The Policy in the Room

POSC 459 — Social Welfare Politics and Policy
Week 2 · Wednesday, September 2

Howard ch. 2, "Tracks of My Tiers" · Desmond ch. 2

Today

  1. Research pre-survey (~10 min)
  2. PapyrusAI sign-up, everyone, in the room (~10 min)
  3. My AI policy — as a policy (~35 min)
  4. Why haven't we made more progress? (the rest)

First: the pre-survey

uci.questionpro.com/t/Acs2vZ9boH
  • It is part of a research study, and it is voluntary.
  • No credit either way. Declining costs you nothing with me.
  • Absent today, or want to do it later? Complete it on your own by Friday 9/11.

Sign-up, together, now

  1. Canvas course nav → PapyrusAI. First access creates your account.
  2. Choose Google or a password — decide now, not in October.
  3. Course code: FALL2026-POSC459
  4. Confirm you're enrolled before you close the tab.

One entry, one new conversation.

The + button. Top right. Every time.

You are carrying five AI policies right now

Five courses. Five different rules. Zero tolerance for mixing them up.

That is administrative burden, and I am contributing to it.

By Week 3 you'll have the vocabulary for exactly this.

§12, in your hands

Read my AI policy the way this course reads any policy:

  • Where is it unclear?
  • Where does it impose burden it doesn't justify?
  • Where would a reasonable student get it wrong?

My stake, on the record

  • I use these tools.
  • I'm in a grant-funded faculty learning community promoting AI-integrated teaching.
  • This course runs on PapyrusAI.

Asking you to evaluate something I benefit from without saying so would be the exact failure the disclosure rule exists to prevent.

Scale it up

The CSU–OpenAI contract.

  • 2025: systemwide ChatGPT Edu deal, ~$17M for eighteen months.
  • Renewed 2026: ~$13M a year, for three years.
  • Sold as equitable access. Contested in public — EdSource, LAist, the AAUP.

You have an account because of this decision.

The exercise — three groups, eight minutes

  1. Does "equitable access" describe what the contract accomplishes? What would have to be true for the claim to hold?
  2. Who bears the cost — and is it the group that gets the benefit?
  3. The beneficiaries were not consulted. Compare: how are TANF or SNAP beneficiaries consulted — or not?

Report a position and the strongest argument against it. A group that summarizes both sides goes back for one minute.

Before we leave it — a poll

Does receiving something make you more likely to defend the decision that provided it?

Monday I left you a question

If poverty is a knot and not a line — what number would tell us we're making progress?

Here is the official number.

The flatline

  • Official poverty rate, 1970: 12.6%
  • Official poverty rate, 2019: 10.5%

Fifty years. And over that stretch, spending on the thirteen largest means-tested programs went from $1,015 to $3,419 per person — up 237%.

Desmond's chapter 2 walks the suspect list: stinginess, immigration, single parenthood. Each one fails the evidence.

But which number was that?

The official measure never sees SNAP, the EITC, housing help, or a stimulus check. Count those — the Supplemental Poverty Measure does — and poverty falls meaningfully over the same years.

Fantasy, in the film: "I have received extra food stamps cause of COVID. That stimulus, that paid half my bills."

The official measure records her as untouched.

Where the dollars leak

  • 22 cents of every budgeted TANF dollar reached families as cash in 2020. About $6B sat unspent.
  • Becky, in the film — approved for pandemic unemployment: "Everything is still pending, though, so I haven't gotten anything yet."
  • Crystal — "I am still required to work my hours for my assistance … maybe my assistance could get shut down."

And Howard's map: Crystal's household runs on food stamps, HUD, disability-linked cash, and free school lunch at once — four programs, both "tiers," one trailer.

Due and next

  • Discussion Prompt 1 — the CSU AI contract: due Tuesday 9/8.
  • Pre-survey, if you missed it today: by Friday 9/11.
  • Discussion Paper 1 (both parts of Growing Up Poor in America + Howard ch. 3 + Pierson): Friday 9/18. Keep film notes now — a person, a scene, a moment.
  • Labor Day Monday 9/7 — no class. Week 3 is Wednesday 9/9 only: Howard ch. 1, Desmond chs. 3–4.

Heavy procedural day and the room should hear that named: three pieces of housekeeping before content, all three date-bound. The content tail flexes; the survey and sign-up do not.

Say the two things plainly: part of a study, and voluntary. Copy the URL to the projector and the same-morning announcement — do NOT retype it; the post-survey URL differs only in the case of the final letter.

Student Guide on the screen for the setup steps. Watch for the quiet failure: an account created but the course code never entered — that student sees no error and lands nowhere. Ask for thumbs up on the enrollment screen, not on "did it work."

Say it out loud; a sentence on the assignment sheet will not prevent it. Reopening the module resumes the last conversation, and ten journal entries stacked in one thread count as one.

Name the whiplash before they do. Saying it is more honest than presenting paragraph 12 as self-evidently correct.

Printed copies out. I would rather hear this in week two than read it in evaluations — say that. Have the structure/"outlining arguments" contradiction ready: I wrote it, a tool caught it, I adopted the tool's fix in the tool's own words — which under my own rule is disclosable use. It is mine, and it is the honest version of the problem I'm asking them to navigate.

Sources on the screen during group work. Disclose your stake before they start, not after. Hold the line on the no-fence-sitting rule — it is the discussion-paper rule too, so this is practice.

Count hands and WRITE THE NUMBER DOWN — Week 15 asks this again after a semester of policy feedback, and the callback only works if today's number exists. Then assign Discussion Prompt 1, the CSU AI contract, due Tuesday September 8.

The elimination march, fast: Reagan couldn't dismantle it; CA/TX/FL absorbed half the foreign-born and poverty fell; single mothers aren't poorer in universal-program democracies, and New Hope raised marriage rates by raising incomes. Save the leaky bucket for two slides from now.

This is the payoff of Monday's parked question: the answer depends on what the yardstick counts, and the yardstick is a 1965 food budget times three. Wimer et al. on the SPM decline; Burtless raised exactly this against Desmond's flatline rhetoric. Both slides are true — that is the point.

Howard's numbers if the room wants them: TANF max for a family of three, 2002 — $201 in Texas, $636 in Connecticut. Same family, same program, same year. The two-tier map can't survive the middle programs; the film's households never heard of the tiers. Say "Crystal in the film" — not Desmond\'s Crystal Mayberry from Monday; the name collision is real.

Part 2 of the film arrives in Week 4. The transcript for Part 1 is on the PBS page for anyone who needs text access.