I very curious about the performance of this query. If you have 10M relationships, but only 10 are of type KNOWS, will this query be optimized to lookup the relationships first (eg. relationship type index), or will it do what it appears to do, which is to iterate through every node in the database and test its relationships (which will be very, very, very much slower)?
For smaller databases this is not relevant, but for very large ones, some clever query pre-processing-optimization could make amazing differences in performance.