WARNING: prioritizing work on this task cannot happen until T387918 is completed. Reason: this analysis will confirm the extent to which the findings from T394463 prove to be statistically significant.
This ticket holds the work of running a follow-on experiment to T387918 wherein we will learn how different Tone Check model confidence thresholds impact editing behavior.
Background
The idea/need for this experiment emerged in response to the following:
- In T394463 we learned that Tone Check appears impactful and targeted
- In T394463#11282441, we decided not to explore this adjustment in the A/B test that's ongoing