Why a group_by + summarise of thousands of observations yielding a dataframe 1x1?

Question

Please, consider the following:

edx2 is a dataframe with 9000055 obs. of 8 variables;

movieId is a column of edx2 with more than 10,000 levels;

rating is a # column of edx2; mean(edx2$rating) is a valid number.

naive is a # constant;

My objective with the code below is to get a table relating a # outcome for each movieId level:

movie_dev <- edx2 %>%
group_by(movieId) %>%
summarize(movieid_dev = mean(rating - naive))

The result I'm getting is a 1x1 dataframe, instead of +10,000 x 2 variables. I don't understand why. Any clues?