Prep handout — Synthesis Discussion

What is SDS for?

Pick the principle from Joi's nine 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?

• Joi's nine principles: Resilience over strength · Pull over push · Risk over safety · Systems over objects · Compasses over maps · Practice over theory · Disobedience over compliance · Crowds over experts · Learning over education.
• The Safecast story (post-Fukushima citizen radiation monitoring) — what made it work was that volunteers, sensors, and a missing institution showed up at once.
• The SDS-vs-traditional-institutions framing: diversity and heterodoxy as design, not accident.
• Re-skim: Week 1 lecture — Why are we here?

Bring one principle and one concrete example — your own work, your own training, or something you have read.

Systems, leverage, and your own theory of change

Pick a problem you actually care about. Where on Donella Meadows' leverage-points hierarchy is most of the public conversation about it happening — and where do you think the leverage actually is?

• Meadows' hierarchy: tweaking parameters (lowest leverage) → information flows → rules → goals → paradigms (highest leverage).
• The antidisciplinary diagrams: disciplinary (parallel), multidisciplinary (side-by-side), antidisciplinary (deeply interwoven).
• Joi's Resisting Reduction manifesto — the answer to a complex future is not a smarter model but a richer relationship between models, people, and feedback loops.
• Re-skim: Week 2 lecture — Systems Thinking and Antidisciplinarity

5–10 minutes. The point of this one is to make your own theory of change explicit, not to be right.

Choosing lenses honestly

Take one method or framework you actually use in your own work. Name three things it reveals and three things it obscures. Was the second list harder to write?

• Week 3: minimal information is enough for the brain to fill in gender / race / age / weapon (Wapman & Belle 2014; Banaji 2022 replication; the train-mugging case; Southport vs. Ulsan).
• Week 4, Lens 1 (EpiSim): trajectory, timing, intervention placement, individual heterogeneity — at the cost of universality and clean R₀.
• Week 4, Lens 2 (percolation): thresholds, universality ("same math everywhere") — at the cost of which node, which time, which intervention.
• Re-skim: Week 3 lecture — Prejudice & Bias · Week 4 Lens 1 — Complex Systems Lenses (Ira) · Week 4 Lens 2 — Percolation (Hiroki)

The reveal/obscure exercise is the most important homework. Do it slowly.

Ethics under uncertainty

The systems you are about to 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 turn that default into a choice.

• Utilitarianism (Bentham): the right action is the one that produces the most aggregate good. Easy to compute; punishes minorities when convenient.
• Kant: the right action is the one whose rule you could will to be universal. Robust to individual cases; can be paralyzed by conflicts of duty.
• Week 5's Part II frames modern AI ethics as largely utilitarian by default — recommender systems, content moderation, autonomous-vehicle trolley framings. The question is whether that default is a choice or an inheritance.
• Week 5 lecture page: not yet linked — Mizuki Oka & Catharina Maracke's deck will appear under docs/lectures/ once published. The Week 5 class is on the Friday before this one.

5–10 minutes. If you want, jot one example to the side you can pull out in class.

One principle

Be ready to name one principle, sentence, or question you want to carry forward from the course. It can be from any week, from this discussion, or one you came in with and want to defend. We will go around the room once, no commentary, only listening.

If you want a thought-partner

If you would like to talk through any of these prompts with an LLM (Claude, ChatGPT, your local model — whatever you use), here is a prompt that sets it up to be useful rather than answer-shaped. Paste it, replace the bracketed text, and go. It also tells the model to respond in Japanese if you write to it in Japanese, so you can hold the entire conversation in 日本語 if that's easier.

You are a thought-partner, not an answer-engine.

I am preparing for a graduate-class discussion in the DNA course
("The Antidisciplinary Approach to the Modern World"), taught at the
Henkaku Center for Radical Transformation at Chiba Institute of
Technology. The discussion covers five prior lectures:

  1. Why are we here? (Joi Ito — SDS story, nine principles)
  2. Systems Thinking and Antidisciplinarity (Joi Ito — Meadows
     leverage points, Resisting Reduction)
  3. Prejudice and Bias (Daum Kim with Joe Austerweil — how the brain
     fills in defaults from minimal information)
  4. Ways of Knowing Complex Systems (Ira Winder with Hiroki Kojima —
     EpiSim agent-based modeling and percolation theory as two lenses
     on epidemic spread)
  5. Utilitarianism, Kant, and AI Ethics (Mizuki and Catharina)

I want help thinking through ONE prompt:

  [PASTE THE PROMPT YOU PICKED HERE]

Rules:
  - Do not write my answer for me. Help me find mine.
  - Ask me sharpening questions when my thinking is vague.
  - Push back when I am being lazy or generic; do not flatter.
  - When I name an example, ask me what it would mean if the
    opposite were true.
  - If I write to you in Japanese, respond in Japanese
    (日本語で書いたら日本語で返してください).

My first attempt at the prompt is below. Engage with it.

  [WRITE YOUR DRAFT ANSWER HERE — TWO OR THREE SENTENCES IS PLENTY]

Click inside the box, ⌘/Ctrl-A to select, then copy. The whole thing is meant to be one paste.