The Comparison Table
Overview
Section titled “Overview”Section 1 was about being someone whose reasoning gets read. This section is about the reasoning.
The central artifact is unglamorous: a table of options scored against criteria. Module 3 introduced it as a decision method. This topic is about it as a justification — a thing other people read, challenge, and act on — and about the one property that decides whether it persuades anyone.
The order of operations is the whole thing
Section titled “The order of operations is the whole thing”Criteria are fixed before options are scored, and the fixing is visible in the artifact.
That sentence is the difference between a justification and a rationalisation, and the second half matters as much as the first.
Everyone, including people with no agenda, adjusts criteria until the preferred option wins. It happens unconsciously: a dimension where your favourite scores badly starts to feel less important, and one where it excels starts to feel decisive. The only protection is sequence — write the criteria, commit to them, then score.
But a reader cannot see your sequence. They see a finished table in which your preference won, which is exactly what a rationalisation also looks like. So the artifact has to show the order:
- Criteria dated, or in a section that visibly precedes the options.
- Weights drawn from an earlier source — Module 2’s Utility Tree — rather than invented here.
- Any criterion added after scoring began flagged as such, with the reason.
- The scoring evidence per cell rather than a bare number.
A table where the criteria were visibly fixed first survives challenge. One where they might have been fitted afterwards does not, however honest its author was.
Veto columns are not scored columns
Section titled “Veto columns are not scored columns”Module 3’s distinction, and it is the most common structural error in these tables.
A constraint is pass or fail. An option that violates it is eliminated, not scored low. Giving a constraint a numeric score lets a strong total carry an option that is actually disqualified — which is how a compliance boundary gets quietly traded away by arithmetic nobody questioned.
graph LR
A["Candidate<br/>options"] --> B{"Violates a<br/>constraint?"}
B -->|"yes"| C["<b>Eliminated</b><br/><i>never scored</i>"]
B -->|"no"| D["Scored against<br/>criteria fixed earlier"]
D --> E["Decision, with<br/>evidence per cell"]
classDef start fill:#5B8FB9,stroke:#4A7A9E,stroke-width:2px,color:#ffffff
classDef veto fill:#D9695B,stroke:#B4483A,stroke-width:2px,color:#ffffff
classDef out fill:#D8D3CB,stroke:#B8B2A8,stroke-width:2px,color:#3F3F3F
classDef score fill:#E8A33D,stroke:#C4832A,stroke-width:2px,color:#ffffff
classDef dec fill:#7D9A4E,stroke:#5F7A38,stroke-width:2px,color:#ffffff
class A start
class B veto
class C out
class D score
class E dec
Elimination precedes scoring. The drained box is where disqualified options go, and they leave the table rather than competing on points.
Criteria that actually discriminate
Section titled “Criteria that actually discriminate”Module 3 made the point and it applies directly to persuasion: a criterion earns a column only if serious options score differently on it.
Dimensions where everything ties are padding, and padding is how a table becomes impressive and useless. Worse, it is how a table becomes gameable — a preferred option can accumulate soft points across many tied dimensions and win on total while losing on everything that matters.
Six discriminating criteria beat twenty with fourteen ties, and the shorter table is easier to defend because every row is doing work.
Evidence per cell, not scores
Section titled “Evidence per cell, not scores”The upgrade that converts a table from an opinion into an audit trail.
A cell reading “4/5” is a judgement the reader must accept. A cell naming the measured figure, the dataset it came from and the date it was taken is a fact they can check. Module 4 made this the standard for model selection; it generalises to every justification you write.
Where you have no evidence, say so in the cell — not measured; would require a two-week shadow run — rather than supplying a number. An honest gap is more credible than a confident guess, and it has the useful side effect of showing exactly what somebody would need to fund to close it.
What the table costs
Section titled “What the table costs”It is slow. Fixing criteria before scoring means a day or two before anyone sees anything, and in an organisation expecting answers that reads as delay.
It also exposes you. A table with explicit criteria invites challenge on the criteria, which is a more sophisticated attack than challenging a conclusion. That is the point — a challenge to your criteria is a conversation worth having, and far better than a conclusion that gets quietly ignored.
And it cannot rescue a bad decision. A well-constructed table around a weak choice makes the weakness legible, which is a feature.
Feeds the workbook
Section titled “Feeds the workbook”This is the core of your decision justification pack, and two rubric items land here: criteria fixed before scoring with that order visible, and the table structured so constraints eliminate rather than score.
→ A table shows what you chose. Credibility comes from what you turned down. — Rejected options and uncertainty