### rng_58's blog

By rng_58, history, 4 weeks ago, ,

We are planning to introduce a division system to AtCoder in the near future.

Here are our current plans (but this is tentative, we may change it based on your feedback):

• The division cutoff is 2000. I think this is about the same as CF's 1900.

• We'll hold div2 contests (called ABC) every week. Usually, 100 minutes and 5 tasks: 100, 200, 300, 400-500, 600-700. Rated for 0-2000.

• Sometimes we'll hold div1 contests. There are two types of div1 contest: AGC and ARC. The first two tasks are shared with div2.

• ARC: Usually 100 minutes and 4 tasks: 400-500, 600-700, 800-900, 1000-1200. Rated for 2000-2800.

• AGC: Similar to current AGC. Rated for 2000-inf.

One major problem is that even strong people have to spend a few matches in Div2. Two "red performance" is good enough to reach Div1. Is it fine?

For example, it says that a gray coder solves a 200-point problem with probability 63%. Note that we use "estimated rating" instead of actual rating for this table (i.e., we don't subtract 1200 from newcomer's rating).

•
• +84
•

 » 4 weeks ago, # |   +190 Just wondering, what is the problem with having no lower bound on AGC (and now ARC)?
 » 4 weeks ago, # |   +48 I think the rating lower bound for AGC should be lower.
 » 4 weeks ago, # | ← Rev. 4 →   +9 Hi, rng_58. The idea seems pretty good, but I'd like to share my question and opinion here: Question 1. For ARC and AGC, the rated range is above 2000. Is it able for Div2 people to participates in these contest? Question 2. Is there a contest which is rated for like 0-1199? I think it's good motivation for beginners. Question 3. You said that you may apply this Division Rules in "near future". When (after about how many months) is "near future"? Question 4. In current ARC, the upper bound of performance is 3200. However, the rated range can change to "2000 -" 2800. Are you going to add lower bound of performance (like 1600)? Opinion 1. For current ABC, the rated range is below 1200 and the upper bound of performance is 1600. The rated range of current ARC is below 2800 and the upper bound of performance is 3200. If you apply this rule, the rated range of new ABC is below 2000 and the upper bound of performance is 2400. And the competitors whose rating is below 2000 only can participate in ABC. This means, even for target-level newcomers, they should compete at least 6 ABCs, applying current AtCoder rating formula. I think it's too much. Opinion 2. If the new rule is applied, the contest name Atcoder "Beginner" Contest should be changed. Obviously, rating 1999 is not a "beginner" at all. It's almost same level of "Candidate Master" in CF. I don't think expert == beginner holds.
 » 4 weeks ago, # |   +33 I took part in the recent ARC, it was my first contest, I showed perfomance 2071 and got rating 871. How long should I play ARCs to reach my 2000 rating?
 » 4 weeks ago, # | ← Rev. 8 →   +84 I think the division system in AtCoder now is good enough (AGC rated for all, ARC rated for 2800-, ABC rated for 1200-). Just remain as it is. There should NOT be rating lowerbound for ARC or AGC. No need to set divisions like CF and TC due to AtCoder's unique rating system.It's not necessary to start from ABC. Tasks in ABC are quite typical, and a coder without too much experience, like me, can manage to solve them all easily. Many CF Div.2 contestants, however, are NOT Beginners. Tasks from CF Div.2 Rounds often offer us challenges, but ABC never does. ABC is, as its name suggests, just for Beginners.I really enjoy participating in both ARC and AGC because they both have enough interesting tasks. It's not good to stop unrated people from participating in ARC and AGC, since this will prevent people from finding their place in a few contest. For coders managing to solve all tasks in ABC, it's not hard for them to solve first two tasks in AGC and ARC, I think.Almost all other platforms, like CF, TC, CC, and CSA, have two types of contests (Div.2 Only, Div.1 + Div.2), different from AtCoder. That's one of the reasons why I like AtCoder. AtCoder needn't be the same as other online judges.
 » 4 weeks ago, # |   +10 Increasing the cutoff for the ABC seems like a nice idea, because currently the ABC are not of large use — 0-1199 targets a few people and as far as I know, most of them aren't actually new to CP but just started using AtCoder. So they simply compete once or twice and then ignore the ABCs. Having the it with cutoff of 2000 will surely increase the number of people competing in the ABC (as a lot of people ignore the ABC when it's not paired with a ARC). Having lower bound for ARC also seems like a nice idea, because there will be some competition in changing division. Like on CF. But why do we need lower bound for AGC if it will be like the current one? The first two problems are easy and everyone who will have the contest rated has > 70% solving them (judging from the table you linked). Why do we need them if almost everyone will solve them?So in my opinion the change is a nice idea except for the lower bound of AGC.
 » 4 weeks ago, # |   +61 What's the output of diff proposed_rules current_rules?
 » 4 weeks ago, # | ← Rev. 2 →   +10 Are these Div2 contests held on the same day as Div1 contests?
 » 4 weeks ago, # |   +43 Some motivations for the division system: Actually, even the first tasks of AGCs are not easy for everyone. Let's think about brown coders. If samples are strong enough, they end up without submitting anything in a half of AGCs (and unrated). They only solve the first task in another half of AGCs, and increase ratings. Even for greens, AGCs are mostly about the speed of the first task. I think the contests will be more interesting if everyone usually solve at least 1-2 problems, but judging from your feedback, maybe people like harder tasks? I'm used to TC/CF division systems. I feel we should increase the difficulty of ARCs a bit. Usually the last task of ARC is around 800, but it may be too easy to rate oranges. (And we have lots of unused 900-1000s). But now I'm thinking about changing the cutoff.
 » 4 weeks ago, # |   +110 My opinion is like that saying — "if it ain't broke don't fix it" :)
 » 4 weeks ago, # | ← Rev. 2 →   +43 So, wouldn't it be too difficult for ARC, how do you think? I'm writing almost all AtCoder`s Contests which you hold, and, despite to my 2760, that is almost the high border for ARC, I feel I will not even have significant time for plunge into the last problem, not talking about quality ideas\solutions, because 900-points problem often demands significant efforts for me, and this is during 100-minutes only contest.
 » 4 weeks ago, # |   +18 I am just curious, with the division cutoff, those below 2000 points can only take part in div 2 contest. Therefore, can they even obtain a performance of "red" within the div 2 contest?I personally don't like this system. I have got 4 out of my last 5 contest above 2k, but I am only at 1762. As I don't do contest often, this seems to be quite punishing, and requires me to do a lot more ABC to able to reach the cutoff compared to other site, which is quite disappointing as I personally like AGC problems a lot.
 » 4 weeks ago, # |   +5 Only ABC is rated for 0-2000 Do you want to increase the difficulty of the contests? ABC isn't really interesting for ~1600+, I think.
 » 3 weeks ago, # |   0 There is suggestion..There should be problemset based on tags.i don't do atcoder contest bcoz their question are beyond my level but can practise if ther is problem tags eith increasing difficulty.. :D
 » 3 weeks ago, # | ← Rev. 2 →   -13 Here are my thoughts.Feedback 1: The proposed new ARC looks too difficult for rated participants i.e. rating <2800. ARC: Usually 100 minutes and 4 tasks: 400-500, 600-700, 800-900, 1000-1200. Rated for 2000-2800. The first two problems are more difficult than current ARC. In such a set, most people with rating <2800 won't have enough time to challenge the last problem. If you target <2800, just making the last problem a little more difficult would be sufficient to differentiate upper-orange users.Feedback 2: I'm not sure sharing the last two of ABC and the first two of ARC is a good idea.IMO a good problem as a difficult one in ABC and a good problem for an easy one in ARC are different. - A good difficult problem in ABC doesn't require strong math skill but requires more coding. Many good practical programmers can enjoy the problem. IMO, considering the nature of the participants, ABC should be useful not only as introduction to sports programming, but also as practice to improve practical programming skill. - A good easy problem in ARC doesn't require much coding. It'a warm-up for more challenging problems, it shouldn't eat up much time of participants if they are strong enough to come up with a solution in short time.To realize this, I think it's better to make it flexible which problems to share between ABC and ARC. For example, sharing ABC-CD and ARC-AB and having a unique problem for ABC-E is an option, depending on what problems you have for the contest.