Hello Codeforces!
Lately, I’ve been looking closely at the difficulty distribution in recent Division 4 rounds, and something feels fundamentally off.
When MikeMirzayanov originally introduced Div. 4, the core philosophy was clear: create an accessible, welcoming environment for absolute beginners and lower-rated participants to experience competitive programming without hitting an immediate, insurmountable wall. It was meant for people who are still mastering basic concepts.
But let’s look at the reality of recent contests. We are routinely seeing Problem G (and sometimes F) hitting 1700–1800+ difficulty ratings.
Is this really a natural "evolution" of the platform, or have we completely lost the original purpose of the division?
The Two Sides of the Coin
Like any controversial topic, there are two main perspectives here:
- The "Top Separator" Argument: Supporters argue that the last problem must be difficult. It acts as a hard separator for the top of the leaderboard and challenges those who are on the verge of outgrowing Div. 4.
- The "Beginner Shield" Argument: Div. 4 was created strictly for newcomers. The entire problemset should remain within their theoretical reach.
Personally, I strongly lean toward the second opinion.
Put Yourself in a Newbie's Shoes
Imagine being a true beginner. You are just learning how to handle arrays or basic loops. Then you open the latter half of a Div. 4 round and face a task that requires complex data structures or advanced dynamic programming to process thousands of elements, rather than a simple single-pass iteration. You physically and theoretically do not possess the tools to solve it yet.
If you were that newcomer, would you honestly feel motivated to participate in a contest where the final goals feel completely alien?
Right now, Div. 4 has become the most controversial contest format on the platform. Even though it is explicitly labeled for beginners, the back-end of the problemset is not adapted for them at all. It feels like inviting someone to a friendly introductory run, only to put a marathon-level wall at the very end.
Conclusion: It's Time to Rethink the Concept
We shouldn't just accept 1800-rated problems in a beginner round as a given. If a problem requires an Expert-level skillset, it belongs in Div. 3 or Div. 2, not here.
We need to modify the entire concept of Division 4 to keep it a genuine gateway for the community, rather than a format that intimidates newcomers at the finish line.
What are your thoughts on this? Should Div. 4 strictly cap its difficulty to protect the beginner experience, or is having an 1800-rated "boss fight" at the end healthy for growth?
Let’s discuss in the comments below!
I fully support this!
In my opinion, Div. 4 should be a challenge primarily for Newbies and Pupils. In reality, most of them struggle at Problem F and don't even attempt G.
A difficulty cap around 1500-1600 would be much more reasonable. It would still offer a solid challenge for Specialists, without completely alienating the beginners.
Auto comment: topic has been translated by Arbuzik342 (original revision, translated revision, compare)
that is not just in div 4!
you can see some problems in div 2 which has got 2300-2700, I do not say I do not agree with these problems' rating tag, but still I am confused.
Nah, Codeforces is supposed to be a difficult platform. You arent supposed to AK the whole contest. If there are 8 problems i think having the first 5 under 1400 difficulty is enough, and the last 3 can be between 1500-2000.
Look at the difficulty caps for div 2, where the last problem routinely reaches 2700, 600 pts above max rating. Should we change div 2s as well then?
However, I think that even if the last problem is hard it can use less advanced data structures to still be accessible to beginners (which is what current rounds do anyway).
True, but Div. 4 was specifically created to be easy and AK-able.
As an expert I don't want to load up a recent contest only for every single problem to be excruciatingly trivial. These hard div4 problems give an extra challenge for participants and higher rated people to solve in or out of contest, and are usually quite accessable. Being able to solve problems above your rating or even division is very exciting and taking that away in favor of handholding newbies more than div4 already does is ridiculous. If this same idea was applied to div2 I would've AKed the last one in 50 minutes, which is obviously unsound for slightly less than CM performance.
Contests aren't supposed to be easy. I feel like div3s and div4s already suffer a lot from that, and getting rid of the most interesting problems only makes that worse.
I think G 1800 is normal, most newbies struggle on other problems, so making G problem won't change a lot, it would change for specialists tho, harder problems are for specialist/good pupil, with harder problems, they separate better from worse people
Almost all the div four have like 10-50 people who full solve and imo this feels like a good range such that for people on the upper end of the rating limit a full solve is possible but difficult. An argument could be made however for the types of tasks that should be at the end of a div four. imo interactive tasks probably wouldn't be great for div fours nor tasks that require very advanced knowledge of some niche data structure. Also as for newbies, I don't think the entire problemset should be catered to them. I've seen div fours where 5 give or take problems were in 800-1200 rating range and having 5 appropriate problems is in my opinion enough for a contest. Also in many contests, the last two problems are very difficult and not realistic for 95% (and sometimes more) of participants to solve and I don't think this is a bad thing and keeps the contest from being speedforces. Also the last (rated) div 4 I participated in, I believe that although I did not solve the last question, I made a fair amount of progress on it and it did not feel as though a task completely out of reach.
Today is the $$$4$$$th of October, which makes it the $$$6$$$ month anniversary of the last Div. 4 contest. Of all the issues in cf rounds, (cheating, queueforces) Div. 4 suffers the most in almost all of them. Imagine having to wait $$$40$$$ minutes in queue just to get your submission judged. This is a regular scenario in Div. 4. And not to mention, the mass cheating.
I really enjoyed Div. 4 rounds, but I don't think we're getting another Div. 4 soon.
has this spread to div3 too? 8 problems with a 2200-2400 H is for Div1 participants to relax rather providing meaningful improvement for Div3 users???????
but its fun for me so i guess
Since 1400+ cannot be rated participants of Div. 4 contests, the 1800 range seems like an acceptable stretch for the last problem to me, especially since problem difficulty may not match expectations during testing. On the other hand, the most recent Div. 3 (Codeforces Round 1122 (Div. 3)) had FGH difficulties of 2000, 2200, 2400, which seemed a very difficult curve for participants capped at 1600 rating.
However, this concern may not be so relevant, as there have only been 3 Div. 4 contests since October 2025. I'm not sure if the issue is server overload, lack of interested problem setters, or intentional focus by this platform, but I think it's potentially a huge missed opportunity to attract newcomers to the community. By comparison, AtCoder has the weekly ABC contests for beginners, and although the problems are not as novel, the consistency is a good experience (and arguably newcomers don't need as much novelty, as the concepts are new anyway).
The last problem of a div2 can be 3000, the last problem of a div3 can be 2500... The point of div4 is giving beginners an accessible competitive programming experience, and regular AKs in your division are actually not the standard CF experience.
Maybe the key is making G feel like an optional stretch goal, not a wall. A hard final problem can be motivating if beginners can still have a satisfying round before they reach it.
Problem ratings are calculated by rated participants' performances on the round. The possibly "inflation" of the problem rating comes from the fact that most rated div4 participants cannot solve A-F in short enough to get them enough time for G. And also, div4 does not mean every participant should be able to AK very quickly.