What do we need to examine now?

• Will multiple raters come up with the same scores for the same output?

• Can multiple labs set up on their own and do the same?

• Can a rater come up with the same result reliably?

• What is the relationship between scores and user satisfaction/intelligibility?