Who Signs? Decision Accountability as a Forensic Standard

A decision is undocumented if a stranger cannot answer five questions in five minutes

The paper defines the Five-Minute Test as a forensic standard for corporate decisions. A record passes only if someone outside the room can reconstruct the decision without tacit institutional knowledge.

• The five questions are: what was decided, who decided it, why, under what authority, and what would trigger reconsideration.
• A single 'I can't tell from this record' means the reasoning has not been retained.
• The intended reader is a future stranger: regulator, board member, journalist, prosecutor, investigator, bankruptcy examiner, or plaintiff's lawyer.

The stranger removes insider assumptions from the record

The test works because insiders know the code words, hierarchy, and omissions that outsiders cannot supply. The stranger reads only what the record says.

• Five minutes is the paper's boundary for examiner patience, not a metaphor for casual review.
• A record that depends on 'we all understood' fails because future investigators were not in the room.
• The forensic risk is a vacuum: when the record cannot establish what happened, anyone can assert what happened.

The Meridian loan passes because every answer is named, quantified, and testable

The paper's model record is a $4.2M commercial real estate loan to Meridian Development LLC for the River Walk mixed-use project. It shows the level of specificity the standard demands.

• Decision: 4.2M loan, 7.1% fixed rate, 25-year amortization, 75% LTV, based on a 5.6M independent appraisal dated March 12.
• Decision-maker and authority: Sarah Chen, VP Commercial Lending, Branch 14, within a $5M approval limit under Policy CL-2019-04.
• Review triggers: River Walk corridor occupancy below 85%, or Meridian DSCR on any existing project below 1.4x.

The ordinary risk committee minute scores zero because it records discussion, not decision

The paper contrasts the Meridian record with a typical committee minute: 'reviewed the portfolio,' 'discussed market conditions,' and 'agreed to maintain current risk limits.'

• What fails: 'current risk limits' does not identify which limits or levels were maintained.
• Who fails: 'the committee' names a forum, not an accountable person.
• Why, authority, and reconsideration fail: the record gives no evidence, alternatives, delegation, threshold, or review condition.

The 737 MAX single-sensor MCAS choice lacked accountable approval rationale

The paper treats MCAS as the central case. Boeing chose an MCAS architecture using one angle-of-attack sensor, without an accountable rationale showing who approved it, why alternatives were rejected, or when it would be reviewed.

• The known failure mode: one faulty AOA sensor could send false data to MCAS and produce repeated nose-down commands overriding pilot inputs.
• The House Transportation Committee sought design review records, risk assessments, authority records, and alternative analysis; the paper says there was nothing.
• The decision killed 346 people in Lion Air 610 and Ethiopian Airlines 302.

MCAS partially answers 'what' and fails the other four questions from the record

The paper applies the Five-Minute Test question by question to Boeing's single-sensor MCAS architecture.

• What: partially reconstructible as single-sensor MCAS, but not from one accountable approval rationale.
• Who: unknown from the record; multiple engineers were involved, but no individual signed the paper.
• Why and authority: cost reduction was reconstructed from emails and testimony; ODA delegation existed, but its use on this specific safety-critical decision was not recorded.

The paper contrasts about 1M per aircraft with 20B and 346 lives

The paper does not claim the cost logic was recorded contemporaneously. It says the rationale was reconstructed later: dual-sensor architecture would have required wiring, a comparator module, revised pilot training, and revised type certification.

• Estimated saving: approximately $1 million per aircraft.
• Crash cost cited by the paper: $20 billion in settlements, grounding costs, lost orders, and reputational damage.
• Review trigger: none in the record; no threshold, review date, or condition required reconsideration before the crashes.

SVB, Purdue, and FTX repeat the same forensic failure: no usable accountability trail

The paper uses later cases to show that the MCAS problem is not industry-specific. The accountability trail disappears in banking, pharmaceuticals, and crypto governance.

• SVB held $91 billion in held-to-maturity securities, but its interest rate risk model did not include rapid rate increases
• Purdue's chronic non-cancer pain strategy for OxyContin relied on a 101-word NEJM letter, with reasoning reconstructed decades later through litigation discovery.
• FTX was a 32 billion exchange using QuickBooks, with no board meetings, no minutes, and no record of the 8.7 billion customer-fund transfer to Alameda Research.

The paper says missing records are structural, not accidental

The absence of records protects decision-makers from being measured against what they knew, predicted, and chose at the time.

• Written reasoning can later be tested against reality; silence allows post hoc defenses such as 'we didn't know' or 'the risks were unforeseeable.'
• Discovery makes documents costly in litigation, so legal caution can produce weak governance records.
• Records require time and expose disagreement: the paper names Boeing engineers, Purdue scientists, and SVB risk managers as examples of warnings that complete records would have surfaced.

Decision Accounting expands the five questions into named fields

The Five-Minute Test is the diagnostic. Decision Accounting is the paper's treatment: a 17-field record that forces specificity where corporate minutes usually stay vague.

• WHAT maps to Field 2: the decision must be stated precisely enough to verify implementation.
• WHO and authority map to Fields 1 and 7: a name without authority is a scapegoat; authority without a name is a shield.
• WHY maps to Fields 5 and 6: reasoning plus evidence makes the decision reconstructable; either one alone is incomplete.

The paper ties each missing field to a named organizational failure

begins showing that each Decision Accounting field answers a question whose absence has destroyed organizations.

• WHO: Volkswagen's defeat-device decision affected eleven million diesel vehicles, yet no single decision-maker could be identified.
• WHAT: FTX had no board resolution, minutes, or memo describing the $8.7 billion transfer to Alameda Research.
• WHERE: Danske Bank's Estonian branch processed approximately $230 billion in suspicious transactions from 2007 to 2015, making jurisdiction and oversight part of the decision record.

Accountable rationale capture would have made MCAS visible before the crashes

The paper is careful: a stronger accountability regime would not have automatically vetoed Boeing's choice. It would have made the choice, its rationale, its authority, its known failure mode, and its system consequences explicit before deployment.

• The MCAS accountability statement would have stated: single AOA sensor input without cross-check or pilot notification.
• It would have listed FAA service difficulty reports, failure-mode analysis, approver qualifications, and the consequence of uncommanded nose-down trim.
• The paper's governing line is concrete: if you cannot explain what you did to someone who was not there, in terms they can verify, you did not govern.