@admin,@vijju123 rating of cookoff is updated from the ratings which we have got before plagarism check but shouldn’t we actually get it from the new ratings after plagiarism checking.please check into the issue because rank predictor also shows we should get more in Nov long according to present ranklist if in Nov long http://codechef-rating-predictor.7e14.starter-us-west-2.openshiftapps.com/contest/NOV17/all
this user yadnesh_viper has a positive NOV17 rating. Even though his submission was caught by the plagiarism test.
He even posted a thread : here
It was an honest mistake, forgetting to hide your solution while using online ide.
But shouldn’t the ratings be recalculated after plagiarism test ??
@vijju123 ping the admin about this matter.
Nov long rating were to be recalculated.On this recalculated ratings cookoff ratings were to be calculated. So in a way the cookoff ratings calculated presently are not final. So I think there was no use of calculating the ratings for cookoff
Cheaters are to be penalised later , that’s okay but ratings of NOV long will change for all right??
Thanks in advance
It seems that there is no update related to ratings issue. It has been more than a month now and still the ratings have not been recalculated.
Ratings of Lunchtimes and Cook-off’s gets updated over wrong ratings of previous contests which makes all the rating system faulty.
I understand that managing such a site as Codechef is very difficult. May be you are already working to solve the issue, but keeping your user ratings filled with errors for almost 2 months is not alright.
I can say that because there are some users who have not been penalized for July and August Contests also.
Leave penalizing the cheaters, consider those who deserve more ratings but are still waiting for updates.
I also don’t have enough karma points so I had to address this issue here, an old question most relevant to the issue.
How much do you think they will differ by? Can you give some rough number?
This matter is pinged to them. They will penalise the cheaters later. Currently they are working on discuss and some bugs. Cookf-off was updated cause probably it would be huge delay elsewise.
So finally what are coming to say ,will are rating in Nov long be updated and will our cookoff rating be updated from it? Or this is finsl@admin@vijju123
For me the difference in Nov long will be approx 24
rank predictor shows 1781(+101) but I got only 1757(+77)
Everything will be eventually recalculated. I got a word from her.
Hope to see everything justified.
The ratings of non-cheaters wont be recalculated as difference is hardly 1-2. The cheaters will be penalized later.
This means that the ratings predictor is not correct because it shows a difference of 20-30 points from the real ratings received by the users.
These are some of the users which should have a higher rating.
Please can you check whether this is a bug or not because I also think that ratings of some of the users should increase.
And about penalizing the cheaters, hope it happens soon because there are some users which have not been penalized for long challenges since July.
It seems like wrong ratings are not a big issue here.Then why should I put any more efforts??
We don’t know how the Rating Predictor works, and are under no obligation to ensure that the real ratings match with it. Given that entire rating calculation algorithm isn’t even publicly available, it makes no sense for us to go by that prediction.
As for re-calculation after penalizing the cheaters, it is part of the system, and it is evolving and a process under experiment. We do ensure that it is always fair though. So you can forget about how it is exactly implemented.
Finally, yes, the cheaters from the last many months have not been penalized yet because some features to do so are being built. It will be done soon. Rest assured that they will be penalized. We apologize for the delay.
@vijju123 I mostly agree with your views in the forum but I think that the idea “The ratings of non-cheaters wont be recalculated as difference is hardly 1-2” isn’t right. I mean shouldn’t all recalculations should be done ? We are Computer Science people, imagine if some important calculation in our work is off by 1 or 2, won’t it result in a faulty program ? For a funnier version, imagine our answers in codechef submissions being off by 1 or 2. Would it work ? Just my views, but 2 months & 6 contests later, even I don’t support total recalculation, but not for the above mentioned reason.
@dwij28 - At present, since the system of recalculation is still under testing as @admin said, only for major discrepancies can we support recalculation. I understand how valuable ratings are to people, but if for trivial differences we start recalculation then the rating system will slow down to great extent. Which will bring us back to square one.
My viewpoint is- Do other sites do that? Other popular sites like codeforces dont do that- and I heard that some user got unfair rating decrease for some reason and it wasnt rectified since “he was the only one” and it will be lot of work.
I understand if my view isnt the popular one out here, but measure it with the question on scale of efficiency v/s effort and you will at least partially get the situation I am trying to convey.