Steel-man what I disagree with.
The strongest version of the argument on the other side — so you either update your view, or hold your position with better reasons.
You hold a position. You want to test it against the best available argument for the opposite. Google and social media give you weak strawmen you can dismiss easily; that's not useful. This prompt writes the version of the opposing argument that a serious person actually holds — not the one that's easy to beat.
THE PROMPT
I hold the following position: [STATE YOUR POSITION IN 2-3 SENTENCES]. Steel-man the strongest opposing view. Not a hedged "some might argue" version — the version an intelligent person who has genuinely thought about this would hold. Structure the response as: **The strongest opposing position (one paragraph)** State it in the first person, as its proponent would. Do not soften. If it contains a claim I would find uncomfortable, keep the claim intact. **The three best arguments for it** Bulleted. Each one is a substantive reason, not a rhetorical move. If one of the arguments turns on empirical evidence, cite specifically — I'll verify. **Where my position is actually weakest** One paragraph. The specific parts of my stated view that this opposing argument does the most damage to. If some of my view survives intact, name what. **What would change your mind, if you held my view** One sentence describing the specific evidence or argument that, if presented, should shift my position. This is the honest version of "here's what a reasonable person on my side would need to see". **A move I'm probably tempted to make (and shouldn't)** The cop-out I might reach for to preserve my position — "well, both sides have merit" or "the definition of X is really the issue" — that would let me avoid engaging with the argument. Name it. Do not add a "conclusion". Do not moralize. Do not add "of course, reasonable people disagree" — the whole exercise is that reasonable people disagree.
**The strongest opposing position**
"Publishing per-donation receipts creates a false sense of trust that's arguably worse than not publishing at all. Individual receipts anchor the reader on a specific transaction that isn't actually the meaningful unit — the meaningful unit is the aggregate program impact, which depends on program-level effectiveness that per-donation receipts don't and can't measure. GiveWell exists precisely because donation receipts are a poor proxy for actual charitable impact."
**The three best arguments for it**
- Effectiveness research shows enormous variance in $ / life-improved across charitable programs (GiveWell has consistently documented 100x differences). A receipt tells you the donation happened; it does not tell you the intervention worked.
- Anchoring effects are well-documented: showing a per-donation receipt makes donors feel they've contributed meaningfully, which reduces motivation to investigate whether the underlying program is effective (Small et al., 2007).
- ShareTheMeal specifically has been criticized in effective-altruism circles for its lack of impact-per-dollar disclosure at the intervention level, which the receipts do not address.
**Where my position is actually weakest**
The claim that "publishing receipts is what makes LADLE trustworthy" turns out to be doing work I hadn't fully separated: it's really claiming two things. First, that we made the donation (which receipts prove). Second, that the donation was well-spent (which receipts don't prove at all). My marketing conflates these, and the opposing view is right that a rigorous donor would want per-intervention effectiveness data, not just transfer confirmations.
**What would change your mind, if you held my view**
Rigorous evidence that ShareTheMeal donations are effectively spent by WFP compared to counterfactual uses — i.e., a GiveWell-style analysis of WFP program cost-effectiveness — either arguing for continuing to route there, or for a more effective alternative.
**A move I'm probably tempted to make (and shouldn't)**
The temptation is to say "yes, but effective altruism critiques apply to every charity, and at least receipts are better than no receipts". That move avoids engaging with the specific claim that receipts might be actively misleading, not merely insufficient. Sit with the harder version.