ACXLW #94 — Saturday, May 17 | “Cheating, Storks … and What They Teach Us About Incentives”

Hi everyone,

After spelunking through AI constitutions and super-egos, we’re pivoting to the messier side of incentives:
Why people cut corners when the game looks pointless, and when a well-timed nudge can coax entirely new humans into existence.

Our twin readings make a perfect (if slightly mischievous) pair:

1. “Cheaters Gonna Cheat Cheat Cheat Cheat Cheat” — Zvi Mowshowitz (May 9, 2025)

Text: https://open.substack.com/pub/thezvi/p/cheaters-gonna-cheat-cheat-cheat • Audio: https://youtu.be/OepfOvEFqpY?si=0xdWcM6nSUZcKMBB

Deep-dive recap

Zvi surveys the “ChatGPT plagiarism panic” on college campuses and argues the real story isn’t ethics; it’s economics.

• Students don’t see value in rote essays designed to police effort, so the moment a frictionless cheat appears, they flock to it.
• Professors, administrators, and even ed-tech firms mostly shrug, tacitly admitting the assignments were signaling games all along.
• Attempts to detect or punish AI help are a losing whack-a-mole. Instead universities must either (a) test in person/orally; (b) redesign coursework so AI becomes a partner in genuine learning.
• Analogy: Scott Alexander’s “Whispering Earring.” Once a tool can optimize every micro-move, the temptation to surrender agency is overwhelming, so society must decide where deliberate effort still matters.

Starter questions

1. If an assignment’s only purpose is to grade conscientiousness, is using AI actually “cheating”—or merely proof the test was poor?
2. Zvi argues that higher ed’s main output has become signaling, not skill. Do you agree? If so, what should replace the college credential?
3. Could universities embrace AI the way calculators transformed math classes? What would a “pro-AI” syllabus look like?
4. “Cheaters escalate until caught.” Are there domains (politics? research?) where AI-assisted corner-cutting could be existential?
5. How do you personally decide when using ChatGPT is a legitimate productivity boost versus a learning dodge?
6. Is cheating actually developing more technology-relevant skills for the real world?

2. “Storks Take Orders From the State” — Cremieux (May 1 2025)

Substack link

Deep-dive recap

Cremieux dismantles the popular claim that “pronatal benefits never work.”

• Misread counterfactuals. Yes, French fertility kept falling after 2013 benefits were trimmed—but it fell twice as fast as the no-cut scenario. You must compare with-vs-without.
• Magnitude blindness. France’s €180/mo child allowance lifts TFR only ~0.15—but that’s 5–10 million extra citizens over a generation.
• Global evidence. Germany’s 2006 welfare cut cut births 6.8 %. Quebec’s baby bonus raised them 12–25 %. Norway, Spain, Israel, Korea, Argentina all show large elasticities (0.4–4.0) of births to cash.
• Cost-benefit. Typical American nets the Treasury >$200 k in lifetime taxes; generous benefits cost <$50 k → 3× fiscal ROI. Spending ~3 % GDP could push U.S. fertility above replacement and pay for itself.
• Growth matters. Income booms—Appalachian coal, lottery wins—increase births; busts shrink them. So pro-growth reform is itself pronatal.
• Social multipliers. Bigger families shift norms; minor policy-driven bumps can snowball culturally.
• Design trumps hype. Hungary’s flashy but narrow incentives flop; Mongolia’s simple universal stipend worked. Key levers: cash-like, broad, easy, predictable.

Starter questions

1. Which elasticity estimate (0.4? 3.9?) feels most plausible to you—and why do studies diverge so widely?
2. If child benefits pay for themselves fiscally, why aren’t treasuries racing to adopt them? Political optics? Time-horizon bias? Moral objections?
3. Could pro-growth housing or energy policy raise fertility more cheaply than direct cash? Where are the synergy opportunities?
4. Should benefits be universal, means-tested, refundable tax credits, or targeted to IVF? What equity or “dysgenic” concerns matter?
5. Social contagion: Have you personally felt family-size norms shift based on peers or media? How might governments harness that?
6. Immigration vs. pronatalism: With global TFR falling, can rich countries still rely on inflows, or must they grow their own future citizens?

Logistics & ritual

• When? Saturday · May 17 · 2 pm (hangout), 2 : 30 pm sharp discussion
• Where? 1970 Port Laurent Place, NB CA 92660‑7117

Whether you’re horrified by AI essays, thrilled by baby bonuses, or just wondering how to design incentives that work, come share the debate.

See you Saturday!

— Michael