Friday, May 20, 2011

2010's gcj prediction about 2012, how is it going?

In 2010 I wrote some predictions about Google Code Jam 2012. If I use gcj 2011 do the trends seem to agree?

- There will be 5 advancers to on site finals.
- The world champion will win 100 dollars.

Those were just rants about the ever shrinking prizes. They supposedly shrank after the economic crisis but they are not being restored yet, and it is 2011... Everything indicates that 2012 will be as disappointing in these regards as 2011 and 2010.

- Rounds will last for four hours.

Didn't tonight's round last 3 hours? It at least seemed like it. Maybe trying to guess up errors for problem B makes the time feel slower.

- Problems will be 350% more about implementation than they are now.

Still likely. Problem A's values overflew 32 bit integers for no reason. With 2000 test cases, such large numbers were not necessary to prevent time outs. Problem B-small was 100% about making sure to fully find all implementation details hidden under "is consistent with what you have written down so far and the result of all of Sean's previous guesses,"

- Problems will be 10% as interesting as they are now.

A and B seemed to be following the trend towards less interesting problems. It turns out problem B-large was fairly interesting and does not need fancy data structures or premade algorithms but something clever which I have not finished decoding out of the winning submissions. Apparently this trend is getting averted, for now.

- Filtering scoreboard by country will still be impossible.

This limitation will last at least 10 years.

I am not sure what to make about round 1A, it felt lack luster. Most people were only able to submit A-small, A-large and B-small which were not very fun. The other problems were actually interesting but out of reach of the level of most people. I have also detected typos in the statements.

That's not to say that the contest was bad. But I thought that gcj had a larger standard. (Above every other source of contests/tournament). 2008's gcj was one of the best tournaments I can remember. It feels that quality declined a little. It is still better than most tournaments out there, but still... It may have started with the increase in round duration.

I remember gcj 2008 because the problems were all great and distributed correctly among the rounds. Also the huge number of advancers made it clear that google was a very resourceful company. Nowadays it appears that even a medium-sized IT company like topcoder can create more rounds and accommodate travel for more people than Google, it also appears that a Russian Google clone can easily have a contest the same size with equal rewards as the real Google. It sounds like Google does not really care. I know that the Googlers that organized gcj really do care about programming contests, but from their own statements, Code jam was work for their 20% extra work so. It sounds like there is not a lot of real involvement or interest from Google itself.

On the bright side, I didn't advance, that should give me a chance to participate in other rounds, which I hope are better in regards to problem design.

No comments :