Apply Poison Cure Diagnostic

---
name: apply-poison-cure-diagnostic
description: Use when an urgent problem has a fast "fix" available — before adopting it, check whether the fix's relief mechanism is the same channel that will make the underlying problem worse and require a larger dose of the same fix next time, versus an ordinary cost/benefit tradeoff where the cost is separate from the relief.
source: '后汉书·霍谞传 (Book of Later Han, Biography of Huo Xu, compiled by Fan Ye, ~5th century AD, describing events ~1st century AD) — 譬犹止渴於鴆毒 (like quenching thirst with poisoned wine — 饮鸩止渴); Cagan, "The Monetary Dynamics of Hyperinflation" (1956); Olivera-Tanzi effect; CFPB, "Payday Loans and Deposit Advance Products" (2013); WHO, "Global Action Plan on Antimicrobial Resistance" (2015); WADA Anti-Doping Code'
tags: [decision-making, risk-management, debt-trap, escalation, tradeoff-analysis, strategy]
related: [apply-goodharts-law, apply-early-intervention, audit-systemic-credit-bubble-risk, apply-sunk-cost-discipline, apply-earnings-myopia-defense]
---

# Apply Poison-Cure Diagnostic

Before adopting a fast fix for an urgent problem, check whether the fix's relief mechanism is structurally the same channel that worsens the underlying problem and escalates the dose needed next time — reject it regardless of how small the first dose looks, because no dose of a self-reinforcing trap is safe.

## Why This Is Best Practice

后汉书·霍谞传 (Book of Later Han, ~5th century AD, describing events ~1st century AD):

> 譬犹止渴於鴆毒，非不暂止，而肠胃已溃矣。

"Like quenching thirst with poisoned wine — it does stop the thirst for a moment, but the gut is already ruined." The idiom 饮鸩止渴 ("drink poison to quench thirst") names a specific, narrow failure pattern: not "any solution has costs," but a solution whose relief mechanism and its future harm are the *same physiological or causal channel* — so using it doesn't trade a cost against a benefit, it compounds the very problem it appears to solve.

**Why best:** An ordinary tradeoff (cut a budget now, pay for it in reduced capability later) has a relief mechanism and a cost mechanism that are separable — you can evaluate them independently and the cost is bounded and known. A poison-cure trap has no such separation: the relief itself is what enlarges the deficit, so each use both solves today's instance and manufactures a bigger instance of the same problem tomorrow, at an accelerating rate. This is a distinct, falsifiable, checkable pattern across finance, medicine, and organizational decisions — not a generic warning against short-termism.

**Cagan (1956) and the Olivera-Tanzi effect:** Monetizing a fiscal deficit by printing currency relieves the immediate cash shortfall, but the resulting inflation erodes the real value of future tax collection (taxes are assessed and collected with a lag, so inflation shrinks their real worth by the time they're paid) — which widens the real deficit that has to be monetized next cycle, at a compounding rate. This precise mechanism is documented in Weimar Germany (1922–23), Zimbabwe (2007–09), and Venezuela (2016–19, IMF and central-bank data): each round of financing the shortfall through the money supply required a larger nominal round the next time, not because policy makers were undisciplined but because the relief mechanism and the harm mechanism were the same lever.

**CFPB, "Payday Loans and Deposit Advance Products" (2013):** Regulatory study finding that the majority of payday-loan volume comes from repeat borrowing and rollovers, not one-off use — the loan relieves this cycle's cash shortfall by creating next cycle's larger shortfall (principal plus fees due against the same or lower income), the defining signature of a poison-cure trap as distinct from ordinary short-term borrowing with a bounded repayment plan.

**WHO Global Action Plan on Antimicrobial Resistance (2015):** Documents that using antibiotics to resolve symptoms faster than clinically indicated (or without completing a full necessary course) accelerates bacterial resistance, so the next equivalent infection requires a stronger antibiotic to achieve the same relief — the treatment mechanism and the resistance mechanism are the same exposure event.

**WADA Anti-Doping Code:** Performance-enhancing substances relieve the immediate competitive gap, but detection risk and physiological dependence compound with each use, and the substances that produce the fastest short-term gain are consistently the ones with the steepest long-term health and career cost — documented across banned-substance case histories used in athlete education globally.

**Adopted by:** IMF and central-bank post-mortems on Weimar Germany, Zimbabwe, and Venezuela cite deficit monetization via the same compounding mechanism; the Consumer Financial Protection Bureau regulates payday lending specifically around the repeat-borrowing/rollover pattern; WHO's Global Action Plan on Antimicrobial Resistance is adopted by health ministries worldwide to restrict antibiotic use on exactly this escalating-dose logic; WADA's Anti-Doping Code is enforced across all Olympic and major professional sports federations.

**Impact:** Cagan's monetary-dynamics model, validated against Weimar Germany, Zimbabwe, and Venezuela, shows deficit-monetization cycles requiring geometrically larger nominal issuance per cycle once started; CFPB data shows the majority of payday-loan fee revenue comes from repeat/rollover borrowers rather than one-time borrowers, confirming the escalating-dose signature; WHO estimates antimicrobial resistance driven by this mechanism contributes to roughly 700,000 deaths annually worldwide, a figure projected to rise sharply without intervention — the clearest large-scale evidence that the shared-channel trap, once entered, compounds measurably rather than staying flat.

## Steps

1. **Name the specific relief mechanism.** State exactly what action removes the immediate pain (borrow cash, print currency, take the drug, cut the corner, give the concession) — not the goal ("solve the cash crunch") but the literal mechanism ("obtain currency by issuing new debt/currency against future revenue").

2. **Ask whether the relief mechanism and the harm mechanism are the same channel.** Test: if you use this fix once, does the thing that produced today's relief directly enlarge or accelerate tomorrow's instance of the identical problem — through the same channel, not a separate cost? If the channel is genuinely separate (e.g., spending down a cash reserve doesn't make the next reserve harder to build), this is an ordinary tradeoff, not a poison-cure trap — stop here, this diagnostic doesn't apply.

3. **Run the escalating-dose test.** Project two or three repeat cycles: does resolving each instance require a larger dose of the same fix than the last, to produce the same relief (more debt, more currency issuance, a stronger drug, a bigger concession)? An escalating required dose across repeat cycles is the confirming signature; a flat or shrinking required dose is not a trap.

4. **If both tests confirm a trap, reject the fix — even a "small," "one-time," or "just this once" dose.** The first dose is what creates the dependency that produces the second, larger instance; there is no dose size that avoids starting the cycle. Look for an alternative that solves the immediate problem through a different, non-compounding channel, even if it is slower or costs more upfront (e.g., renegotiating terms, cutting real spending, seeking equity rather than debt, treating the actual infection source rather than just the symptom).

5. **If no non-compounding alternative exists and the fix must be used anyway (true emergency), cap it explicitly before adopting it.** Set, in writing, before the first dose: the maximum total exposure, the maximum number of cycles, and the specific exit condition that ends reliance on the fix — because a poison-cure fix adopted without a pre-set exit is the documented path (Weimar, payday-loan rollover data, antibiotic resistance) to the compounding trajectory this diagnostic exists to catch.

## Rules

- A poison-cure trap is defined by the relief mechanism and the harm mechanism sharing one causal channel — not by the mere presence of a downside. Confirm the shared-channel test before applying this diagnostic; otherwise this is ordinary tradeoff analysis (see `apply-early-intervention` for cost-of-delay tradeoffs where the channels are separate).
- Never adopt the fix based on the size of the first dose. The trap's mechanism is that the first dose looks negligible; the escalating-dose test, not dose size, is what determines whether to reject it.
- If the fix must be used under genuine emergency constraints, the cap and exit condition must be set before the first use, not after — a cap set retroactively has already absorbed the first round of escalation.
- Distinguish from `apply-goodharts-law`: Goodhart's mechanism is a proxy metric diverging from the truth once it becomes an optimization target. This diagnostic's mechanism is a chosen remedy whose relief channel directly compounds the problem it treats — a decision-adoption trap, not a measurement-gaming trap. The two can co-occur (a gamed metric can itself become the poison-cure fix) but are separate failure modes.

## Examples

**Finance:** A government facing a cash shortfall considers monetizing the deficit (having the central bank purchase government debt directly). Relief mechanism: new currency covers the shortfall today. Harm mechanism: inflation erodes real future tax revenue via the Olivera-Tanzi lag effect, widening next year's real deficit. Escalating-dose test: confirmed — each cycle requires a larger nominal issuance. Diagnostic rejects direct monetization; alternative pursued: real spending cuts and external financing with a fixed repayment schedule, which do not share the same compounding channel.

**Consumer finance:** A household short on rent takes a payday loan. Relief: cash today. Harm: the loan plus fees comes due against the same or lower income next cycle, per CFPB rollover data. Escalating-dose test: confirmed for repeat borrowers. Diagnostic rejects repeat payday borrowing; alternative: one-time negotiated payment plan with the landlord, which resolves this cycle without creating a larger shortfall next cycle.

**Organizational:** A team facing a slipping deadline proposes shipping without the planned test coverage to hit the date. Relief: the date is hit. Harm mechanism check: untested code increases defect rate, and the resulting production incidents consume more engineering time than the coverage would have — the same channel (engineering time) that was "saved" is what gets spent, at a larger multiple, next cycle. Escalating-dose test: confirmed if the team repeats the shortcut on subsequent deadlines under the same pressure. Diagnostic rejects cutting test coverage as the fix; alternative: renegotiate scope or date, which resolves the deadline pressure through a separate channel.

## Common Mistakes

- **Treating every short-term-for-long-term tradeoff as a poison-cure trap.** Most tradeoffs (deferred maintenance, temporary budget cuts, working overtime for a sprint) have separable cost and relief channels and are bounded — applying this diagnostic to them produces false rejections of ordinary, sound tradeoffs.
- **Judging the fix by the size of the first dose instead of the escalating-dose trajectory.** "It's just this once" is exactly the reasoning the trap depends on; the diagnostic question is whether the second instance requires more, not whether the first instance is small.
- **Adopting the fix under emergency pressure without setting the cap and exit condition first.** Once the fix is in use, the same pressure that justified the first dose will justify the second; the cap only works if it precedes use.
- **Confusing this with Goodhart's Law.** A poison-cure trap is about a chosen remedy's causal structure, not about a metric being gamed once targeted — don't reach for `apply-goodharts-law` when the actual question is whether a proposed fix compounds its own problem.

## When NOT to Use

- When the relief mechanism and any downside are on separable channels — that's ordinary cost/benefit or cost-of-delay analysis (`apply-early-intervention`), not this diagnostic.
- When the "fix" is a one-time, non-repeatable action with no plausible future cycle (e.g., a single emergency surgery) — the escalating-dose test requires a repeat-cycle structure to apply.
- When evaluating whether a metric itself has been gamed rather than whether a remedy is self-reinforcing — use `apply-goodharts-law`.