>>Another way is your log idea, but then select all records from the main file that match the basic criteria, with the summed total number of "featured appearances" from the MOTD table, ordered by the number of appearances, then just pick the first two.
>
>Whoah! Lets not get drastic here, maybe some of the duplicates are duped for a reason? :-) I think you're just jealous. ;-)

This is not fair - somebody was three times and somebody - 0 ... The theory of probability just doesn't work :)