BrowserUk,
I have already maximized the efficiency in determining the distance between two selections using bitmasks. As stated, I know that full cross comparison is necessary in some cases. I was just hoping that I could prune that full cross comparison in cases where I knew a compare couldn't lead to a higher distance than I already have. There is one trivial case where this short circuiting works (when 2 items are the maximum distance possible). I was hoping there might be others assuming the data will be mostly similar.