Direct i has been providing excellent services and codechef is now undoubtedly one of the topmost platforms for sport coding competitions and good problem sets with people like Tourist, ACrush and Anton contributing to the problem sets. Although as an ardent user , I feel that some easy changes might be helpful for the betterment of the platform .For this , a lot of things Codeforces does , a lot of things hackerearth does and a lot of things ACM contests do might add to the benefits of this system.
- HackerEarth has been very innovative in setting up the problems. There are problems to design a game bot and design an AI based environment. What I think that problems like coding the bot firstly expels a lot more cheating because all the solutions are tested against each other in the end and secondly people get to use a lot of alternatives and data structures to improve their code . They are like partial problems of long contests but are more real life ones.
- The fact that problems for contests are created to be as unique as possible , its very much possible that psetters and testers leave some crucial test cases as what happened with TkConvex problem. So the best possible thing is introduce the hacking system for 1 day after the coding phase is over and award those people suitable points who can present test cases which wrong an AC solution .This way at the end,during system testing,we would have even the cornerest of test cases.
- June’s contest involved a lot of mathematical problems . Predict was mathematics , Collect was mathematics apart from implementing basic seg-tree , SpMatrix was mathematics ,permute was mathematics . 4 problems out of 8 problems(not including the easiest one and the partial one) being of pure mathematics made the competition too monotonous . 70% People who did SpMatrix googled for series and I don’t believe that all of them know the logic of that problem which was very very good. I feel that there should be mathematics but not more than two questions. April’s contest was very good in contrast where that editor problem was one of the best real life problems I could have learnt . More hard DP, max-min flow, non-deterministic problems should be there according to me .
- People should stop worrying about cheaters. Those who cheat do bad for themselves and good for us.
- I learnt about the Moss system http://theory.stanford.edu/~aiken/moss/ which is widely and actively used and updated . That could be well the answer for plagiarism detection.Apart from this , this particular discussion on SO has suggested very good tools http://stackoverflow.com/questions/546487/tools-to-identify-code-duplications