# Lecture Plan — *Synthesis Discussion* DNA course · Week 6 · Friday, June 5, 2026 · ~120 min · Joe Austerweil. This is the final session of DNA. It replaces the original "present-back to the faculty" assignment with a guided discussion across the semester's five prior lectures. Students prepare four reflective prompts at home (see [`homework.html`](./homework.html)) but **submit nothing** — the homework exists so the conversation can be theirs. ## Framing The semester arc, in one sentence: *antidisciplinary practice is the discipline of choosing your instruments knowingly, naming what each one obscures, and acting under uncertainty without pretending the uncertainty is gone.* The four discussion topics walk that sentence: 1. **What is SDS for?** — antidisciplinary as a stance, not a label. 2. **Systems, leverage, and your own theory of change** — once you accept that the world is a system, you have to say which part of it you intend to push on. 3. **Choosing lenses honestly** — the moral content of method choice; the bias of choosing only one. 4. **Ethics under uncertainty** — when the system you build inherits an ethics whether you wrote one or not. The close brings the room back together by asking each student to name one principle they will carry forward. ## Learning goals 1. Articulate, in their own words, what SDS's "antidisciplinary" stance means — and where they personally agree and disagree with it. 2. Practice naming *what each lens or framework obscures*, not just what it reveals — extending the Week-3 (bias) and Week-4 (two-lenses) move to their own working methods. 3. Notice the ethical commitments their work already encodes, and describe at least one place where they would prefer to make that commitment a choice instead of an inheritance. 4. Hear the cohort say, in their own voices, what each is taking away — so the course ends as a group, not as a lecture. ## Class shape | § | Topic | Lecturer | Time | |---|---|---|---| | 0 | **Open.** Ground rules (conversation, not assessment; equal voice; bring an example). 5-minute recap of the course arc — one sentence per prior lecture, ending with the antidisciplinary one-liner above. | Joe | 5 min | | 1 | **What is SDS for?** Antidisciplinary as a stance, not a label. Anchor: Joi's nine principles + Safecast. | Joe + room | 25 min | | 2 | **Systems, leverage, and your own theory of change.** Anchor: Meadows hierarchy + antidisciplinary diagrams + Resisting Reduction. | Joe + room | 25 min | | 3 | **Choosing lenses honestly.** Anchor: Week-3 bias cases + Week-4 reveal/obscure framing. The hinge of the course. | Joe + room | 25 min | | 4 | **Ethics under uncertainty.** Anchor: Week-5 utilitarianism / Kant + the "default ethic in your field" prompt. | Joe + room | 25 min | | 5 | **Close — one principle each.** Around the room, one sentence each, no commentary; Joe closes with ≤2 min synthesis. | Room → Joe | 15 min | **Pacing discipline.** Topic 1 is the one most likely to overrun (students will want to argue Joi's principles). Plan to interrupt at the 25-minute mark, naming explicitly that good arguments are being left on the floor and that the close is non-negotiable. ## Opening recap script (≤5 min) A one-sentence reminder of each prior week, on the way in: - **Wk 1 — Why are we here?** SDS is built on resistance to naive compliance, learning by doing, and the comfort of working *between* disciplines. - **Wk 2 — Systems Thinking and Antidisciplinarity.** Complex systems don't yield to one instrument; the highest-leverage interventions usually change the paradigm, not the parameters. - **Wk 3 — Prejudice and Bias.** The brain happily fills in detail the world didn't give us — and what it fills in becomes the basis for action and policy. - **Wk 4 — Ways of Knowing Complex Systems.** Two methods on one phenomenon: EpiSim shows the trajectory, percolation shows the threshold. Each is honest only when you name what it obscures. - **Wk 5 — Utilitarianism, Kant, and AI Ethics.** When we build, we inherit an ethics by default — and the question is whether we notice in time to choose differently. End the recap with: *"Today we put these together. Not because the course tied them up — because you are about to leave and act on them."* ## Discussion questions — full menu per topic The homework gives students **one** prompt per topic (the headline). In the room, Joe can reach for any of the alternates below if a topic stalls or needs a redirect. ### Topic 1 — What is SDS for? (Wk 1, Joi Ito) **Anchor material.** Joi's nine principles. The Safecast story. The diversity-vs-standardization tension. The SDS vs. traditional institutions framing. 1. *(homework headline)* Pick the principle that felt most *wrong* to you the first time you read it. Why did it sting, and has anything in the last five weeks moved you on it? 2. *Disobedience over compliance.* Where in your own training so far has compliance been rewarded in a way Joi would call "naive"? What was the cost? 3. Joi's Safecast story works because volunteers, sensors, and an institution that didn't exist yet showed up at the same time. What's the modern problem you would Safecast — and what would the sensor be? 4. SDS frames itself in opposition to traditional credentialing. If that's true, what is your *degree* from SDS evidence of? Say what it should mean to a future employer, collaborator, or government. ### Topic 2 — Systems, leverage, and your own theory of change (Wk 2) **Anchor material.** Donella Meadows' leverage-points hierarchy (paradigms > rules > information flows > parameters). The disciplinary / multidisciplinary / antidisciplinary diagrams. The AI Governance multi-lens example. Theory-of-Change diagram. 1. *(homework headline)* Pick a problem you care about. Where on Meadows' ladder is most of the public conversation about it happening? Where do *you* think the leverage actually is? 2. *Antidisciplinary* is easy to say and hard to do. Name a problem you've worked on where you noticed your own discipline pulling you toward its instruments. What did you reach for? What might a different lens have caught? 3. Joi's *Resisting Reduction* manifesto says the answer to a complex future isn't a smarter model — it's a richer relationship between models, people, and feedback loops. Where in your own field is "the smarter model" being sold as the answer? ### Topic 3 — Choosing lenses honestly (Wk 3 + Wk 4) **Anchor material.** Week 3: Wapman & Belle 2014, Banaji 2022; the train-mugging case; Southport vs. Ulsan. Week 4: each lens is honest only when you name what it obscures. 1. *(homework headline)* Take a method or framework you actually use. Name three things it reveals and three things it obscures. Was the second list harder? 2. The Week-3 cases show how confidently the brain infers detail from very little signal. In your own research or design work, what's a "lens" you've been using that quietly fills in defaults? What defaults are they? 3. When (if ever) is it ethical to pick a single lens and commit to it? Is "I know it's a partial view but I'm acting on it anyway" antidisciplinary, or just disciplinary in self-aware clothing? ### Topic 4 — Ethics under uncertainty (Wk 5) **Anchor material.** Wk 5's Part I (Bentham / utilitarianism / consequences / Kant) and Part II (utilitarianism in modern AI ethics). The previous topic's "name what your lens obscures" habit is the bridge into this one. 1. *(homework headline)* The systems you build will have an ethics whether or not you wrote one. Name a default ethic that gets baked into systems in your field by accident, and say what it would take to make it a choice. 2. Utilitarianism asks *what produces the most good.* Kant asks *what could be willed as a universal rule.* Pick a real AI deployment decision you've read about in the last year — does it look better through one lens than the other? Where do they disagree? 3. In Wk 2, the highest leverage point was *changing the paradigm.* In Wk 5, utilitarianism IS a paradigm. What happens when you treat it as a *leverage point* (a way of intervening on a system) instead of as the truth? 4. The course defines antidisciplinary as choosing instruments knowingly. Apply that to ethical frameworks themselves: when would you reach for utilitarianism, when for Kant, when for something else? What's the question that makes you choose? ### Close — One principle (15 min) Around the room, once. Each student names one principle, sentence, or question to carry forward. No commentary. Joe closes with ≤2 min of synthesis — explicitly tying the most-named principle (or the most-different ones, if they cluster) back to the antidisciplinary one-liner from the opening recap. ## Facilitation notes - **Equal-voice rule.** Open by saying you'll be tracking who has spoken; if the same three voices dominate, you will name names to bring others in. - **Examples beat essays.** When a student answers in abstractions, ask for one example. When they answer with an example, ask what the opposite case would look like. - **Pushback on flatness.** "Both have merit" is the most common bad answer to Topic 4 prompts. When you hear it, ask: *which one would you bet your work on, if you had to pick before lunch?* - **The homework's claude-cli affordance.** Some students will arrive having argued with an LLM. Treat that as legitimate prep — but if their answer sounds polished, ask the same example question and watch what they do without the model. ## Verification - Read the four homework prompts out loud. Each should be answerable in 5 minutes with no research. If a prompt requires a citation, move that citation into the anchor bullets, not the prompt itself. - Mentally rehearse Topic 1 with two follow-up questions. If it spills past 30 min, drop one alternate question from the menu *before* class — not during. - Local preview after edits: `python3 -m http.server 8000` from repo root, then open `http://localhost:8000/docs/lectures/synthesis-discussion/`.