I competed in ACM-ICPC for the last three years. I was also an ACM-ICPC World Finalist in 2016. That year was a peak in my ACM-ICPC career as I competed in six(!) contests: three local contests, two regionals, and the world finals.
This year, I retired from competing in the ACM-ICPC. My main reason is that the ACM-ICPC takes too much time. (After concentrating during the five-hour contest, I would not be productive for weeks). I wanted to focus more on my research topics. However, I still stay as volunteer for this year contests that my university hosts.
How wrong I was. I now think that being a volunteer judge actually takes more time than to compete. This year was not my first time as a judge, as I have been a judge and problem setter at a Pre-ACM contest (organised by SIPA) and a local ACM-ICPC contest last year. This year, however, I was also a judge of the regional contest.
Before the regional, I have also been a judge and problem setter in this year Pre-ACM contest, a local contest (Thailand Central-1A), and the Thailand First Round Online Contest. I was also a problem setter for Thailand First Round Onsite Contest.
And finally, in the ACM-ICPC Asia-Bangkok Regional 2016, I was a judge. My responsibilities include:
- Organising problem statements including format and layout of the problem statements as a whole.
- Coordinate with the system team for testcases and validators.
- Check correctness of the problem statement and supplied solutions.
- Assign balloons according to difficulty.
- Handle clarifications from the contestants
- Handle result announcements.
(That’s probably everything a judge team needs to do already…)
I don’t want to go into detail of what I have done for the contest. I am also not very sure if I could anyway. I will just put up a bullet point of what I think are more important reasons for successful contests:
- Always scrub the problem statements to eliminate all ambiguity. This lessened the load during the contest as there will be very few clarifications.
- PC^2 are very buggy. Be prepared.
- There will always be machine-destroying submissions. I have seen submissions that crash PC^2 almost always. In this contest, we have ubuntu-crashing submissions.
- We always use full Autojudge. I don’t understand why someone wants the autojudge and manual-review. I don’t really see the point of manual review.
- Make sure you have enough machine for judging. We ran 20 machines during the contest just for autojudge, and separate machines for answering clarification.
That is probably everything. Overall, it was a fun experience, but also a very tiring one.