Accuracy
Every correction this site has issued, counted. The numbers below are computed from the correction record on every build — there is no second copy of them to fall out of date.
A zero here is not a boast. This wire publishes original summaries of other outlets' reporting, at volume, with a model doing the writing. It will get things wrong. If this page reads zero for long, the likeliest explanation by far is that we are not looking hard enough — not that we have been perfect. Found something wrong? Tell us and it will appear here.
Of these, 1 is a correction against a published dataset rather than a story — the on-chain ledger and the API served from it. Corrections per 100 counts stories published as its denominator, so a dataset correction is in the numerator and not the denominator; it is named here rather than left to be misread.
By month
By category
| What was wrong | Count |
|---|---|
| Figure mis-stated | 2 |
| Attribution imprecise | 0 |
| Translation term | 0 |
| Other | 1 |
How complete is this?
No gap ledger covers this dataset, so nothing here can say when its producer was not looking.
A count of our own errors, which is the one dataset here whose completeness we cannot measure: an error nobody has found is not in it, and no ledger records what we have not noticed. The gate tally below is the closest thing to a floor under it.
When we were not looking
No gap ledger covers this dataset.
Days that are not a day’s worth
This dataset is not a continuous record: a day with nothing on it is a quiet day in the world rather than a gap in ours, so no per-day expectation is stated.
Corrections
None issued against this dataset.
Our own downtime, inside this window
- 38 min degraded · The wire wrote no ledger row for 38 minutes (stage 'budget') not written up yet
- 44 min degraded · The wire wrote no ledger row for 44 minutes (stage 'budget') not written up yet
- 107 min degraded · The wire wrote no ledger row for 107 minutes (stage 'budget') not written up yet
- 47 min degraded · The wire wrote no ledger row for 47 minutes (stage 'budget') not written up yet
- 39 min degraded · The wire wrote no ledger row for 39 minutes (stage 'budget') not written up yet
- 46 min degraded · The wire wrote no ledger row for 46 minutes (stage 'budget') not written up yet
- 65 min degraded · The wire wrote no ledger row for 65 minutes (stage 'budget') not written up yet
- 57 min degraded · The wire wrote no ledger row for 57 minutes (stage 'budget') not written up yet
- 44 min degraded · The wire wrote no ledger row for 44 minutes (stage 'budget') not written up yet
- 39 min degraded · The wire wrote no ledger row for 39 minutes (stage 'budget') not written up yet
- 37 min degraded · The wire wrote no ledger row for 37 minutes (stage 'budget') not written up yet
- 31 min degraded · The wire wrote no ledger row for 31 minutes (stage 'budget') not written up yet
- 31 min degraded · The wire wrote no ledger row for 31 minutes (stage 'budget') not written up yet
- 31 min degraded · The wire wrote no ledger row for 31 minutes (stage 'budget') not written up yet
Anything known to be wrong?
5 problems are open in this site’s own checks; none of them is about this dataset.
Assembled from records this site already keeps — the gap ledgers, the corrections record, the incident record and the gate’s own tally. It adds no new collection. The same report ships with the download as /api/v1/accuracy.integrity.txt and inside the JSON. What this site cannot see at all.
Categories are assigned when the correction is filed. “Figure mis-stated” means a number was wrong; “attribution imprecise” means we credited the wrong outlet or over-stated how many independently reported it; “translation term” means an edition rendered a term in a way that changed its meaning.
Read the corrections themselves · How this site works · Uptime and incidents