Here is a writeup for our participation in the ICFP contest:
I had lots of fun. We didn't do very well with the assigned task but we managed to send submissions (lightning and normal).
I really enjoyed working with Andrés and with Juan Manuel for the first time in a contest. And of course working again with Luis Felipe was a good experience.
In previous years we had issues with coordination because we were working from different cities and countries and it made our experience seem like a contest between team mates.
I don't think it happened because we're evil or because we didn't want to work with each other but rather because in previous years we had people wanting to write code in Python, in Common LISP and in Scheme. This year we standardized in Python even when half of the team didn't know the language very well before the contest and it did pay off. Even the new Python users liked the idea.
We used some C++ code from Python thanks to swig.
We also met in real life and this helped a lot. The inter-personal communication bandwidth of a blackboard and a real-life meeting is by far bigger than the bandwidth of IRC.
Perhaps in the future if we work from different cities or countries we should try conference calls with Skype (I'd prefer a free software alternative but Skype is OK if we can't find one) and a virtual blackboard. Not all the time but for periodic team meetings and and peer to peer meetings.
Working in teams has a big overhead but it is more fun, specially if you don't care -that much- about the scores. When you work in a team in a contest like this one (with a single goal) you might not achieve your typical maximum performance but it don't think it really matters. The performance of the team is what matters and we have to improve it.
At some moment you should realize that it doesn't matter if you notice that the problem can be solved in a better way (specially if you don't have time to finish and debug a new implementation). You will notice that but it doesn't matter much. When you have reached the point of no return you should focus in making a good submission (specially when you have to send source code).
If the team is big perhaps it can be divided and try different approaches and then send the one that performs better. But it would be better if you do this after you've built the infrastructure that you need (simulators, visualization, etc) so that you can feel comfortable with this code-base and you don't feel like you need to reinvent the wheel.
Let's hope I can participate again next year and share a good time with my friends :-)
Discussion (4)Last update: 2008-07-17 (Rev 14396)