Changelog
Every change to the tests and scoring rules, and every batch of retests, with the date and who it affects.
Rubric v0.1 published
Ten scored dimensions, one published test each, scored from three checkpoints and the number of corrections. Personality is not scored.
Affects: All assistantsOverall score accounts for coverage
Every assistant starts with three neutral scores of 5, which pulls assistants tested on fewer dimensions toward the middle. At least five scored dimensions are needed to be ranked.
Affects: All assistantsScores expire after 90 days
Older scores still count but are shown in grey until the assistant is retested.
Affects: All assistants