Worked examples · What a number is actually made of · BRD-006

One dataset, four audiences

One risk register rendered for a board, an auditor, an engineer and a regulator. Each is correct, each is lossy, and their priority lists barely overlap.

OpenSynthetic dataFixed seed

This run

64 risks · 4 renderings · fixed seed

What the simulation shows

Sees every risk, ordered by money. Five of them carry half the exposure.100%

Sees every risk, ordered by controls remaining. Money is absent entirely.100%

Sees only stale reviews, absent owners and missing dates. Size of risk is absent.73%

Sees reportable domains only, and counts accepted risks as open.36%

Every figure here is generated from a fixed seed, so the numbers on this page are the numbers in the simulation.

The entry

The problem

A board pack, an audit report, an engineering backlog and a regulatory return are described as four views of one position. They are four lossy summaries.

Why it matters

Each drops something a different reader needed, and the disagreement surfaces in the room, in front of the person least able to reconcile it.

The approach

Render one generated register four ways, state plainly what each rendering discards, and measure how little the four priority lists share.

Judgement calls

No fifth rendering is proposed. A summary that drops nothing is the register, and the register does not fit on a slide. The argument is narrower: what should travel with each rendering is the sentence saying what it dropped.

Where else it applies

Any reporting line serving more than one audience from one dataset.

The data

Entirely synthetic, generated from a fixed seed. Prose in the page carries no counts, because a written number is true until the seed moves.

Open the simulation

It asks for your name and email, then opens. The page carries a watermark with the name of whoever opened it.

Open BRD-006 in a new tab →