Please remove the partial grading system in long contests.

long-contest

#4

Due to partial grading, the first 5 or 6 problems can be solved by Bruteforce only.So many people are getting more number of ACCEPTED’s.PLZ CHANGE THE FORMAT.


#5

I like the partial grading and I believe I’m not the only one…

There are few side-effect (I won’t call it bugs), for example I do not want to have partially solved problem in my profile as solved - new category like partially solved would be great. Similar problem with marking solved problems on contest page.

On the other hand I believe, while CodeChef’s long term goal is educating, this is beneficial. Good example can be Fombinatorial, for 9 points it’s good for a beginner, 40 points were challenging for me too and 100 is good evaluation of your effort.

Yes, someone can disagree, that 10 point in one problem is not as easy to get as 10 points in another, so it is not balanced, but here I can say just one thing - the environment is the same for all contestants and is you choice which problem you want/are trying to solve…

From a long term point of view I think it is also better for competition, before it was so, that a lof of contestants are in one bucket, let say with 6 points, now there are more buckets…


#6

Completely Agree !!!
it completely ruins the thrill and interest of a long contest …

(also misguides)
for eg.
Fombinatorial(100 points) is harder to solve than Chef and Words (100 points) , but everyone gets a green colour on Fombinatorial getting 9 points which is not worthy of a solution …


#7

Partial grading system was good…but it had it’s own flaws like i could solve 4 subtasks with one method sol1
and 2 subtasks with other approach sol2 .
What i did was guessed what would the length of string in the last 2 cases and i combined both the solutions and voila AC final soln
Once i got AC,i didn’t even bother to check back the mistakes i made in either approaches.

Plus point :Due to this grading system, I attempted questions which i would have left.

So in my opinion, there should be 2-3 problems with partial grading,and the rest should be same as earlier :slight_smile:


#8

I think that partial grading is a good move, but IMHO, there is an issue introduced. For instance, there is no clear objective about how difficult particular problem is. The excellent examplas are Fombinatorial and FCNS problems. If the grading system was like for the previous contests, then these problems would have very low percentage of AC. That percentage was a good indicator about the difficulty of each problem and I used to solve problems according to their difficulty. However, in this contest, solution Fombinatorial has a very high percentage of AC due to the fact that there were a lot of partially accepted solutions. To mitigate this issue, you should introduce a list of acceptance rate of fully accepted solution. In this manner, the contestants would be still able to solve solutions partially and learn more. On the other side, by having the list of acceptance rate of fully accepted solutions, the contestants would be able to get a good sense about problems difficulties.


#9

I think it’s a very good initiative taken by codechef team by introducing partial grading system. I myself felt very motivated to get full points in the FNCS by trying so many times. By this i was able to know that my approach was right, giving right answer but was not the optimized one. So it helps the beginners and motivates them to try more and more so they can earn full points for the problem. This time i tried to attempt and solve the question that i would have left by seeing its acceptance rate in the old system. So i think it’s good for the beginners. There are some drawbacks of the system like we are now not able to judge the problem complexity by seeing its accept rate and some others related to ranking. This may take time , we should give chance to codechef team to work on it. I’m sure that they will definitely come up with some solution.


#10

yes I completely back the idea that a partially solved question must not appear in the solved section of profile of a codechef user.
Moreover, I will suggest that partial marking should me removed when contest’s problems are introduced in practice arena,I think when someone is practising ,he must solve fully.


#11

What you could have tried is to divide the tasks into subtasks and show subtask-specific information upon each submit. But, one would get score (1 point per problem) for a problem, only if all subtasks pass. (Hard-earned points!!!)

For folks who want to try harder problems, they can still get ideas on how and where they should optimize, but, unless and until they reach the final destination, no points.

Thoughts?

Disclaimer: Personal opinion.


#12

It’s a long contest. You have ample time to judge and correct your approach to come up with the perfect solution in 10 days.


#13

But the situation was the same in previous contests - you had just 1 point and it was not important how hard the problem was…


#14

I still cannot see the problem? Ok, they have some additional points, and? Someone solving the problem for 100 points has always higher rank.


#15

I will tell you my IOI experience. When I started programming and came to know that IOI had partial grading, I often used to focus on getting partial points rather than going for full solutions. Basic human instinct - maximisation of points. A beginner who managed to score 40 on Fombi would not make a real effort to move to 100. This happens; I know it pretty well. If there has to be a change, I would suggest that we can have different points depending on difficulty. Hard problems can be worth 3 points and easy worth 1. Partial grading in a 10 day contest is absolutely ridiculous.


#16

I like the way you put it @dextrous. It actually ruins the thrill of managing to do the questions which <100 people have done.


#17

1 point for each problem was scrutinized but partial grading is even worse …
when one see’s that only one test case is wrong then he/she go for corner cases straightforward … previously it made you think more about the problem with your solution … it widens your way of thinking (sometimes corner case check and sometimes we have to completely change the approach)… partial scoring hints whether approach is wrong or only corner cases are left .!!!


#18

That’s why I counted the statistics offline. You need more?


#19

Who marked this as “community wiki” and why? This option is, according to my understanding, for editorials…


#20

you are correct on your point … but these are just my personal views on the format …


#21

Here I think the situation is different in long contest - you have a lot of time to implement first easier approach and later the more advanced one, while you have a lot of time to do so… Motivation is personal thing, for you it’s a challenge maybe to get 20 WAs, for others it might be rewarding to have at least some points…


#22

too much of hassle. Its better to remove it.


#23

This “hacking” is your personal choice, it’s not contest format problem…