EJIC_B_KEDAX's blog

Codeforces Round sponsored by NEAR (Div. 1 + Div. 2)
By EJIC_B_KEDAX, 8 hours ago, In English

Hello, Codeforces!

I am pleased to invite you to participate in Codeforces Round sponsored by NEAR (Div. 1 + Div. 2), which will be held at Jul/18/2024 17:35 (Moscow time). You will be given 9 problems and 2 hours and 30 minutes to solve them. It will be rated for all participants.

The theme of the round is computer games!

Tasks for the round were prepared by EJIC_B_KEDAX, zwezdinv, green_gold_dog, molney, azureglow and Sokol080808.

We would like to thank everyone who assisted in the preparation of this round:

  1. Our coordinator 74TrAkToR for his useful advices and help in preparing the tasks!

  2. Qwerty1232 for red interest in testing the round.

  3. turmax for the red-black testing.

  4. makrav, arbuzick, Anonymous_Noob, Hyperbolic for red testing.

  5. [user:ivan. alexeev], alenenok, Kihihihi, azureglow, pakhomovee, plagues, IzhtskiyTimofey, FelixArg, XaRDKoDblCH, meowcneil, AndreyPavlov, djm03178, MylnikovNikolay for yellow testing.

  6. Mr_Ell, krigare, marzipan, TimVen74, coder8080, MichsSS, PieArmy, tvladm, Vladosiya, bashem for purple testing.

  7. Algolagon, PoDuReMaN, zarubin, IgorA, leoper, kovir_aleksey_korroy, Kosya, Vamperox for blue testing.

  8. MakGeoKar, viteli for cyan testing.

  9. Sonya_2009, ishaandas1 for green testing.

  10. MikeMirzayanov for the excellent Codeforces and Polygon systems.

I also congratulate green_gold_dog on his birthday and wish him good luck on IOI.

This round is supported by NEAR!

NEAR was founded in 2017 by Illia Polosukhin, one of the creators of Transformers, and Alex Skidanov as an attempt to build an artificial system capable of solving competitive programming problems. You can read more about that attempt here.

Ultimately, NEAR pivoted into building a blockchain protocol, which it launched in 2020.

This year, NEAR started NEAR.AI, a new lab with a mission to build AI systems that are open and available to everyone, instead of being controlled by a few mega-corporations.

One of the areas of focus is making models capable of reasoning reliably, and for that, competitive programming problems provide a great environment. To help us build it, we invite all members of the Codeforces community with a rating of Specialist or higher (1400+) to help us annotate step-by-step explanations of solutions to competitive programming problems. We want to annotate a very large set of problems of all difficulty levels, and pay relatively high rewards in NEAR per annotation. You can join the system following this: https://codeforces.com/settings/near-connect-account

Thanks to NEAR for providing the following prizes to the contestants:

  • 1st place: 512 Ⓝ;
  • 2nd and 3rd places: 256 Ⓝ each;
  • 4th to 7th places: 128 Ⓝ each;
  • and so on;
  • 512th to 1023th places: 1 Ⓝ each.

Good luck!

  • Vote: I like it
  • +76
  • Vote: I do not like it

»
3 hours ago, # |
  Vote: I like it 0 Vote: I do not like it

I hope to become expert after this contest ^_^

Good luck everyone and have fun <3

→ Reply
»
3 hours ago, # |
  Vote: I like it -18 Vote: I do not like it

Seeing 74TrAkToR as coordinator is quite scary, change my mind.

→ Reply
»
2 hours ago, # |
  Vote: I like it 0 Vote: I do not like it

praying for 1434 rating

→ Reply
»
76 minutes ago, # |
  Vote: I like it 0 Vote: I do not like it

Finally not 3h Div1 but 9 tasks in 150min, I hope the round is not too speedrun.

→ Reply
»
17 minutes ago, # |
  Vote: I like it +5 Vote: I do not like it

Hello, I am writing this comment, possibly hoping for officials at NEAR to see.

Recently people on Baekjoon Online Judge (BOJ) have seen many DMs on Codeforces, apparently related to the payout for NEAR.AI. They have been asking for solutions (assumably they annotated the tasks themselves but did not have solutions) to many tasks on BOJ. I am gently asking you to remove the BOJ platform from the automated payouts for the AI. Here are a few reasons why.

The BOJ Rules state this is bannable offense.

Through investigation, the users at BOJ had found out that the BOJ account sahera5474 is related to verifying the solution for the payout (though, we did not know exactly what kind of service it was at that time). This kind of automation, not only is condemned by the users, but is also bannable offense according to the rules.

The following is part of what is stated on the rules of the platform (link):

...

Cheating: Cheating states of the behaviour of submitting code that one did not write.

The code that is considered cheating is removed from the site, and according sanctions follow the following rules.

  • User A submits copied code on task B: User A's submissions on task B are removed entirely
  • User A commits repeated cheating: All of User A's submissions are removed entirely

The user reported for cheating is penalized accordingly.

  • First case: 7 day submission ban
  • Second case: 31 day submission ban
  • Third case: 2,147,483,647 second submission ban

...

Clearly, this kind of behaviour is considered cheating on the platform, and is bannable offense.

Most tasks on Baekjoon are not worth any payout for solutions, as they have the solutions publicly available.

For whoever does not know how the Baekjoon platform is maintained, the website contains the following kinds of tasks.

  1. Tasks from user contests/local contests (mostly by Korean people)
  2. Tasks from any other contest apart from the platform (OI, ICPC, foreign local contest, etc)
  3. Personal task contribution (also mostly by Korean people)

For case 1, most such contests contain official editorials, so the solutions are publicly available. For case 2 it is almost the same as case 1, but you will likely have to search outside the platform for the publicly available solution. Most cases where the solutions are not publicly available belong to case 3, but recently case 3 appears much less than case 1 or 2 in quantity. Therefore, an automated payout for training the AI on tasks of case 1 or 2 would make no sense.

Conclusion

Even if one suggests the two points above might not hold for this specific case, it is common sense to consider the DMs from the payout assignees to be considered as fraudulent use of both the Baekjoon platform and your payouts. Therefore, I am publicly asking to remove the BOJ platform from automated payouts and possibly the training data.

→ Reply
»
9 minutes ago, # |
  Vote: I like it 0 Vote: I do not like it

As a tester, I want comments again.

→ Reply