Eval Integrity: order an audit
We check the "best known" values your benchmark or leaderboard scores against, against the primary literature, using the published method. Every verdict carries its source and a cut-off date.
Single-table freshness check
$990
Turnaround: report within 24 hours, with a human second pass.
- One stored table of best-known targets, at the version you pin
- A verdict per row: CURRENT, STALE, MALFORMED or UNCLEAR
- For every non-current row, the primary-source quote and citation
- A stated cut-off date for the search
- UNCLEAR rows reported, not counted
- Stale-target rate with a Wilson 95% interval
Full benchmark audit
$4,900
20% deposit ($980) now; balance on delivery.
Turnaround: report within 10 business days, with a blind second pass.
- Every target table in the benchmark, at the pinned version
- A verdict per row, with primary-source quote and cut-off date
- Blind second pass: a second reviewer re-checks every row without seeing the first verdicts
- UNCLEAR rows reported, not counted
- Leaderboard re-scored against audited values, with rank changes listed
- A certificate for any re-scored record
Continuous monitoring
$14,900
20% deposit ($2,980) now; balance on delivery.
Turnaround: a report every week of the term.
- A weekly literature sweep against every target in the benchmark
- A verdict per changed row, with primary-source quote and cut-off date
- UNCLEAR rows reported, not counted
- Alert on any target superseded since the last report
- A certificate for any re-scored record
Deposit refundable if no finding.
Questions, or a scope that does not fit a tier: support@ripplarity.com
See the method.