It was more work than expected. I actually started doing a manual count, then decided halfway that it was too hard and error-prone, and started writing the vote counting script.
The script is here: http://unicast.org/download/vote.tcl
The consensus after last election was that the next election should be done with more automation, perhaps based on the simple-survey package. This was complicated by the fact that last time there were a couple of eligible voters who did not have openacs.org accounts.
