The ladder is printed before the candidate is read
Each heading carries nine concrete policies at 100L, 75L, 50L, 25L, 0, 25R, 50R, 75R and 100R. Placement is choosing the printed policy that matches the candidate's own stated position. It is not an impression of how far one way somebody seems, and a rung that two careful readers place differently is a badly written rung, not a hard call.
Resolution comes from more printed rungs, never from interpolation
The obvious way to sharpen a scale is to let a grader place someone "between" two rungs. That quietly restores the judgement the ladder exists to remove, and it does it invisibly, because there is nothing printed for a reader to argue with. So the in-between positions were written out instead: 44 additional rungs at 25 and 75 on each side, each one a distinct policy with its own statute. A half rung had to pass one test to be allowed on — somebody real advocates this and not its neighbours — and anything that failed it was cut rather than kept as padding.
Every rung names a real law
Each of the 99 rungs carries a referent: a bill, a rule or an enacted programme that actually occupies that position. A rung defined only by adjectives can be slid quietly in whichever direction suits the desk; a rung fixed to a statute cannot. Where a grader is unsure what 50R means on a heading, they read the law printed beside it.
The spacing is checked against positions already in circulation
Under each ladder is a short list of well-documented positions and where they land. If the rungs were badly placed, real positions would bunch at one end or disappear into the middle — so a reader who thinks the ladder leans can point at a specific row rather than at a feeling. No candidate on the board is used for calibration, because measuring an instrument against the thing it measures proves nothing.
The centre rung is a real position
Zero on this scale is a policy somebody actually holds — enforce current law, extend current rates, keep the current structure — not 'we could not tell'. A scale whose middle means uncertainty converts every silence into moderation, which is the most common way these instruments lie.
Silence is excluded, and counted out loud
Headings a campaign has not addressed are unscored and appear in full underneath every placement. No aggregate is published below 6 of 11 headings — the instrument declines instead of estimating. A thin record produces no number here, however tempting the number would be.
Every heading weighs the same
Salience weighting is the strongest lever on the final figure and the hardest to falsify, so this scale does not use it. If that ever changes, the weights are published and dated here before any candidate is placed against them.
The number never travels alone
A placement is published only with its coverage and its spread. A candidate placed hard one way on some headings and the other way on others is not a centrist, and the spread is what prevents the mean from claiming otherwise.
Thin evidence widens the figure, it never moves it
A placement supported by one general line of campaign copy is not a placement toward the centre — it is a placement the record cannot pin down, and the honest way to publish it is wide. Each placement carries one of three printed evidence grades whose widths are fixed in advance: explicit publishes a point, adjacent a band 12 either side, thin a band 25 either side. The grader chooses a grade, never a width, and the band travels through the arithmetic into the overall figure.
The published figure is coarser than the arithmetic, on purpose
The axis runs to 100 a side, but the ladder's finest real distinction is 25 — the gap between two printed rungs. A figure like 37L would therefore claim a precision the record cannot supply. Published figures are quantised to the nearest 5, 20 stops a side: coarse enough that no digit outruns the ladder, fine enough that two different band edges do not collapse into one number. The rounding happens at the glass and never in the arithmetic — aggregates are computed from unrounded parts — and the computed value is printed beside every published figure so the mean can be checked rather than taken.
Placements are drafted until they are signed
A drafted placement has been made against the ledger cell and the printed ladder. It becomes signed when the editor of record countersigns it. The badge on each row says which, and it never claims the state above.