Bright searches through all of the people who starred a particular repo, then looks at how many stars each of their repos have. It sums the total number of stars from each user to compute a "Brightness" score for a repo. It also creates a tree for visualizing what repos that your stargazers have made and which of those repos have a lot of stars.
The intuition behind this is that a user who has a cumulative 1000 github stars is likely more experienced and knowledgable than a user with 0 github stars. Repositories which have thousands of stars from fake bot accounts (whose accounts would have 0 cumulative stars) would have a brightness score of zero, but a repository which seems unimportant but had a highly experienced developer star it would have a high brightness score.
The intuition behind this is that a user who has a cumulative 1000 github stars is likely more experienced and knowledgable than a user with 0 github stars. Repositories which have thousands of stars from fake bot accounts (whose accounts would have 0 cumulative stars) would have a brightness score of zero, but a repository which seems unimportant but had a highly experienced developer star it would have a high brightness score.