They were sorted into categories (wikimedia, technology, free culture, etc), and slots were apportioned by approximately how many high ranking (>8) submissions there were in each category. The idea was to provide a good programme for a diversity of interests.
I had to make a couple of judgement calls, especially on sessions that asked for a longer slot. How you compare a one hour session rated 8.6 to two 30 minute sessions rated 8.4?
Bear in mind the ranking system isn't perfect; it's hard to rank consistently across 500 items, and the difference between an 8.4 and an 8.6 is just noise, really. Anything above 8 is already a "strong accept"!