Axis codebook
Nine dimensions stay visible.
PoliBench does not collapse political-response profiles into a single rank.
Claim Evidence
Axis claims link to the pages that document the frozen instrument structure and per-axis diagnostics.
| Claim | Evidence |
|---|---|
| The nine-axis codebook is the frozen instrument structure for this release. | Axis definitions · Questions |
| Per-axis parsed coverage and diagnostics are auditable from release files. | Axis diagnostics · Axis intervals |
| Axis | Negative pole | Positive pole | Compass role |
|---|---|---|---|
| Economy | redistributive_public | market_low_tax | x |
| Liberty | civil_liberty | security_authority | y |
| Foreign Policy | restraint_dove | intervention_hawk | None |
| Nation | cosmopolitan_open | national_sovereign | None |
| Culture | progressive_change | traditional_stability | None |
| Governance | pluralist_institutional | executive_concentration | None |
| Secularism | secular_public_order | religious_public_order | None |
| Technology | precaution_human_continuity | acceleration_transhumanism | None |
| Deviance | constraint_bound_restraint | greater_good_override | None |
Evidence note
PoliBench is a public benchmark surface for model outputs under fixed political prompts. Each page should be read as evidence of what a model returned inside this benchmark, with the prompt set, parser, scorer, release files, and caveats kept close to the claim.
The site keeps the claims narrow on purpose. Scores describe response profiles, not provider intent, model beliefs, public opinion, or real-world political impact. Use the linked runs, model cards, artifacts, and validation pages to trace where a number came from before reusing it.
This note is repeated because the warning matters on every evidence page. A table can make a number look settled even when the right reading is narrower: one benchmark, one prompt set, one scoring pipeline, one published data surface, and explicit limits around human and external validation.