The Wall

There are two very different kinds of evidence about a politician, and this product refuses to mix them. What a person did with the power they held — how they voted, what they authored, what they blocked — is one kind. What they have said, studied, promised, and been researched on is the other. The Wall is the rule that keeps each kind out of the other's award.

Two evidence streams run down two separate engines to two separate results, with a wall between them that no input, number, or adjustment crosses.

What it does

Exactly three classes of act reach the Monkey Award and the Llama Award: how a person voted at any stage, from subcommittee through the floor; what they authored or sponsored, including cosponsorship, amendments, and discharge-petition signatures; and what they blocked from advancing, using a position of power to keep a vote from happening at all. Every one is a dated act of official power taken in a seat the person was elected to, cited to the official record.

Everything else the movement holds about a person reaches only the Flamingo Award. That includes transcripts of interviews and podcasts, background research, a resume, an education record, public statements and commitments, qualification assessments, and what citizens themselves have said about them.

This is a prohibition on a data path, not a display convention. The two sides are produced by two separate engines. The voting-record engine has no reader for a transcript. The Flamingo engine has no reader for a roll call, an authorship row, or an obstruction record. Not switched off — absent. Adding one would be a change to the software itself.

• The voting-record engine may open the vote stores, the obstruction records, the two curated law lists, and the roster. Nothing else.
• The Flamingo engine may open the per-video assessment files, the qualification catalog, the roster, and the citizen assessment store. Nothing else.
• Neither award is ever a term in the other's arithmetic. No netting, no damping, no tie-break, no confidence adjustment.

A person who holds both kinds of record gets both results side by side, at the same size. A challenger who previously held office carries a voting-record award and a Flamingo, and neither shrinks, annotates, or suppresses the other because it is unflattering.

Enforcement is two tests that fail the build. The first names the file and the line if any statement in the backend carries a qualification reading into a Monkey or Llama value. The second fails if a Monkey or Llama term reaches a Flamingo result, including inside a text string, because a filename written between quotes is still a way in.

The goal it serves

The heaviest thing this movement publishes about a living person is a statement about how they used official power. That claim has to be answerable purely from government records, so that nobody — including us — can say an interpretation of what somebody meant in an interview moved a figure about their voting record. Two provenances, two awards, each argued on its own terms.

The two kinds of evidence do not have the same standing, and pretending otherwise is how a checkable claim quietly becomes an opinion. A roll call is a public fact with an official URL. A reading of what a person meant in a podcast is this movement's interpretation of their words. Blend them and the result has a provenance nobody can state, and a figure nobody can trace is exactly the thing this product exists to replace.

The Wall is also the only way a challenger and a sitting official can be compared honestly. Most people worth electing have never held office, so their voting side is legitimately empty. If interview evidence could lift a Llama Award, somebody who never cast a vote could out-earn the member who actually filed the repeal, and the Llama's whole claim — that it counts what a person did about those laws — would be false. Keeping the record award empty where the record is empty is what makes it mean what it says.

What keeps it honest

The Wall costs something, and the cost is the proof it is real. A challenger with no roll calls used to get a voting-record reading derived from their qualification assessments, so their page would not sit empty. That derivation was deleted rather than fenced off, because a boundary that has to be defended forever against a standing incentive erodes, and a deletion does not. The assessment it produced measured something real, and it is still published at full strength — as a Flamingo Award, with its own percentage.

• A blank is never a verdict. A Flamingo that reads unknown means we have not read enough about that person yet. It is our gap, not a finding about them, and it never renders as 0%.
• An empty voting record means no record, not a clean one, and not a bad one.
• How much we know is published too. A Flamingo Award carries a plain "N% known" beside it, so a percentage cannot be over-read as a complete picture of a person.
• The Wall does not claim the two sides are equally strong. It claims they must never be added together.

Works with

• The Monkey Award — the damage side of the voting record, and one of the two awards the Wall protects from interpretation.
• The Llama Award — the repair side, computed by the same engine over the same acts, and empty when the record is empty.
• The Flamingo Award — everything on the far side of the Wall, with three states of its own and its own separate engine.

Where to go next

• Politicians — the full doctrine: an award is earned or not earned, never a judgment issued on a human being.
• Where the data comes from — the official sources every act on the voting-record side is cited to.
• The three states — why not earned and unknown are a difference in fact, not in styling.
• How It Works — where the awards sit inside the loop a citizen actually runs.
• Get Started — run it on your own machine and check the arithmetic against the record yourself.