By dividing the spectrum of ORES scores into discrete levels and labeling these in a way that makes their predictive value clear, enable users to effectively use the system to meet their reviewing needs. In the immediate future, this system will be used on the [[ https://phabricator.wikimedia.org/tag/edit-review-improvements-rc-page/ | ERI revisions to Recent Changes ]] and in the [[ https://phabricator.wikimedia.org/tag/edit-review-improvements-reviewstream/ | ReviewStream ]] feed.
At present, here's how we're defining and describing these levels :
**CONTRIBUTION QUALITY [Damaging]**
**Very likely good**
Edits with 90-100% probability of being helpful.
**Likely good**
Edits with 50-100% probability of being helpful.
**May have problems**
Edits with a 20-100% probability of being flawed or damaging.
**Likely have problems**
Edits with a 50-100% probability of being a flawed or damaging.
**Very likely have problems**
Edits with a 90-100% probability of being flawed or damaging.
**USER INTENT (PREDICTED) [Good Faith]**
**Very likely good faith**
Edits with 93-100% probability of being in good faith.
**Likely good faith**
Edits with 50-100% probability of being in good faith.
**Likely bad faith**
Edits with 50-100% probability of being in bad faith.
**Very likely bad faith**
Edits with 90-100% probability of being in bad faith.