That produces pretty much the same results as the CROSS JOIN I was using before. Because each "my_value" in the table are different, if I group on just their value then I will always have the full result set and a bunch of essentially duplicated results.
Any other ideas/options?
how do you want to solve the problem:
A is similar to B
B is similar to C
A IS NOT similar to C
?
I'm not sure that PostgreSQL is the best tool to solve clustering problems...though my experience with them is minimal.