Party tells you what they're called. Impressions tell you how they make you feel. Neither tells you what they've done.
This is a procedure for closing that gap. It does two things in order: it helps you name the standard you're actually judging by, then it shows you how to test a candidate against it with evidence instead of instinct.
Yours to name
What you valueSubjective. There is no correct answer, and this page will never suggest one. But it has to be conscious and specific, not absorbed from a party or a mood.
Checkable
Whether they meet itObjective. Once your standard is named, this is a matter of record — passed laws, who benefited, loops closed or left open. Two people can verify it and agree.
Most voting collapses these two into one move: you read the value off the party label and read “meets it” off a feeling. Keeping them apart is the whole method. You own the standard out loud; the record either clears it or it doesn't.
Name what you actually want, before you look at anyone.
Do this first, and do it blind — without a candidate in mind — so you're describing a standard, not reverse-engineering a justification for someone you already like. These prompts are structural on purpose: they ask for the shape of your criterion and leave the substance entirely to you.
What would disqualify a candidate no matter how good the rest of their record?
Your dealbreaker is the most useful thing you own, because it's a value you can check a record against directly. “I don't like them” isn't a floor. “They used someone's private life as a weapon” is — it's a thing that either happened or didn't.
When you say you want someone who “gets things done,” done how?
Passing new laws, blocking bad ones, and building coalitions are three different skills with three different evidence trails. Naming which one you mean tells you what to go looking for — and stops a candidate's strength in one from masquerading as proof of another.
When power is used, who do you want it pointed at?
Every record has a direction: it tends to benefit those with more leverage or those with less. Neither answer is the right one — that's yours. But naming it converts a vague allegiance into a test you can run on who actually benefited from what a candidate built.
Does it matter to you how a result was reached, or only that it was?
Some voters weight outcomes only; some treat the method as part of the result. Decide which you are. If method matters, you've just named a second, independent test — one a strong record can still fail.
If two of your criteria collide, which one wins?
A candidate may clear your floor on direction but not on means, or deliver on the skill you wanted toward an end you don't. Knowing your priority order in advance is what keeps the final call from collapsing back into an instinct at the moment it's hardest.
Four lenses, in order. Each one asks for evidence, not impressions.
Run them as a sequence — a candidate has to clear the earlier gates before the later ones matter. The lenses are the same for everyone; only your rubric from Engine 01 decides what counts as passing. That's what makes the procedure objective even though the verdict is yours.
Can they build at all?
Has this person ever turned intent into a finished outcome — in any arena?
- What counts as evidence
- A completed thing with their name on it: a law, a program, an organization, a case won. Capacity demonstrated somewhere, not merely claimed.
- What doesn't
- Titles held, years served, intentions stated, problems described. A résumé is a list of positions, not a list of outputs.
Did they build in this role?
Not capacity in the abstract — shipped work in the seat they're asking you to extend or hand them.
- What counts as evidence
- Legislation passed with a date and a vote count. Initiatives that closed. Things you can point to with a citation, ideally from a neutral record rather than their own campaign page.
- What doesn't
- Bills introduced but never passed. Press releases. “Fought for” and “championed” with no outcome attached. Read the verb: introduced, fought, and championed are not passed.
Toward what?
Describe the direction of the record before you judge it — characterize, then evaluate.
- What counts as evidence
- Who measurably benefited from what they passed, traceable from the text of the bill: which group, which ward, which side of a power gap. This step is descriptive — you're mapping the record, not scoring it yet.
- Where your rubric enters
- Now lay Prompt C against the map. The direction is a fact; whether it's the direction you wanted is your call. The lens finds the pattern; your standard grades it.
By what means?
How they operate, tested against the floor and the method-question you set in Engine 01.
- What counts as evidence
- Documented conduct: how they treat opponents, whether they deal directly or maneuver, whether their coalitions are built with the people affected or around them, whether they own mistakes or bury them.
- Where your rubric enters
- This is where a strong record can still fail your test — and where two honest voters most often part ways. A method that crosses your floor outranks any achievement above it, if that's how you weighted it in Prompt E.
A real method works even for a candidate you'd never pick.
If your framework only ever validates the people you already favored, it's a rationalization, not a procedure. The proof that you're reasoning rather than rationalizing: someone applying these same lenses with a different rubric should be able to reach a different candidate — honestly, and by the same route.