Appearance
Session 7: Facilitator Notes
Duration: 60 min · Deliverable: three governance docs (drafted) + one working enforcement mechanism
Before the session
- I've reviewed the four enforcement options myself (Husky, ESLint rule, coverage gate, PR template) and can debug each.
- The starter templates in
starter/are in shape; review them once. We fill these in together, live; they're not take-home. - I have a real-world governance doc to reference (my own team's, or one of the publicly available ones from Stripe, GitLab, etc.).
- Everyone has their course logbook (
/logbook) open. The governance docs and the enforcement we ship today get recorded there live as we go.
Time budget
| Block | Mins | Slides |
|---|---|---|
| 1. Tech debt + cognitive offloading | 10 | 1–7 |
| 2. Real-world governance examples | 10 | 8–10 |
| 3. Draft three governance docs | 15 | 11 |
| 4. Implement one enforcement | 15 | 12–15 |
| 5. Debate: freedom vs standards | 10 | 16–22 |
Block-by-block
All core work today is done live, together, in session. I demonstrate the enforcement mechanism (see live-demo.md), and we draft the governance docs as a group. Students record the docs and the shipped enforcement in their logbook (/logbook) as we go. Nothing is assigned as take-home; a second enforcement mechanism is optional polish only.
Block 1. Cognitive offloading (slide 6) is the new concept; most students haven't thought about it. The line "AI multiplies your output. It can also multiply your debt." is the anchor. Personal story helps: a time I used AI and shipped something I later regretted.
Block 2. Show one real governance doc on screen, mine or a public one. Read it together. Note how short it is. Most students expect 30 pages; show them it's 2.
Block 3 (drafting, 15 min, live together). Three docs, but each is one page. We open the templates in starter/ and fill in the bracketed sections together, live. They'll feel like they're "not writing enough." That's the point. Cue: record the three docs in the logbook as we go.
Block 4 (enforcement, 15 min, live together). I demo one mechanism live; each student ships ONE alongside me. Don't let anyone do two; depth over breadth here. A second mechanism is optional polish, never required. Walk the room; common issues:
- Husky setup confusion (npm scripts vs husky install).
- ESLint flat config (eslint.config.js) syntax for custom rules.
- GitHub Actions branch protection rule must be set on GitHub UI, not just in YAML.
Block 5 (debate, 10 min). Split the room. Side A argues against the enforcement they just built. Side B argues for it. Then swap. The goal is empathy with the opposing view; every senior engineer needs both perspectives.
Anticipated questions
"Does this stuff actually slow teams down?"
Yes, a little. The trade is consistency for time. Worth it past ~5 engineers.
"What if a team member objects to the policy?"
Talk to them privately first. If you can't change their mind, escalate to the EM. Don't enforce in PR reviews unilaterally.
"How do I know when to remove a standard?"
When you can't remember why it was added. (Document the why, always.)
"Should the AI usage policy ban AI for security-related code?"
Reasonable to require human review on security-touching files. "Ban" is strong; "requires sign-off" is better.
If you run short
- Cut block 5 debate to 5 min: quick poll instead.
- Compress block 2 to 5 min: show one example, move on.
- Never cut block 4. The working enforcement is the deliverable.
Post-session
In chat: "Everything we built today is in your logbook (/logbook): the three governance docs, the enforcement mechanism you shipped, and a pointer to your improved repo. Keep your logbook; in Workshop 8 your proposal comes straight out of it (it already holds your W1–W7 outputs). Optional: practice your 8-minute story out loud before next session."