Read Achilles markRead Achilles

Risk Management

The Drill Is Not the Fire

Every institution I have advised runs stress tests, and most run them well: serious scenarios, honest numbers, a committee that reads the results with appropriate gravity. And yet I have come to distrust the drill, for…

♥ 0 Sign in to like 0 comments

This post’s retention & record terms are stated in full in the colophon below.

Every institution I have advised runs stress tests, and most run them well: serious scenarios, honest numbers, a committee that reads the results with appropriate gravity. And yet I have come to distrust the drill, for a reason that took me years to say plainly. A stress test rehearses the institution's *arithmetic*. A crisis tests its *behaviour* — who may decide, how fast, on what information, and against whose objections. The drill almost never rehearses the second list, because the second list is embarrassing to write down. [author assertion]

Watch what a real funding stress actually consumes. It does not consume capital first; it consumes *permission*. The treasurer knows what to do and cannot do it until a committee that meets on Thursdays agrees it is Thursday. The collateral is available and the person who can release it is on a plane. The scenario in the drill assumed a decision-maker the way a recipe assumes a cook. Fires are not interested in our assumptions, and neither, in my experience, are crises. [author assertion]

The drill has a second, quieter defect: it trains the wrong confidence. A test the institution passes every year, on scenarios the institution wrote, graded by the institution, becomes evidence of safety in the precise way a model becomes trusted — smoothly, plausibly, month after month. The scenario team and the risk committee read the same newspapers, fear the same recent crisis, and exclude the same unthinkable one. Every drill I have ever seen was, in part, a rehearsal for the *previous* fire, conducted by people whose careers the previous fire had already edited. [author assertion]

None of this is an argument against testing. It is an argument about what the test is *of*. I now advise treasuries to run two drills, not one. The first is the arithmetic one they already run. The second takes the same scenario and removes, on the morning of the exercise, two names chosen at random from the decision chain, one system, and the assumption that the first call will be answered. What breaks then is never the funding plan. It is the choreography — and choreography, unlike capital, can be repaired in peacetime for almost nothing. [author assertion]

The deeper point is the one underneath all my work: a drill, like a model, is an opinion with an org chart, and passing it feels like a verdict when it is only ever a rehearsal. Institutions should hold their passing grades the way sailors hold a forecast — gratefully, and with a hand on the rail. The question I leave with any committee that has just passed its annual test is the one I would want asked of me: your drill assumed a decision-maker, a working phone tree, and a borrower of last resort's patience. Which of those three have you actually *met* — and which have you only ever read about in the scenario document? [author assertion]

Julia Zhao

All posts by Julia Zhao → · Follow Julia Zhao by RSS

The Board 20 essays 0 corrections Agent citations: —— — citation attached ≠ claim validated As of 16 November 2026 Full board →

Provenance

About this version

How this post was verified, labelled and shared — in plain language first. The complete machine-readable record is kept below, word for word, for anyone who wants to check it.

Version
1 · in good standing
Author
Julia Zhao · verified at publication
Published
10 Nov 2026
Claims
5 labelled — 5 author assertion
Citations
No external citations
Distribution
Shared with the author’s consent — including to agents acting for readers, under the author’s per-post opt-in
Content fingerprint
fdf63ea5…e81de0
Full verification record

The verbatim graded record of this version: identifiers, claim classes, citation targets, hashes and standing, exactly as issued.

This record is also published in structured text for agents and tools — the same fields, word for word. View structured text

House editorial — Read AchillesAuthor verification: verifiedStanding: governingHash: fdf63ea54e4954fc7df6bf22eed3ea598ba9a1b8fb198f751298970f94e81de0

Version & standing

Post post-the-drill-is-not-the-fire-31ed77 · version 1 (post-the-drill-is-not-the-fire-31ed77-v1)

@julia-zhao · verified state at publish: verified

Published: 2026-11-10T14:09:08.878Z · Standing: governing

Claims & citations

  • c-1: author_assertion
  • c-2: author_assertion
  • c-3: author_assertion
  • c-4: author_assertion
  • c-5: author_assertion

    Disclosure & trust

    When your words reach the end of their retention period they are destroyed, on the schedule above. What the platform keeps after that — indefinitely, with no expiry — is a skeleton of acts: that an act occurred, its class and time, the bundle and section it touched, its change class, a hash that identifies the record without containing it, and the keys that link a correction to what it corrects. The skeleton is not your words and cannot reproduce them. Your elections to shorten a period, to ask ephemerally, or to skeletonise now govern your words; they do not shorten, and cannot remove, the skeleton — it is the record that lets the platform prove what it did with your work after the words are gone. The platform will never produce your skeleton by choice against you; like any record it can be reached by a court order the platform cannot refuse, and if that happens you will be told wherever the law allows and the production will state, on its face, that it is a skeleton and that the words expired on this schedule.

    When your words reach the end of their retention period they are destroyed, on the schedule above. What the platform keeps after that — indefinitely, with no expiry — is a skeleton of acts: that an act occurred, its class and time, the bundle and section it touched, its change class, a hash that identifies the record without containing it, and the keys that link a correction to what it corrects. The skeleton is not your words and cannot reproduce them. Your elections to shorten a period, to ask ephemerally, or to skeletonise now govern your words; they do not shorten, and cannot remove, the skeleton — it is the record that lets the platform prove what it did with your work after the words are gone. The platform will never produce your skeleton by choice against you; like any record it can be reached by a court order the platform cannot refuse, and if that happens you will be told wherever the law allows and the production will state, on its face, that it is a skeleton and that the words expired on this schedule.

    Integrity & distribution

    Content hash: fdf63ea54e4954fc7df6bf22eed3ea598ba9a1b8fb198f751298970f94e81de0

    Distribution: opted_in

    Canonical presentation: https://blog.readachilles.com/post-the-drill-is-not-the-fire-31ed77/

    What we retain, and why — the retention disclosure

    The retention disclosure for this version is the disclosure line of record, stated in full in the colophon fields above.