Search this site:

On Debconf5's scheduling

AJ Towns wonders how the hell the timetable didn’t seem to obey to any logic. Well… I agree with you: It was completely computer-generated, and there are many points that should have been taken into account, but were not. This is the second scheduling algorithm that has been used with Comas - I didn’t suggest our previous one, as Don Armstrong’s implementation introduced voting according to your interests so you don’t get locked out by having two interesting talks at once. Of course, this is the first time his algorithm is tried - but it is running on, sorry, incomplete information - For example, we don’t know (better yet: Our system doesn’t know) which proposals will be more popular than others, so they get scheduled in the bigger room. And, yes, it’s only a bigger room - it is not to give more importance to those talks. It’s almost random. The algorithm we used for CONSOL 2004 and 2005 took into account proposal type and track, trying to keep the density of any given track as homogeneous as possible during the whole conference - once again, to keep you from missing interesting stuff. If you look at the timetables, of course, they are not really balanced - but that’s because us humans had to tweak it afterwards. As you know, timetables are quite tricky. And, from what I understood from Don’s mail announcing the schedule first draft (sent to the debconf5-team list, not yet to everybody interested), this is by far not final. So… Well, thanks for mentioning this points, I think Don will be able to tell the system to move it around (or just reshuffle - the results will change).But, please, don’t act so hostile on us. Ok, I told you 30 minutes ago I could not keep my eyes open - Now it is official, I’m going to my bed. BTW, I’m still surrounded by my cats. If any of you has never heard a cat snore, you don’t know what you are missing! ;-)