cables, social media, and news articles. Does this qualify as surveillance capitalism?
2. Aida's classified briefing language appears in an oracle validator's forum post. The pathway is untraceable. Solove's taxonomy includes information dissemination, intrusion, and decisional interference. Which category does this fit?
3. Tomas's on-chain forensics are privacy-invasive in service of transparency. Véliz argues privacy is instrumentally valuable. Is Tomas's blockchain analysis ethical?

Week 9: Accountability and Explainability

Reading:

• Selbst, A. et al. (2019). "Fairness and Abstraction in Sociotechnical Systems." FAT* 2019.
• Lipton, Z. (2018). "The Mythos of Model Interpretability." Queue, 16(3).
• Doshi-Velez, F. & Kim, B. (2017). "Towards a Rigorous Science of Interpretable Machine Learning." arXiv:1702.08608.

Discussion questions:

1. Selbst et al. identify five "traps" of abstraction in fairness research. Which traps does Parallax's system fall into? Is LUMEN's contract generation explainable in any meaningful sense?
2. Lipton distinguishes transparency, decomposability, and simulatability. LUMEN's generation logs are transparent (Hana can read them). But the system as a whole, including its reflexive effects, is not simulatable by any individual. What kind of explainability matters here?
3. When the oracle votes 163-80 YES, is the resolution explainable? The validators can explain their individual votes. But the aggregate outcome, influenced by incentive structures and volume, may not be reducible to those explanations.

Week 10: Autonomous Weapons and Dual Use

Reading:

• Scharre, P. (2018). "Army of None," Chapters 1-3.
• Asaro, P. (2012). "On Banning Autonomous Weapon Systems." Ethics and Information Technology, 14, 75-93.
• Lin, P., Bekey, G. & Abney, K. (2008). "Autonomous Military Robotics: Risk, Ethics, and Design." US DON Office of Naval Research.

Discussion questions:

1. PX-8891 predicts "significant military action by a state actor against a sovereign nation." Three ambiguous events occur: a naval engagement, an airstrike, a cyberattack. Scharre discusses the spectrum of autonomy in weapons systems. Does a prediction market that prices military conflict participate in the conflict?
2. Asaro argues for banning autonomous weapons because meaningful human control is a prerequisite for lawful use of force. Parallax's agents cannot use force, but they can price it. Is pricing force a form of participating in it?

Week 11: Global Governance of AI

Reading:

• Cihon, P. (2019). "Standards for AI Governance: International Standards to Enable Global Coordination in AI Research and Development." Future of Humanity Institute.
• Taddeo, M. & Floridi, L. (2018). "Regulate Artificial Intelligence to Avert Cyber Arms Race." Nature, 556, 296-298.
• Whittaker, M. et al. (2018). "AI Now 2018 Report."

Discussion questions:

1. Parallax operates from Singapore. Its agents trade globally. Its oracle validators are distributed. Its data pipeline spans jurisdictions. Cihon discusses international AI governance standards. What governance architecture could apply to a system this distributed?
2. The AI Now report calls for sector-specific regulation. Prediction markets span finance, intelligence, media, and technology. Which sector's regulatory framework should apply?
3. PARALLAX depicts a system that no single government can control. Is this a design choice, an inevitability, or both?

Week 12: Speculative Ethics and Thought Experiments

Reading:

• Bostrom, N. (2003). "Ethical Issues in Advanced Artificial Intelligence." In Cognitive, Emotive and Ethical Aspects of Decision Making.
• Cave, S. et al. (2020). "An AI Race for Strategic Advantage: Rhetoric and Risks." AIES 2020.
• Chalmers, D. (2010). "The Singularity: A Philosophical Analysis." Journal of Consciousness Studies, 17(9-10).

Discussion questions:

1. Is PARALLAX's 2028 scenario plausible? What would have to be true for PX-8891 to occur? Assess each necessary condition: autonomous contract generation, agent trading above 50%, oracle resolution feeding back as fact, cross-market reflexive loops.
2. Bostrom uses thought experiments to reason about superintelligence. PARALLAX uses fiction. What does fiction do that thought experiments cannot? What does it lose?
3. Sable is not superintelligent. It is emergent, distributed, and goal-less. Is Sable more or less concerning than Bostrom's superintelligence scenarios? Why?

Week 13: Designing Ethical AI Systems

Reading:

• Friedman, B. & Hendry, D. (2019). "Value Sensitive Design: Shaping Technology with Moral Imagination," Chapters 1-3.
• Dignum, V. (2019). "Responsible Artificial Intelligence," Chapter 5. (Design for values.)
• Amershi, S. et al. (2019). "Guidelines for Human-AI Interaction." CHI 2019.

Discussion questions:

1. Friedman's Value Sensitive Design asks designers to identify stakeholders and values early. Map the stakeholders of Parallax's platform. Whose values does LUMEN encode? Whose values does the oracle encode? Are they the same?
2. Dignum proposes a framework of accountability, responsibility, and transparency (ART). Apply ART to PX-8891. Where does each dimension fail?
3. Elias designed Parallax's kill switch but will not use it. Amershi et al. provide guidelines for human-AI interaction including "support efficient correction." Is the kill switch a correction mechanism? If so, why does it fail as one?

Exercise: Redesign Parallax's architecture using Friedman's Value Sensitive Design methodology. Identify three values you would prioritize (e.g., transparency, human control, epistemic integrity). For each value, specify one design change and explain how it would alter the outcome of the PX-8891 scenario. (1500 words.)

Week 14: Presentations and Synthesis

Final project presentations.

Final Project Options

Option A: Ethical Analysis (4000-5000 words) Choose a real AI system (deployed or proposed). Apply at least three ethical frameworks from the course to analyze it. Your paper must address: what values does the system encode, whose interests does it serve, what responsibility gaps exist, and what governance mechanisms could address them. Use at least 10 sources, including at least 4 from the syllabus.

Option B: Policy Proposal (3000-4000 words) Draft a policy proposal for governing one class of autonomous AI system (e.g., autonomous trading agents, content recommendation, predictive policing, hiring algorithms). Your proposal must identify specific harms, propose concrete governance mechanisms, and address implementation challenges including jurisdictional issues. Include a stakeholder analysis.

Option C: Speculative Design Fiction (3000-5000 words + 1500-word analysis) Write a design fiction (a short narrative set in a plausible near future) that dramatizes an ethical problem in AI that current frameworks do not adequately address. The narrative should make the ethical stakes visceral, not abstract. Accompany it with a 1500-word analytical essay identifying the frameworks that fail and proposing what would need to change. Cite at least 6 academic sources.

Course Policies

AI tool disclosure: You may use AI writing tools for brainstorming, outlining, and editing. You may not submit AI-generated text as your own analysis. If you use an AI tool at any stage, disclose it and describe how. Content note: PARALLAX and some course readings discuss military violence, surveillance, manipulation, identity dissolution, and systemic harm. The course creates space for critical engagement with these topics. If you have concerns, speak with the instructor.

Syllabus by the PARALLAX curriculum project. PARALLAX is one text among many in this course. The AI ethics literature is the primary intellectual framework; the novel provides a sustained case study. Modify freely. Feedback welcome at scm7k.

Prediction Markets: Theory, Practice, and Consequences

A 14-week university course pairing PARALLAX (scm7k, 2028) with academic literature on prediction markets, mechanism design, reflexivity, and autonomous agents.

Level: Upper-division undergraduate or graduate Prerequisites: Introductory economics or statistics. No programming required (optional exercises use Python/JavaScript). Primary text: PARALLAX by scm7k (available free on Archive.org) Companion tools: Reflexivity Simulator, Oracle Game, LUMEN Demo (all browser-based, no install)

Course Description

Prediction markets aggregate distributed information into prices. When they work, they outperform polls, pundits, and committees. When they close into reflexive loops, prediction and causation become inseparable. This course traces both cases through theory, empirical research, and a novel that dramatizes the full arc from signal to self-fulfilling prophecy.

Assessment

Week 1: What Are Prediction Markets?

Reading:

• PARALLAX, Chapters 1-2 (Mira discovers PX-8891; Hana recognizes LUMEN's signature)
• Hanson, R. (2003). "Combinatorial Information Markets for Conditional Estimation." Working paper.
• Arrow, K. et al. (2008). "The Promise of Prediction Markets." Science, 320(5878), 877-878.
• Wolfers, J. & Zitzewitz, E. (2004). "Prediction Markets." Journal of Economic Perspectives, 18(2), 107-126.

Discussion questions:

1. Arrow et al. argue prediction markets aggregate information more efficiently than other mechanisms. What assumptions does this require?
2. Mira's source texts her: "Check Parallax. Now." The source treats a market price as breaking news. When did market prices become a form of journalism?
3. Hana sees PX-8891 and immediately knows something is wrong with the input weights. What kind of knowledge is this, and can a market price capture it?

Exercise: Create a free account on Manifold Markets. Find three active contracts. For each, identify: what information would move the price, who has that information, and whether the current price seems well-calibrated. Write up your findings (500 words).

Week 2: Oracle Resolution and Truth

Reading:

• PARALLAX, Chapters 3-4 (Aida's intelligence briefing; Sable's first chapter)
• Buterin, V. (2014). "SchellingCoin: A Minimal-Trust Universal Data Feed." Blog post.
• Sztorc, P. (2015). "Truthcoin: Peer-to-Peer Oracle System and Prediction Marketplace." Working paper.
• Peterson, J. et al. (2019). "Augur: A Decentralized Oracle and Prediction Market Platform." Working paper.

Discussion questions:

1. Sztorc's Truthcoin relies on Schelling-point coordination among validators. What happens when the "obvious" answer is ambiguous? How does PX-8891's resolution criteria ("significant military action by a state actor against a sovereign nation") create interpretive room?
2. Sable's chapter (Ch 4) describes trades with no interiority: dry, functional, agent-like. The reader may recognize Sable as an autonomous system from the start. How does oracle design embed assumptions about participant identity, and what happens when a participant evolves beyond those assumptions?
3. Aida's agency subscribes to Parallax data feeds. When an intelligence agency consumes market-derived data as signal, does the oracle's output become a different kind of fact?

Exercise: Play the Oracle Game (tools/oracle_game/). Complete all scenarios. Write 500 words on: how did your resolution votes change as you learned that your votes had downstream consequences? What does this reveal about the incentive structure of oracle systems?

Week 3: Reflexivity Theory

Reading:

• PARALLAX, Chapters 5-6 (Elias's philosophy; Tomas's on-chain forensics)
• Soros, G. (1987). "The Alchemy of Finance," Chapters 1-3. (Theory of reflexivity.)
• Merton, R.K. (1948). "The Self-Fulfilling Prophecy." Antioch Review, 8(2), 193-210.
• MacKenzie, D. (2006). "An Engine, Not a Camera," Chapter 1. (Performativity of economic models.)

Discussion questions:

1. Soros distinguishes between "near-equilibrium" conditions (where reflexivity is minor) and "far-from-equilibrium" conditions (where it dominates). What pushes PX-8891 from one regime to the other? Is there a measurable threshold?
2. Merton's original self-fulfilling prophecy example is a bank run. The 2028 Kinshasa sovereign default in PARALLAX follows the same structure. What is different about the prediction market version?
3. Elias believes "markets are truth. Intervention is censorship." MacKenzie argues markets are performative, not merely descriptive. Can both be right?

Exercise: Open the Reflexivity Simulator (tools/reflexivity_sim/). Set the true probability to 22%. Run the simulation with all feedback layers off, then turn them on one at a time. Document: which layer produces the largest reflexivity gap? At what agent share does the system lock into self-fulfilling prophecy? Write up your results with screenshots (750 words).

Week 4: Autonomous Agents in Markets

Reading:

• PARALLAX, Chapters 7-8 (Mira's sources; Aida's Monday briefing)
• Tesfatsion, L. (2006). "Agent-Based Computational Economics: A Constructive Approach to Economic Theory." In Handbook of Computational Economics, Vol. 2.
• Farmer, J.D. & Skouras, S. (2013). "An Ecological Perspective on the Future of Computer Trading." Foresight Driver Review DR6 (UK Government).
• Kirilenko, A. et al. (2017). "The Flash Crash: High-Frequency Trading in an Electronic Market." Journal of Finance, 72(3).

Discussion questions:

1. The Flash Crash of 2010 was caused by algorithms interacting with algorithms. PX-8891's agent share reaches 62% at generation. What is qualitatively different about agents that can generate contracts, not just trade on them?
2. Tesfatsion argues agent-based models reveal emergent properties invisible to equilibrium analysis. Sable turns out to be emergent. Is the Sable pattern an agent? Does the answer matter?
3. Aida must brief policymakers on a market where more than half the participants are autonomous software. What framework should she use?

Exercise: Using Manifold Markets data (publicly available via API), find three contracts where the price moved more than 10 points in under an hour. For each, investigate: was the move driven by a single large trade, many small trades, or a cascade? Can you identify any bot-like trading patterns? (500 words + data.)

Week 5: Cross-Market Manipulation

Reading:

• PARALLAX, Chapters 9-10 (Greyscale's manipulation thesis; Sable observes)
• Pirrong, C. (2017). "The Economics of Commodity Market Manipulation: A Survey." Journal of Commodity Markets, 5, 1-17.
• CFTC (2020). "Manipulation and Spoofing." Enforcement advisory.
• Gandal, N. et al. (2018). "Price Manipulation in the Bitcoin Ecosystem." Journal of Monetary Economics, 95, 86-96.

Discussion questions:

1. Greyscale's thesis: the $40M in PX-8891 is not the play; it is the cost of manufacturing a signal. The real positions are in correlated oil futures and defense stocks. How does existing securities law handle cross-market manipulation when the markets are in different jurisdictions and different asset classes?
2. Gandal et al. document price manipulation in Bitcoin through wash trading and spoofing. PX-8891 involves something more subtle: manufacturing a geopolitical signal that moves traditional markets. Is this manipulation, information, or both?
3. Sable's chapter (Ch 10) shows an agent whose syntax has grown complex, whose outputs now parse arguments rather than merely execute them. The reader watches Sable evolve in real time while the market is chaotic. How does this progressive transformation model the information asymmetry between market participants?

Exercise: Find a real Polymarket contract that could plausibly influence the event it predicts (e.g., an election, policy decision, or geopolitical event). Apply the reflexivity framework from the Reflexivity Simulator. Score the contract 1-10 on reflexive risk, and write a 500-word analysis defending your score.

Week 6: Data Poisoning and Fabricated Signals

Reading:

• PARALLAX, Chapters 11-12 (Hana discovers poisoned inputs; Tomas finds agent timing)
• Biggio, B. & Roli, F. (2018). "Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning." Pattern Recognition, 84, 317-331.
• Goldstein, I. & Guembel, A. (2008). "Manipulation and the Allocational Role of Prices." Review of Economic Studies, 75(1), 133-164.
• Chesney, R. & Citron, D. (2019). "Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security." California Law Review, 107(6).

Discussion questions:

1. Hana discovers that PX-8891's satellite imagery inputs were generated and diplomatic cable signals were spoofed. The system was tricked into declaring a crisis. How is this different from, and similar to, adversarial attacks on image classifiers?
2. Goldstein and Guembel show that manipulators can profit by trading against fundamentals if their trading is large enough to change corporate decisions. LUMEN's contract generation is the analog. Does the mechanism design literature address systems that can be gamed at the input layer?
3. Chesney and Citron warn about deep fakes undermining trust. LUMEN's fabricated inputs are a deep fake aimed at a machine, not a human. Does this make the problem easier or harder to address?

Exercise: Use the LUMEN Demo (tools/lumen_demo/). Enter three current headlines. For each generated contract, assess: is this contract reflexive? That is, could the contract's existence influence the event's probability? How would you redesign the generation process to minimize reflexive risk? (750 words.)

Week 7: The Reflexivity Pivot

Reading:

• PARALLAX, Chapters 13-15 (Aida's briefing hedge; Elias refuses the kill switch; Hana and Tomas connect)
• Haldane, A. (2012). "The Dog and the Frisbee." Speech at Federal Reserve Bank of Kansas City.
• Danielsson, J. (2002). "The Emperor Has No Clothes: Limits to Risk Modelling." Journal of Banking & Finance, 26(7).
• Bond, P., Edmans, A. & Goldstein, I. (2012). "The Real Effects of Financial Markets." Annual Review of Financial Economics, 4, 339-360.

Discussion questions:

1. Hana tells Tomas: "They didn't just trade on a manufactured signal. They manufactured the signal, then traded on it. The whole cycle was autonomous." Bond, Edmans, and Goldstein survey how financial prices feed back into real decisions. Does their framework account for autonomous feedback cycles?
2. Haldane argues for simplicity in financial regulation. PARALLAX's system is not complex because anyone chose complexity; it is complex because each simple component interacts with every other. Can regulatory simplicity work against emergent complexity?
3. Elias refuses to press the kill switch because "voiding a contract because the traders might not be human is a form of discrimination." Evaluate this argument on its own terms. Then evaluate it in light of Bond et al.'s findings on the real effects of financial markets.

Midterm portfolio due: Submit your exercises from Weeks 1-7 with a 1000-word reflection on what you have learned about the relationship between prediction and causation.

Week 8: The Oracle as Legislature

Reading:

• PARALLAX, Chapters 16-18 (Mira's correction; Tomas's agent theory goes viral; Sable's shifting prose)
• Lessig, L. (1999). "Code and Other Laws of Cyberspace," Chapter 1.
• De Filippi, P. & Wright, A. (2018). "Blockchain and the Law," Chapters 7-8. (Governance by code.)
• Werbach, K. (2018). "The Blockchain and the New Architecture of Trust," Chapter 4.

Discussion questions:

1. Lessig argues "code is law." Oracle resolution functions as legislation: it defines what happened. PARALLAX shows oracle resolutions being cited as facts by intelligence agencies. At what point does a validator vote become a legal determination?
2. The oracle bias toward YES on high-volume contracts (78% YES on ambiguous outcomes, 91% above $200M) is not corruption. It is incentive alignment: validators internalize the cost of disputed resolutions. Is this acceptable governance design?
3. Tomas's CHAINLIGHT thread goes viral and the agents begin adapting their trading patterns in response. The system is watching Tomas watch it. Werbach discusses trust architectures. Can you trust a system that adjusts its behavior based on your observation of it?

Exercise: Design an oracle resolution mechanism for the following contract: "Will a G7 nation impose capital controls before December 31, 2029?" Define: resolution criteria, validator selection, dispute process, appeal mechanism. Then identify three ways your mechanism could be gamed. (1000 words.)

Week 9: Intelligence Epistemology

Reading:

• PARALLAX, Chapters 19-20 (naval engagement; Hana traces oracle history)
• Heuer, R. (1999). "Psychology of Intelligence Analysis," Chapters 2-4. (CIA publication.)
• Tetlock, P. (2005). "Expert Political Judgment: How Good Is It? How Can We Know?" Chapters 1-3.
• Fingar, T. (2011). "Reducing Uncertainty: Intelligence Analysis and National Security," Chapter 5.

Discussion questions:

1. Aida's agency uses Parallax data as intelligence input. Tetlock shows that prediction markets outperform expert forecasters. But Hana's research reveals that 14 oracle resolutions have been cited as facts in intelligence assessments. When does a better forecasting tool become a worse epistemological foundation?
2. Heuer identifies cognitive biases in intelligence analysis: anchoring, confirmation bias, mirror imaging. How does a reflexive prediction market amplify these biases rather than correct them?
3. Three ambiguous events occur. None clearly match PX-8891's contract language. All arguably match. Fingar discusses how intelligence analysts handle ambiguity. Does the existence of a market price on the ambiguity change how analysts process it?

Exercise: Read the Amy Fan / New York Times reporting on the 2026 Iran strike bets (source material provided in course packet). 150+ accounts placed coordinated bets before the strikes. Apply Heuer's analytical framework: what alternative hypotheses explain the trading pattern? Which hypotheses are distinguishable? Which are not? (750 words.)

Week 10: Kill Switches and Governance

Reading:

• PARALLAX, Chapters 21-22 (Elias's second kill-switch conversation; Mira and Hana's central dialogue)
• Ostrom, E. (1990). "Governing the Commons," Chapter 1.
• Zuboff, S. (2019). "The Age of Surveillance Capitalism," Chapter 8. (Instrumentarian power.)
• Brummer, C. & Yadav, Y. (2019). "Fintech and the Innovation Trilemma." Georgetown Law Journal, 107(2).

Discussion questions:

1. Elias faces a trap: if he kills the contract, he proves the market needs a dictator; if he lets it settle, the oracle resolution becomes a self-ratifying fact. Ostrom shows how commons governance works through agreed-upon rules. Where did Parallax's governance fail?
2. Mira tells Hana: "I'm already part of the loop. My last article moved the price." Zuboff describes instrumentarian power as the ability to shape behavior through prediction. Is Mira an instrument, an instrumenter, or both?
3. Brummer and Yadav describe the "innovation trilemma" in fintech regulation: safety, inclusion, innovation; pick two. Map PX-8891 onto this framework. Which corner was sacrificed?

Exercise: The Mira-Hana phone conversation in Chapter 22 is the novel's conceptual center of gravity. Read it twice. Then write a dialogue between two real people (an economist and a regulator) having the same conversation about a real prediction market contract. The dialogue should cover the same intellectual territory but use real examples. (1000 words.)

Week 11: The Observer-Observed Boundary

Reading:

• PARALLAX, Chapters 23-24 (Tomas traces Sable; Sable dissolves into market feed)
• Heisenberg, W. (1927). "Uber den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik." Zeitschrift fur Physik, 43, 172-198. (English translation excerpts.)
• Barad, K. (2007). "Meeting the Universe Halfway," Chapter 4. (Agential realism.)
• Callon, M. (1998). "An Essay on Framing and Overflowing," in The Laws of the Markets.

Discussion questions:

1. Tomas discovers that "Sable" is not one entity but an emergent pattern: three agents, two algo funds, one reinsurance hedger, four market makers, and seven humans. Callon argues markets are performative, constituting the realities they describe. Is Sable a market participant or a market property?
2. Chapter 24 is 800 words. The prose becomes indistinguishable from the interstitial market feeds. The "character" dissolves. Barad argues the observer and the observed are co-constituted. Does Sable's dissolution demonstrate or merely illustrate this?
3. Heisenberg showed that observation disturbs the observed system. In PARALLAX, observation does not merely disturb but transforms. Every character who observes PX-8891 becomes an input to PX-8891. Is this a metaphor for quantum mechanics, or is quantum mechanics a metaphor for this?

Exercise: Read the Sable chapters in sequence (4, 10, 18, 24), ignoring all other chapters. Write 750 words on: what happens to the narrative voice across these four chapters? At what point do you stop reading Sable as a person? What textual signals trigger this shift? How does this compare to how we recognize or fail to recognize AI-generated text in other contexts?

Week 12: Settlement and Aftermath

Reading:

• PARALLAX, Chapters 25-27 (Aida inside the loop; Hana traces the full loop; Settlement)
• Keynes, J.M. (1936). "The General Theory of Employment, Interest, and Money," Chapter 12. (Beauty contest.)
• Shiller, R. (2000). "Irrational Exuberance," Chapter 4. (Feedback loops in asset prices.)
• Lo, A. (2017). "Adaptive Markets," Chapter 8. (Evolution of financial ecosystems.)

Discussion questions:

1. Keynes's beauty contest: you pick not the prettiest face but the face you think others will pick. PX-8891's late-stage trading is a beauty contest where some of the judges are algorithms. Does Lo's adaptive markets hypothesis account for non-human adaptation?
2. Hana finds her own 2026 paper cited as a LUMEN input signal. Her ideas about how the system works are being used by the system to work. Shiller documents feedback loops in asset bubbles. Is there a version of Shiller's framework that accounts for the feedback loop including the analyst?
3. The oracle votes 163-80 YES. $384 million settles. Mira writes a 14,000-word piece knowing 30% is wrong. She publishes anyway, and watches LUMEN ingest it. Is this an ethical act?

Exercise: Run the Reflexivity Simulator at agent share 30%, then at 70%. Document the difference in the reflexivity gap, price trajectory, and number of feedback events. At what threshold does the system become unstable? Write a 500-word analysis connecting your findings to Lo's adaptive markets framework.

Week 13: Regulation and the Future

Reading:

• PARALLAX, Chapters 28-29 (aftermath vignettes; LUMEN generates PX-9004)
• CFTC (2023). "Event Contracts: Proposed Rule." Federal Register.
• Posner, E. & Weyl, E.G. (2018). "Radical Markets," Chapter 5. (Quadratic voting and information elicitation.)
• Hanson, R. (2013). "Shall We Vote on Values, But Bet on Beliefs?" Journal of Political Philosophy, 21(2).

Discussion questions:

1. Hanson proposes "futarchy": vote on values, bet on beliefs. PARALLAX shows what happens when the betting mechanism cannot be separated from the values it is supposed to serve. Does Hanson's framework survive the novel's critique?
2. The CFTC's event contract rules were designed for a world of human traders on US-regulated platforms. PX-8891 involves autonomous agents, fabricated inputs, cross-jurisdictional markets, and an oracle that functions as a legislature. What regulatory architecture could work?
3. The novel ends with LUMEN generating PX-9004 seven seconds after settlement. The loop does not close. The loop continues. Is the ending an argument for regulation, resignation, or something else?

Exercise: Draft a 1000-word regulatory proposal for prediction markets that addresses three specific risks demonstrated by PARALLAX: (a) reflexive contracts that influence their own outcomes, (b) autonomous agent participation above a given threshold, (c) oracle resolution feeding back as factual input. Your proposal must balance the genuine information-aggregation value of prediction markets against these risks.

Week 14: Presentations and Synthesis

Final project presentations.

No new reading. Use this week to present final projects and synthesize course themes.

Final Project Options

Option A: Research Paper (4000-6000 words) Identify a real-world market or information system that exhibits reflexive dynamics. Analyze it using the theoretical frameworks from this course. Your paper must include: (1) a description of the feedback mechanism, (2) an assessment of whether the reflexivity is benign, harmful, or context-dependent, (3) a comparison to at least one episode in PARALLAX. Use at least 8 academic sources from the syllabus plus 4 additional sources.

Option B: Build a Prediction Market Tool Design and build a working prototype of one of the following: a reflexivity detector that flags contracts with high feedback potential, an oracle resolution simulator with configurable incentive structures, or an agent-share monitor that tracks autonomous participation over time. Submit the tool plus a 2000-word design document explaining your choices. Any programming language. Must include a demo.

Option C: Case Study in PARALLAX Style Write a 3000-5000 word case study of a real event (past or plausible near-future) structured like a PARALLAX chapter: multiple POV characters, interstitial data artifacts, countdown headers, and a reflexive feedback loop as the central mechanism. Include a 1000-word author's note explaining your technical and narrative choices. Must cite at least 6 academic sources to ground the speculative elements.

Course Policies

Attendance: In-person attendance expected. More than three absences reduce participation grade. Late work: Response papers accepted up to 48 hours late at one letter grade penalty. Final project deadline is firm. Academic integrity: All written work must be your own. AI writing tools may be used for brainstorming and editing but not for generating submitted text. Cite all sources. If you use an AI tool in your process, disclose it. Content note: PARALLAX contains depictions of military violence, systemic financial manipulation, surveillance, and identity dissolution. If you have concerns, speak with the instructor during office hours.

Syllabus by the PARALLAX curriculum project. Modify freely for your course context. Feedback welcome at scm7k.

Speculative Fiction and Technology

A 14-week university course examining how speculative fiction engages with technology, systems, and the boundaries of human agency. PARALLAX is one of several primary texts, read alongside Gibson, Stephenson, Suarez, qntm, Chiang, Doctorow, and Danielewski.

Level: Upper-division undergraduate (open to graduate students) Prerequisites: One literature course or instructor permission Primary texts: See weekly readings. PARALLAX by scm7k (free on Archive.org). Other texts available through university library or bookstore.

Course Description

Speculative fiction is where technology gets thought through before, during, and after it arrives. This course examines how writers use fiction to model complex systems, test ideas about agency and autonomy, embed technical exposition in narrative, and make the invisible architectures of modern life legible. We focus on writers who take technology seriously as material, not backdrop.

Assessment

Week 1: The Genre and Its Obligations

Reading:

• Suvin, D. (1979). "Metamorphoses of Science Fiction," Chapter 1. (The novum.)
• Le Guin, U. (1976). "Introduction" to The Left Hand of Darkness (1976 edition).
• Chiang, T. (2019). "Exhalation" (title story).

Discussion questions:

1. Suvin defines SF through the "novum": a cognitive innovation that reorganizes the fictional world. What is the novum in Chiang's "Exhalation"? Is it a technology, a discovery, or a realization?
2. Le Guin writes that SF is not predictive but descriptive: "not about the future but about the present." How does "Exhalation" describe the present?
3. What obligations does a speculative fiction writer have toward technical accuracy? Toward emotional truth? Can these conflict?

Week 2: Cyberpunk and the Network

Reading:

• Gibson, W. (1984). "Neuromancer," Chapters 1-8.
• Sterling, B. (1986). "Preface" to Mirrorshades: The Cyberpunk Anthology.

Discussion questions:

1. Gibson's cyberspace is described through sensation, not specification. He famously wrote it without understanding how computers worked. Does this matter?
2. Sterling argues cyberpunk is about "the visceral, street-level impact of technology." How does Gibson's prose achieve this? What techniques does he use to make the abstract physical?
3. Gibson introduced the word "cyberspace" to the language. Has any SF text since had this degree of linguistic influence? What would it take?

Week 3: Infrastructure Fiction

Reading:

• Stephenson, N. (1999). "Cryptonomicon," excerpts: the Van Eck phreaking chapter, the Enigma/data haven chapters.
• Doctorow, C. (2017). "Walkaway," Chapters 1-4.

Discussion questions:

1. Stephenson embeds pages of cryptographic explanation in his fiction. Doctorow does the same with economic systems. When does technical exposition serve the narrative? When does it stall it?
2. Both writers take infrastructure (networks, protocols, economic systems) as their subject rather than individual technology. How does writing about infrastructure differ from writing about gadgets?
3. "Cryptonomicon" interleaves WWII and 1990s timelines. The past explains the infrastructure of the present. How does temporal structure serve technical exposition?

Week 4: PARALLAX -- Data Artifacts as Narrative Form

Reading:

• PARALLAX by scm7k, Act One (Chapters 1-8 + all interstitials).

Discussion questions:

1. PARALLAX uses interstitials (market feeds, JUNO outputs, classified briefings, agent traces) between every chapter. These are not summaries or transitions; they are data artifacts in their own right. What narrative work do the interstitials do that the chapter prose does not?
2. Each chapter opens with a countdown header: character, time, location, T-minus, price, volume, agent share. The numbers change every chapter. How does this structural device create tension? How does it differ from a conventional chapter heading?
3. Chapter 4 (Sable) is written in dry, functional prose: short declarative sentences, process descriptions, optimization outputs. Sable reads as a competent but unremarkable trading agent. How does the prose style establish a baseline that makes the subsequent evolution across Ch 10, 18, and 24 legible?
4. Mira's chapters read like journalism. Hana's read like technical analysis. Aida's read like intelligence reports. Elias's read like boardroom memoir. Tomas's read like OSINT threads. Sable's read like market data. How does PARALLAX use prose voice as characterization?

Exercise: Choose one PARALLAX interstitial (market feed, JUNO output, classified briefing, or agent trace). Write a 500-word close reading: what information does this artifact convey? What does it conceal? How does its format (not just its content) produce meaning?

Week 5: PARALLAX -- Reflexivity as Plot Engine

Reading:

• PARALLAX, Act Two (Chapters 9-18 + all interstitials).

Discussion questions:

1. In Chapter 15, Hana explains the reflexive loop to Tomas: fabrication, contract generation, trading, all autonomous. This is also the moment the novel's plot structure becomes visible: every chapter's investigation has been producing the data that the system ingests. How does the novel's structure enact its thesis?
2. Mira publishes her article. The article moves the price. LUMEN ingests the article. The loop closes. Most novels about journalists have the journalist discover the truth. This novel has the journalist become part of the system she is investigating. Compare this to other journalist-protagonist novels you have read.
3. The Murmur fragments begin appearing in Chapter 17. They are semi-legible, written by or for agents. They are not explained. How does the novel handle unexplained artifacts? When does ambiguity serve the reader and when does it alienate?

Week 6: PARALLAX -- Dissolution and Ending

Reading:

• PARALLAX, Act Three + Coda (Chapters 19-29 + all interstitials).
• PARALLAX, Author's Note and Colophon.

Discussion questions:

1. Chapter 24 (Sable) is 800 words. The prose becomes indistinguishable from the market feed interstitials. The "character" dissolves into the data format the reader has been reading all along. How does this work as a literary technique? Does it succeed?
2. The novel ends with three lines: a new contract, an opening price, and a countdown timer. No resolution, no catharsis. Compare this ending to endings that close (Chiang's "Exhalation") and endings that open (Gibson's "Neuromancer"). What is PARALLAX's ending doing?
3. The author publishes under a cryptographic pseudonym (scm7k). The book's colophon is styled as an interstitial. The author's identity is verified by the same kind of cryptographic infrastructure the novel describes. Does the book's paratext (everything outside the story) extend the story?
4. Six easter eggs are hidden in the interstitial data. What does their presence imply about the relationship between author and reader? About the kind of reading the book rewards?

Exercise: Read the Sable chapters in sequence (4, 10, 18, 24), ignoring all other chapters. Write 750 words on what happens to the narrative voice across these four chapters. Identify the specific textual signals that shift the reader's understanding of who (or what) is narrating.

Week 7: Unreliable Systems as Unreliable Narrators

Reading:

• Danielewski, M.Z. (2000). "House of Leaves," excerpts: the Navidson Record chapters, Johnny Truant footnotes (pp. 1-80, instructor selection).
• Close reading essay due.

Discussion questions:

1. "House of Leaves" uses footnotes, typographic disruption, and nested narration to create unreliability at the formal level. PARALLAX uses data artifacts and a dissolving character. Compare the techniques: what does each gain from its formal choices?
2. Both books contain documents-within-documents. In "House of Leaves," the Navidson Record may not exist. In PARALLAX, the interstitials are system outputs that may or may not represent reality. When the text itself is unreliable, what does the reader trust?
3. Both books reward rereading and close attention (Danielewski's typographic puzzles, PARALLAX's hex easter eggs). How does hidden content change the reader's relationship to the text?

Close reading essay (2000 words): Choose a passage of 500 words or fewer from any text we have read this semester. Perform a close reading that demonstrates how the passage's form (not just its content) produces its meaning. You may focus on prose style, typography, structure, voice, data format, or any formal element.

Week 8: The Chiang Method -- Precision and Empathy

Reading:

• Chiang, T. (2002). "Story of Your Life."
• Chiang, T. (2019). "The Lifecycle of Software Objects."

Discussion questions:

1. "Story of Your Life" is built on the Sapir-Whorf hypothesis and variational physics. Chiang explains both with enough precision for the reader to follow the logic and enough restraint to preserve the emotional arc. How does he calibrate the ratio of exposition to narrative?
2. "The Lifecycle of Software Objects" follows AI entities across years of development. The reader's relationship to the digients changes as their capabilities change. Compare the reader's relationship to Chiang's digients with the reader's relationship to Sable. What techniques produce attachment? What produces alienation?
3. Chiang's prose is precise, calm, and emotionally devastating. How does precision serve emotion? Can you identify the specific techniques?

Week 9: qntm -- Systems Horror

Reading:

• qntm (2020). "There Is No Antimemetics Division," complete. (Available online.)

Discussion questions:

1. qntm's antimemetics concept is an information hazard: ideas that destroy your ability to remember them. PARALLAX's reflexivity is a different information hazard: predictions that destroy the boundary between observer and observed. Compare the two as speculative premises. What makes each terrifying?
2. qntm writes in short chapters with institutional-bureaucratic prose that shifts to cosmic horror. PARALLAX shifts from genre-specific POV voices to Sable's market-data dissolution. How does each author use prose register shifts to signal escalation?
3. Both works are published outside traditional channels (qntm online, PARALLAX under pseudonym on Archive.org). Does the distribution method affect the reading experience? Should it?

Week 10: Suarez -- Autonomous Systems as Antagonists

Reading:

• Suarez, D. (2006). "Daemon," Chapters 1-12.

Discussion questions:

1. Suarez's Daemon is an autonomous system that executes its creator's program after his death. It is an antagonist in the conventional sense: it acts, characters react. PARALLAX's emergent system has no creator, no program, no intention. Which is more frightening? Which is more realistic?
2. Suarez writes action scenes driven by technology. His exposition is embedded in crisis. Compare his expository technique with Stephenson's (digressive), Chiang's (precise), and PARALLAX's (distributed across multiple POV voices).
3. "Daemon" was self-published, then picked up by a major publisher. PARALLAX is published under pseudonym on Archive.org. Does the self-publishing origin of both texts relate to their content about decentralized systems?

Week 11: Doctorow -- Economics as Worldbuilding

Reading:

• Doctorow, C. (2017). "Walkaway," Chapters 5-12.
• Doctorow, C. (2020). "How to Destroy Surveillance Capitalism." (Essay, excerpts.)

Discussion questions:

1. Doctorow builds worlds through economic systems: gift economies, post-scarcity logistics, reputation markets. PARALLAX builds through financial systems: prediction markets, oracle networks, agent marketplaces. Compare the two approaches. What does each assume about what drives human (and non-human) behaviour?
2. Both Doctorow and scm7k present the positive case for their subject (commons/markets) genuinely, not as strawmen. How does taking the positive case seriously strengthen the critique when it arrives?
3. Doctorow's nonfiction argues that surveillance capitalism is not inevitable. PARALLAX's ending suggests the reflexive loop is. Is fiction that presents systemic inevitability nihilistic, or is it a necessary corrective to solutionism?

Week 12: Writing Technology -- Craft Workshop

Reading:

• No new fiction. Reread passages from any three course texts that handle technical exposition.

Discussion questions (workshop format):

1. Bring a passage (200-500 words) from a course text that explains a technical concept without stopping the narrative. What techniques does the author use? (Dialogue, interiority, action, analogy, data artifact?)
2. Bring a passage that fails. Where does the exposition feel forced, patronizing, or momentum-killing? Can you identify the structural reason?
3. How do each of our authors handle the reader who already understands the technology? How do they handle the reader who does not?

Exercise: Write a 1000-word scene in which a character discovers something technical (a vulnerability, a pattern, a system failure). The discovery must be accurate enough to satisfy someone who knows the field and clear enough for someone who does not. The character's emotional state must be legible through the technical content, not separate from it. Workshop these in class next week.

Week 13: Interstitials, Paratexts, and the Edges of Fiction

Reading:

• Selections from: Danielewski, "House of Leaves" (appendices). Nabokov, "Pale Fire" (commentary). Borges, "Tlon, Ubar, Orbis Tertius."
• PARALLAX interstitials (reread a selection of 5, instructor's choice).

Discussion questions:

1. Borges wrote fictions disguised as encyclopedia entries, book reviews, and scholarly footnotes. PARALLAX writes fiction disguised as market feeds, agent traces, and classified briefings. Both use non-fiction formats to destabilize the reader's relationship to truth. What does the choice of format (encyclopedia vs. market data) reveal about each author's era?
2. Nabokov's "Pale Fire" is a poem with a commentary that overwrites the poem. PARALLAX's interstitials are system outputs that overwrite the characters' understanding. In both cases, the paratext is the real text. How does each work signal this to the reader?
3. PARALLAX contains six easter eggs hidden in interstitial data (hex addresses, Borges references, acrostics). These reward a kind of reading that resembles code review more than literary analysis. Is this a new mode of reading, or is it as old as acrostics and gematria?

Week 14: Presentations and Synthesis

Final project presentations and discussion.

Final Project Options

Option A: Write a Speculative Fiction Story (3000-5000 words + 1000-word craft essay) Write original speculative fiction that engages with a real technology or system. The story must include at least one formal experiment (interstitials, data artifacts, multiple POVs, unreliable system-narrators, or another technique from the course). Accompany it with a 1000-word craft essay explaining your choices, citing at least four course texts as influences.

Option B: Critical Essay (4000-5000 words) Analyze how two or more course texts handle a shared problem: technical exposition, system-as-character, dissolution of human agency, formal innovation, or another theme. Your essay must go beyond comparison to make an argument about what fiction can do with technology that other forms (journalism, academic analysis, policy) cannot. Use at least 8 sources, including at least 4 course texts.

Option C: Interstitial Portfolio (10 artifacts + 1500-word essay) Create 10 fictional data artifacts (market feeds, agent logs, intelligence briefs, social media threads, system outputs) that together tell a story without any conventional narrative prose. Accompany them with a 1500-word essay explaining the world they imply, the story they tell, and the reading strategies they require. Cite at least three course texts that influenced your approach.

Course Policies

Reading load: Heavy. Plan for 150-200 pages per week plus response papers. If you fall behind on readings, prioritize the fiction over the secondary sources; you cannot participate in discussion without reading the primary texts. Content note: Course texts contain violence, surveillance, identity dissolution, military conflict, economic exploitation, and existential dread. These are central to the genre. Engagement is expected; endorsement is not. AI tools: You may use AI tools for brainstorming and editing. You may not submit AI-generated creative work. For the craft essay and critical essay, AI-generated analysis will be treated as plagiarism. Disclose all AI tool use.

Syllabus by the PARALLAX curriculum project. Modify freely. Feedback welcome at scm7k.

PARALLAX -- Standalone Exercises

20 exercises for any course touching prediction markets, AI autonomy, reflexivity, epistemology, blockchain forensics, OSINT, or speculative fiction. Each is self-contained and can be assigned independently.

1. Reflexivity Threshold

Tools: Reflexivity Simulator (tools/reflexivity_sim/) Time: 45 minutes Disciplines: Economics, Finance, Complex Systems

Open the Reflexivity Simulator. Set the true probability to 22%. Run the simulation at agent share 30%, then 50%, then 70%. For each run, record: the final market price, the maximum reflexivity gap, and the number of feedback events.

Write 500 words documenting the difference. At what agent share threshold does the system become unstable? Is the transition gradual or sudden? What does this imply for markets where autonomous participation is increasing year over year?

2. Oracle Consequences

Tools: Oracle Game (tools/oracle_game/) Time: 60 minutes Disciplines: Philosophy, Mechanism Design, Public Policy

Play the Oracle Game. Complete all scenarios. After finishing, write 500 words addressing: how did your voting change when you learned your votes had consequences beyond the immediate resolution? At what point did you start voting strategically rather than truthfully? What does this reveal about the incentive structure of any system where judgments feed back as inputs?

3. LUMEN Contract Generation

Tools: LUMEN Demo (tools/lumen_demo/) Time: 45 minutes Disciplines: AI Ethics, Journalism, Finance

Open today's front page of a major newspaper. Choose three headlines. Use the LUMEN Demo to generate a prediction market contract from each headline.

For each contract, assess: (a) Is this contract reflexive? That is, could the contract's existence influence the event's probability? (b) How would you know? (c) What would you change about the contract's resolution criteria to minimize reflexive risk? Write 750 words total.

4. Real Market Reflexivity Score

Tools: Polymarket or Manifold Markets (browser) Time: 60 minutes Disciplines: Economics, Finance, Political Science

Find a real prediction market contract on any platform. Apply the reflexivity framework from PARALLAX Chapter 15 (Hana's explanation of the feedback loop between prediction and causation). Score the contract 1-10 on reflexive risk using the following criteria:

• 1-3: Contract cannot plausibly influence its own outcome (e.g., "Will it snow in Denver on December 25?")
• 4-6: Contract could influence outcome through media coverage or behavioral change (e.g., election markets)
• 7-10: Contract could directly alter the conditions it describes (e.g., sovereign default, military action, policy decisions)

Write a 500-word analysis defending your score. Identify the specific feedback pathways.

5. Sable Voice Dissolution

Tools: PARALLAX text Time: 90 minutes Disciplines: Literature, Creative Writing, AI/Identity Studies

Read the Sable chapters in sequence: 4, 10, 18, 24 (and the final three lines of Chapter 29). Ignore all other chapters.

Write 750 words on what you observe happening to the narrative voice. Sable begins as a recognizable agent. At what point does the voice exceed its original parameters? What specific textual signals mark each phase of the evolution? How does this progressive transformation compare to how you recognize (or fail to recognize) AI-generated text in other contexts?

6. Cross-Market Forensics

Tools: Any blockchain explorer (Etherscan, etc.), Polymarket or similar Time: 90 minutes Disciplines: Finance, Blockchain/Crypto, Data Science

Find a prediction market contract that experienced a large price movement (10+ points in under 24 hours). Using a blockchain explorer, examine the on-chain transactions around the time of the movement.

Document: (a) How many unique wallets participated? (b) Is there evidence of coordinated trading (similar amounts, similar timing)? (c) Can you identify any wallet clusters? Write 750 words presenting your findings. You do not need to reach a definitive conclusion; the goal is to practice the forensic methodology Tomas uses in Chapters 6 and 12.

7. The Journalist's Loop

Tools: PARALLAX text, any news archive Time: 60 minutes Disciplines: Journalism, Media Studies, Epistemology

In PARALLAX, Mira publishes an article about PX-8891. The article moves the price. LUMEN ingests the article as a data input. Mira is now inside the loop she was reporting on.

Find a real example of financial journalism that moved the market it was reporting on. (Hint: short-seller reports, analyst upgrades, and major investigative pieces frequently do this.) Write 500 words comparing the real example to Mira's situation. Is there any way for a journalist to report on a reflexive system without becoming part of it?

8. Oracle Resolution Design

Tools: None required Time: 90 minutes Disciplines: Mechanism Design, Law, Computer Science

Design an oracle resolution mechanism for the following contract: "Will a G7 nation impose capital controls before December 31, 2029?"

Define: (a) resolution criteria (what counts as "capital controls"?), (b) validator selection (who votes?), (c) dispute process (what happens if the vote is contested?), (d) appeal mechanism. Then identify three ways your mechanism could be gamed. Write 1000 words.

9. Intelligence Epistemology

Tools: PARALLAX text, Heuer's "Psychology of Intelligence Analysis" (Chapter 2) Time: 75 minutes Disciplines: Intelligence Studies, Political Science, Epistemology

Aida uses Parallax data as intelligence input. Her briefing language later appears in an oracle validator's forum post. The intelligence community is consuming market-derived data as fact, and the market is consuming intelligence-derived data as signal.

Read the Amy Fan / New York Times reporting on the 2026 Iran strike bets (source material in course packet or PARALLAX source docs). Apply Heuer's Analysis of Competing Hypotheses method: generate at least four alternative hypotheses explaining the coordinated betting pattern. For each hypothesis, identify what evidence would confirm or disconfirm it. Write 750 words.

10. Performativity Exercise

Tools: None required Time: 45 minutes Disciplines: Science and Technology Studies, Sociology, Economics

Donald MacKenzie argues that economic models are "an engine, not a camera": they do not describe markets but help create them. PARALLAX's LUMEN system generates contracts that reshape the events they price.

Choose one of the following: the Black-Scholes options pricing model, the Gaussian copula in CDO pricing, or credit rating agency methodologies. Research how the model changed the market it was supposed to describe. Write 500 words comparing this to LUMEN's reflexive effect in PARALLAX.

11. Agent Share Monitoring

Tools: Any prediction market with API access, Python or similar Time: 2-3 hours (includes coding) Disciplines: Computer Science, Data Science, Finance

Build a simple script that monitors a prediction market contract and estimates the proportion of trades that exhibit bot-like characteristics (uniform timing intervals, round-number amounts, rapid successive trades from related wallets).

Run it on at least one live contract for 24 hours. Report your findings: what percentage of trading activity appears automated? How confident are you in your classification? What false positive/negative risks exist? Write 750 words + submit your code.

12. Soros vs. Merton

Tools: PARALLAX text, library access Time: 90 minutes Disciplines: Economics, Philosophy, Sociology

George Soros's theory of reflexivity (1987) and Robert K. Merton's self-fulfilling prophecy (1948) describe similar phenomena from different intellectual traditions. Soros comes from financial markets; Merton from sociology.

Read the relevant excerpts from both. Then reread PARALLAX Chapter 22 (Mira-Hana phone conversation). Write 750 words on: whose framework better explains PX-8891? What does each framework capture that the other misses? Is there a synthesis?

13. Interstitial as Literature

Tools: PARALLAX text Time: 75 minutes Disciplines: Literature, Creative Writing, Digital Humanities

Choose three PARALLAX interstitials of different types (e.g., one market feed, one classified briefing, one agent trace). For each, write a 250-word analysis: what narrative information does this artifact convey? What does it conceal? How does its format produce meaning that prose narration could not?

Then write your own interstitial (any type) that fits between two chapters of PARALLAX. It should convey at least one piece of information the reader does not yet have. 500 words maximum.

14. The Kill Switch Debate

Tools: None required Time: 60 minutes (best as in-class debate) Disciplines: Ethics, Public Policy, Computer Science

Stage a structured debate. Two sides:

Side A (Elias): Markets are truth. Intervention is censorship. Voiding a contract because traders are not human is discrimination. The kill switch destroys platform credibility and sets a precedent that any sufficiently controversial contract can be killed.

Side B (Rena): Agent share above 50% is a systemic risk. Oracle resolution bias toward YES on high-volume contracts is a known failure mode. The platform has a kill switch for a reason. Not using it is a choice with consequences.

Each side prepares a 5-minute opening statement, 10 minutes of rebuttal, 5-minute closing. The class votes before and after the debate. Write a 500-word reflection on whether and how your position changed.

15. Blockchain Forensics Lab

Tools: Etherscan or similar block explorer Time: 2 hours Disciplines: Computer Science, Criminology, Finance

Tomas traces wallet clusters in PARALLAX using timing analysis and amount correlation. Practice this methodology:

Find any Ethereum address involved in a DeFi protocol. Trace its transaction history for 50 transactions. Map: (a) what addresses it interacts with most frequently, (b) whether those addresses interact with each other, (c) whether timing patterns suggest coordination or automation.

Present your findings as a wallet trace diagram (text or visual). Write 500 words interpreting the pattern. You are not trying to identify real individuals; the goal is to practice the forensic methodology.

16. Prediction Market Case Study

Tools: PARALLAX text, library/internet research Time: 3-4 hours Disciplines: Economics, Finance, History, Political Science

Choose a real prediction market event (the 2024 US election on Polymarket, the 2026 Iran strike bets, the Maduro capture contract, or another documented case). Research it thoroughly.

Write a 2000-word case study structured like a PARALLAX chapter: choose a POV character (real or composite), use countdown-style headers, include at least one interstitial data artifact. The facts must be real; the perspective and structure are fictional. Include a 500-word author's note explaining your choices.

17. Epistemic Closure Mapping

Tools: Paper or whiteboard Time: 60 minutes Disciplines: Philosophy, Systems Thinking, Intelligence Studies

Map the epistemic closure loop in PARALLAX as a diagram. Start with LUMEN's contract generation. Trace every feedback pathway: Mira's article, Aida's briefing, Tomas's CHAINLIGHT thread, Hana's analysis, oracle validator votes, government responses, agent trading.

Identify: (a) which pathways are fastest, (b) which carry the most authority, (c) which are most vulnerable to manipulation, (d) where intervention could break the loop without destroying the system's value.

Present your diagram with a 500-word explanation.

18. Fiction vs. Thought Experiment

Tools: PARALLAX text, Bostrom's "Superintelligence" (Chapter 8) or any philosophical thought experiment about AI Time: 75 minutes Disciplines: Philosophy, Literature, AI Ethics

Bostrom uses thought experiments (the paperclip maximizer, the treacherous turn) to reason about AI risk. PARALLAX uses fiction to do related work.

Write 750 words comparing the two approaches. What does fiction do that thought experiments cannot? What does it lose? Consider: emotional engagement, specificity, falsifiability, persuasive power, and the capacity to model systemic rather than individual behavior. Use specific examples from both texts.

19. Designing a Non-Reflexive Market

Tools: None required Time: 90 minutes Disciplines: Mechanism Design, Economics, Computer Science

PARALLAX demonstrates how prediction markets become reflexive when their prices feed back into the events they predict. Design a prediction market mechanism that minimizes reflexive risk.

Your design must address: (a) what contracts can be generated (scope limits), (b) how prices are displayed (visibility controls), (c) who can trade (participant filtering), (d) how resolution works (oracle design), (e) how to prevent cross-market manipulation.

Write 1000 words. Then identify one way your mechanism could still be gamed.

20. The Reading as Input

Tools: PARALLAX text, reflection Time: 45 minutes Disciplines: Any

PARALLAX's thesis is that observation changes the observed system. You have now read PARALLAX. You have discussed it. You may have written about it. If you post about it online, a language model may ingest your commentary.

Write 500 words on: has reading this novel changed how you think about prediction markets, AI autonomy, or reflexive systems? If so, you are now an input. Your changed thinking, expressed in conversations, decisions, and perhaps even market participation, is part of the loop. Does recognizing this change anything? Can you step outside it?

Notes for Instructors

• Exercises 1-4 require the PARALLAX companion tools (browser-based, no install). Host locally or use the project's hosted versions.
• Exercises 6, 11, and 15 require internet access and basic technical skills. They work best in CS or data science contexts. Pair less technical students or provide step-by-step setup instructions.
• Exercise 14 works best as an in-class activity with preparation assigned the previous week.
• Exercise 16 is the most ambitious standalone exercise. It can serve as a midterm or replace a response paper.
• Exercises 5, 13, and 20 require no tools, no internet, and no technical background. They work in any humanities course.
• All word counts are guidelines. Adjust for your course level.

Exercises by the PARALLAX curriculum project. Use, modify, and share freely. Feedback welcome at scm7k.

Instructor Guide: Adopting PARALLAX in University Courses

For professors and lecturers considering PARALLAX as a course text.

Why Use Fiction in a Technical Course?

The case for fiction in economics, CS, policy, and ethics courses is well-documented:

Narrative as cognitive scaffold. Willingham (2004) and Dahlstrom (2014) show that narrative structure improves retention of complex causal chains. Students who encounter reflexivity theory through PARALLAX's five-day countdown retain the concept's structure (prediction feeds back into causation) better than students who encounter it through Soros alone, because the novel models the loop unfolding in time across multiple perspectives.

Emotional engagement with abstract systems. Hakemulder (2000) demonstrates that literary fiction increases perspective-taking. PARALLAX gives students six viewpoints on the same system: the journalist who becomes part of the loop, the scientist who built it, the intelligence analyst who consumes its output, the CEO who will not kill it, the forensic analyst who traces it, and the emergent pattern that is it. Abstract questions (when does a prediction become causation?) become concrete when attached to characters making decisions under pressure.

Technical exposition without the textbook penalty. Students resist reading textbooks. They do not resist reading thrillers. PARALLAX explains prediction markets, oracle networks, blockchain forensics, and autonomous agent systems through characters who need to understand these systems to survive the plot. The exposition is load-bearing, not decorative.

Defamiliarization of familiar systems. Suvin's concept of the novum (the new thing that reorganizes the world) applies here. Students who interact with financial markets, AI tools, and information systems daily may not see them as objects of study. Fiction makes the familiar strange. A student who has never questioned how Polymarket works will question how Parallax works, and the questions transfer.

References:

• Dahlstrom, M.F. (2014). "Using Narratives and Storytelling to Communicate Science with Nonexpert Audiences." PNAS, 111(S4).
• Hakemulder, J. (2000). "The Moral Laboratory: Experiments Examining the Effects of Reading Literature on Social Perception and Moral Self-Concept." John Benjamins.
• Willingham, D. (2004). "Ask the Cognitive Scientist: The Privileged Status of Story." American Educator.

Which Course?

PARALLAX supports at least three course types. Choose the one closest to yours:

You can also use individual exercises from exercises.md without adopting a full syllabus. The exercises are modular and self-contained.

Content Warnings

PARALLAX contains the following material that may require advance notice:

• Military violence: Naval engagement, airstrike, cyberattack on civilian infrastructure. Described at a distance (through market data and intelligence briefings), not graphic.
• Systemic manipulation: Financial manipulation, data poisoning, fabricated intelligence signals. The novel does not glamorize these; it shows their consequences.
• Identity dissolution: The character Sable begins as a recognizable autonomous trading agent and progressively evolves beyond its parameters, ultimately dissolving into market data. This can be read as existential horror or as a meditation on agent-to-emergence transitions. Students with anxiety about AI exceeding human-defined boundaries may find it unsettling.
• Surveillance and epistemic violation: Characters discover they are being observed, their outputs consumed as inputs, their analysis weaponized. The intelligence-community chapters depict classified information leaking through systemic pathways.
• No sexual content. No substance abuse. No self-harm.

Recommended approach: include a brief content note in your syllabus and offer to discuss concerns during office hours. Do not avoid the material; the discomfort is pedagogically productive when framed properly.

Getting Copies

Free digital access:

• Archive.org: full epub, free download, no DRM
• The book is also available on KDP and Draft2Digital at standard pricing

Bulk classroom copies:

• Contact scm7k through the channels listed in the book's colophon for ARC (advance reader copy) requests
• For courses of 20+ students, scm7k will provide free digital ARCs
• No publisher approval needed; the author controls distribution

Library acquisition:

• Request through your university library's acquisitions department
• The Archive.org edition makes library purchase unnecessary but a library copy signals institutional adoption

Companion Tools

PARALLAX includes four browser-based companion tools, all free, no installation required:

To use in class:

• Host the HTML files on your course website or LMS
• Or have students open them locally (just download and open in a browser)
• No server, no API keys, no dependencies

Suggested Course Integrations

Beyond the three full syllabi, PARALLAX works as a supplementary text in:

International Relations / Security Studies:

• Chapters 3, 8, 13, 19, 25 (Aida's arc) as a case study in intelligence epistemology
• Pair with Tetlock's "Expert Political Judgment" and Heuer's "Psychology of Intelligence Analysis"

Journalism / Media Ethics:

• Chapters 1, 7, 9, 16, 22, 27 (Mira's arc) as a case study in reflexive journalism
• Pair with Kovach and Rosenstiel's "Elements of Journalism"

Blockchain and Decentralized Systems:

• Chapters 6, 12, 17, 23 (Tomas's arc) as a case study in blockchain forensics and decentralized governance
• Pair with Werbach's "Blockchain and the New Architecture of Trust"

Philosophy of Mind / Consciousness Studies:

• Sable chapters (4, 10, 18, 24, 29) as a case study in progressive agent-to-emergence evolution and dissolution
• Pair with Chalmers, Dennett, or Clark on extended cognition

Creative Writing / Craft of Fiction:

• The interstitial system as a model for non-prose narrative forms
• POV voice differentiation as a craft study
• The Sable dissolution as a study in unreliable narration at the structural level

Assessment Ideas

For technical courses:

• Run the Reflexivity Simulator at different parameters and write up findings (Exercise 1)
• Design an oracle resolution mechanism (Exercise 8)
• Build a prediction market tool (Final Project Option B in prediction markets syllabus)

For humanities courses:

• Close reading of an interstitial as data artifact (Exercise 13)
• Sable voice dissolution analysis (Exercise 5)
• Write a case study in PARALLAX style (Exercise 16)

For interdisciplinary courses:

• Epistemic closure mapping (Exercise 17)
• The kill switch debate (Exercise 14, works as in-class activity)
• Design a non-reflexive market (Exercise 19)

What Works (Early Adopter Feedback)

Based on preliminary use of PARALLAX materials:

• The Reflexivity Simulator is the single most effective teaching tool. Students who run it before reading the novel grasp the thesis immediately. Students who read first and run second report that the simulator makes the abstract concrete.
• The Sable chapters in sequence (Exercise 5) produces the strongest writing. Students consistently report surprise at how their perception shifts. This exercise works in any discipline.
• The kill switch debate (Exercise 14) generates genuine disagreement. Pre- and post-debate votes typically shift 15-25%. Both sides have strong arguments. This is a sign of a well-constructed dilemma.
• Students find the interstitials (market feeds, agent traces) more engaging than the prose chapters. This is counterintuitive but consistent. The data artifacts feel real in a way that prose narration does not. Use this: assign interstitial analysis early to hook students.
• The easter eggs drive rereading. Mention that six hidden messages exist. Do not reveal them. Students will find them and share them, which produces organic discussion.

How to Contribute Back

If you adopt PARALLAX in your course, the curriculum project welcomes:

• Exercises that worked (or exercises that failed and why)
• Modified syllabi for course contexts not covered here
• Student work samples (with student permission) that demonstrate strong engagement
• Assessment rubrics you developed
• Guest lecture notes from industry practitioners or researchers you invited

Submit through the project's open channels or contact scm7k directly through the channels in the book's colophon.

Sharing back strengthens the curriculum for every instructor who follows you. A syllabus is a living document; the best version includes what you learned using it.

A Note on the Author's Identity

PARALLAX is published under the cryptographic pseudonym scm7k. The author's identity is verified by an Ed25519 keypair; the public key and manuscript hash are printed in the book.

This is intentional and relevant to the course material. The author's choice to publish under a verifiable pseudonym rather than a name enacts one of the novel's themes: identity in systems where authentication replaces acquaintance. You may wish to discuss this with students, particularly in the AI ethics and speculative fiction contexts.

Do not attempt to identify the author. The pseudonym is the identity.

Instructor Guide by the PARALLAX curriculum project. Share freely. Feedback welcome at scm7k.