Challenge Problem Score



During the contest my best solution for SEAVEC got 0.085 points and it was in the top 20 solutions for SEAVEC. Now after the rejudge the solution shows [0]points and my contest rank(global) has gone from 26 to 170. I don’t understand how there can be so much difference when the test cases are randomly generated and I didn’t use any random seed in my code. Is it possible that I got lucky on the first two test files and everyone else’s code scored way more than me for the rest 8 files. This is very frustrating and if anyone has got an explanation for this, I’d be grateful.

My best solution during the contest


I was ranked 4(india) with a score of 0.433 initially but now after the rejudging i am ranked 45(india) with a score of 0.002…this is so really demotivating :frowning: plus this rejuding is taking forever and I really feel it’s somewhat buggy because the same codes are performing way better in the practice. Please just take back the idea or have some other method so that we can atleast estimate our performance better. Really annoyed :frowning:


I was ranked 14th during the contest …now i am 126th …I am not getting the reason behind this


Yeah seriously this present idea should be taken back…its really frustating and even if they have to keep it it should be such that the ranks are not affected much mabbe…at max 4-5 …the main motive of a challenge problem is to actually keep improving upon your algorithm to get a better and better result but if such things happen there is no use of keeping the challenge problem only


@sikander_nsit: Yes, this is exactly what happened. Other contestants scored a lot more than you on the remaining 8 files. I intend to write something about it in a comment to the SEAVEC problem editorial. Anyway, your score during the contest was relative to the best score on the 2 test files. After the rejudging, the scores when considering all 10 test files were much higher. Thus, your [0pts] score shown is not really zero, but it is below the precision used for displaying the scores (Codechef displays scores with an accuracy of 0.001 and your score is below that).


@mugurelionut I understand that my score is very low compared to the total best score and that’s why I got 0 points. But the thing is why did so many solutions that performed way worse on the first two test files perform so much better on the rest of the test files. Frankly this is deceiving because when I saw that my solution got a good relative score, I believed that it will also perform good on the total test data and I started trying to solve other problems. I never expected that a solution that got a top 20 score will not even remain in the top 100 scores finally.


@sikander_nsit even though you’re case is every unfortunate that’s why this system was implemented in the first place, to prevent users from guessing the nature of the test cases . I’ve seen my rank drop considerably as well but not as much. Since it’s a new system not everyone is fully aware of it and not everyone is used to its behavior yet. I just think they should at least provide the test case generation method or part of it…


They should consider taking back this idea. It’s really unfair to those who tried hard to obtain better score during the contest. The performance on the two shown files should be an indication of the performance on the hidden files, otherwise we are just shooting in the dark. I was ranked 1 (India) for 2-3 days during the contest and finished the contest at rank 6 but after rejudging, I have been pushed to 26. This is really frustrating.


Now think about my plight. I finished at 8th in India and now I’m over 100.