THE ORIGINAL FRAMEWORK // CC BY 4.0 // DOI 10.5281/ZENODO.21864687

Price What Your Judgment Is Worth.

The Judgment Multiple expresses the return on human verification of AI-generated work. It turns "someone should really check this" into a number a CFO can act on. A multiple of 40x means every dollar spent on verification prevents forty dollars of expected loss.

THE JUDGMENT MULTIPLE (IGNATIUSTHEYOUNGERAI, 2026) = ANNUALIZED EXPOSURE ÷ ANNUALIZED COST OF THE HUMAN CONTROL

PUBLISHED JULY 2026 IN THE AI "KEEP YOUR CAREER" BIBLE · FREE TO USE, TEACH, AND REPUBLISH WITH ATTRIBUTION

01

Compute Yours

YOUR NUMBERS, YOUR MULTIPLE · NOTHING LEAVES THIS PAGE

THE EXPOSURE · WHAT MISTAKES WOULD COST

Quotes, summaries, filings, replies. Whatever your role ships.
Share of outputs with a consequential mistake.

THE CONTROL · WHAT CATCHING THEM COSTS

Minutes per review × reviews per year × loaded hourly rate.
~155x
YOUR JUDGMENT MULTIPLE MODELED — NOT MEASURED
ANNUALIZED EXPOSURE$241,920
ANNUALIZED CONTROL COST$1,560

Every dollar spent on this check prevents a modeled $155 of exposure. Show the assumptions with the number: that honesty is the whole method.

02

The Worked Example

FAILURE MODE 10 · SYCOPHANCY · THE CALCULATOR'S DEFAULT NUMBERS

The Setup

  1. 400 quotes a month at a $42,000 average value.
  2. 3% carry a consequential error, at a 4-point impact on value.
  3. Annualized exposure: ~$242,000.
  4. The control: ten-minute reviews at a $65/hour loaded rate, $1,560 a year.
  5. The multiple: ~155x.

Change any input above and watch the number move. That movement is the conversation with your manager.

Rules For Honest Use

  1. Model, don't assert. Show the assumptions and invite people to change them.
  2. Ranges over false precision. "$180K–$260K" beats "$241,920."
  3. Label modeled vs. measured. Presenting an estimate as an observed loss makes you the problem this framework exists to catch.

The Judgment Multiple is not the accounting term "control ratio." It borrows the language of valuation multiples on purpose: it prices a return.

03

The Range Across All 24 Modes

MODELED · THE LOW END IS PUBLISHED ON PURPOSE

HIGHEST · MODE 22
~19x TO ~385x
MOST MODES
~15x TO ~167x
LOWEST · MODE 5
~0.7x TO ~10x

A framework that only ever produces flattering numbers is a sales pitch. Mode 5 can come out below 1x, and the book says so: sometimes the check is not worth the cost, and knowing that is also worth money. Every mode's range is in the index →

Citing this framework? The Judgment Multiple, IgnatiusTheYoungerAI LLC, July 2026. DOI 10.5281/zenodo.21864687, licensed CC BY 4.0. Ready-to-copy formats in every style: the citation page.

Turn The Number Into The Case

The book carries the full method: the 24 failure modes behind the exposure column, the control that catches each one, and the 90-day plan that replaces these modeled defaults with your own measured numbers.

GET THE BOOK · $29READ A CHAPTER FREE →