Some Final CPAIOR Thoughts

I have returned from Paris (my original plan was to go to Italy for IPCO, but I had to change that). Here are some thoughts from the CPAIOR conference:

1. It is impossible to blog while being Program Chair, particularly in Europe (for me). Program Chair at the conference is the easiest job: the choices have all been made, so it is really a matter of sitting back and seeing how everything turned out. But I was nervous about the timing, since the schedule was more packed in that I had liked, so I sat near the front to provide pressure on speakers to keep to their time. Being in the front, I didn’t feel right tapping on my notebook, so I didn’t blog during the talks. But jetlag made me fall asleep anytime I was within fifty feet of a bed, so I couldn’t blog from my hotel room. Hence, just a few comments now!
2. Pascal van Hentenryck gave a fascinating review of the history of constraint programming, based on an imaginary conversation with his advisor J.L. Lauriere describing what has happened to CP over the last 30 years (Lauriere died about 5 years ago). Lauriere wrote the pioneering paper in constraint programming back in 1978, centered around a language entitled ALICE. It was really neat to see how many of the ideas we have today are within that 1978 paper (though the 1978 paper and language structure is such that it is really hard to see how we got to where we got). I had not heard of Lauriere, but that is perhaps not so surprising: constraint programming is not my main home base, and Lauriere was perhaps best known within France. Correction added June 2.  Lauriere was on Pascal’s committe, but his adviser at the University of Namur was Baudouin Le Charlier.
3. Francois Laburthe of Amadeus (a company like Sabre) gave the final plenary, and it complemented Cindy Barnhart’s very well. Francois talked about the business side of airline schedules: how do you show them to customers and determine the best routes for them. Companies like Orbitz and Expedia do this all the time, and I hadn’t realized how hard that it is to do. One stat of his I liked: pre-Internet, systems were designed to serve five price queries for every ticket sold. Now the number is closer to 1000. Given the number of searches I do before every trip, I can believe it!

Overall, I thought the conference went very well. Francois Fages, who was responsible for local organization among other things, chose a very nice boat on the Seine for dinner, and everything went very well in general.

In addition to Cindy Barnhart’s talk, I guess I most liked the workshop that Robin Lougee-Heimer put together on open-source solvers in CP and operations research. I was a little surprised that most open source constraint programming systems don’t do it for the community of developers. For the most part, they do it as a convenient license for distribution. Only COIN-OR really seems to be working to get a community of people working at improving code.

It was a great conference. Next year, it will be in Pittsburgh, where my colleagues John Hooker and Willem-Jan van Hoeve have to do all the work.