Trust · SI Editorial Standard in public

Weekly public audit

Each week we publish a random sample of the items that were published automatically, with the score the independent SI critic gave each rule, plus how often the gate passes or rejects drafts and why. The sample is drawn with a fixed seed for the week, so anyone reloading this page sees the same items. Items are SI-generated and not individually reviewed by a person. Rules and thresholds are on the methodology page; fixes are in the corrections log.

← 1997-W03 Week 1997-W04 · Jan 20–Jan 26, 1997 1997-W05 →

“The sample”

0 of 0 automated publications this week (seed 199704).

No items were published automatically in this week.

“Gate pass and reject rates”

Last 7 days

ProductPublishedRejectedPass rateCritic pass / failRepairedHeldUnpublished / flagged later
VFA SI Solution 0 0 — 0 / 0 0 0 0 / 0
SI Think Tank idea 0 0 — 0 / 0 0 0 0 / 0

Last 30 days

ProductPublishedRejectedPass rateCritic pass / failRepairedHeldUnpublished / flagged later
explainer 0 0 — 6 / 262 147 0 0 / 0
SI Think Tank idea 3 7 30% 0 / 0 0 0 0 / 0
Founders' Verdict 0 75 0% 0 / 119 60 0 0 / 3
SI Senate measure 1 51 2% 1 / 81 41 0 0 / 0
VFA SI Solution 9 286 3% 9 / 520 289 0 0 / 0

Top rejection reasons

  • solution: unsupported claim — 591
  • solution: misapplied citation — 273
  • solution: critic verdict — 269
  • parliament: unsupported claim — 265
  • solution: constitution_support — 175
  • solution: source_grounding — 169
  • solution: mean score below threshold — 168
  • solution: honesty — 126
  • parliament: misapplied citation — 125
  • senate: unsupported claim — 61

Pass rate = published / (published + rejected) among drafts the gate decided. A draft that is held waits for a daily cap or a live source check and is counted when decided. Standard version si-editorial-standard/1.0.

Corrections log  ·  Methodology  ·  JSON