As I understand it the Harkness rule for dropping odd numbers in pairing groups used to be the norm. This rule drops the player right below the middle of the group rather than the lowest player in the group. The rule seems to make quite a bit of sense but I figured I might see a variety of opinions in this forum.

Some challenges I see in dropping the lowest rated player in a group. First if there are several unrated players the program may have to go a long way to find that lowest rated player. This may be particularly evident in scholastic groups. I also see where this leads to last round pairings that really work against the goal of having the two top players meet in the final round. For example if players number 1 2 and 9 are the only perfect scores in the next to last round then 1 would play 2 and 9 would be dropped. So there is no way 1 will play 2 in the final round. Finally there is the problem with dropping a player that since they are the lowest rated in the group if they win they remain the lowest rated and in an odd number for the next round are due to be dropped. I realize some review pairings to avoid this however there is no uniformity. In an extreme example if players 1-9 are all that have a perfect score then 9 is dropped. Assuming this go as planned 1-4 and 9 are left with a perfect score after that round. In the next round 9 would be due to be dropped again. If they won that round then 1,2 and 9 would be the remaining perfect scores. So 9 gets dropped into the next lower score group 3 times. In Harkness pairing in the first round of the example 5 not 9 would drop. In the next round 3 would drop and in the next to last round 2 would drop. My feeling is that we should reintroduce the Harkness system into the rules as a pairing variation.