Hi,
I happy that I see many contests in the CF calendar. Div $$$2$$$ rounds are good training for me, as I need some time for solving last two tasks in such contests (many times I am not able to solve all tasks and need editorial for some).
But, I really do not see reason why we have so small amount of $$$D1$$$ rounds? I am pretty sure the tasks from two $$$D2$$$ only rounds can be merged to get one combined round in many cases. Remaining easier tasks can be used for some $$$D3$$$ or $$$D4$$$ contests. So I am pretty sure that from current $$$4$$$ $$$D2$$$ rounds we can get at least $$$3$$$ rounds ($$$D1$$$ + $$$D2$$$ + $$$D3$$$).
Again this is not complain, as I am happy when someone gives me opportunity to solve new problems. Only I think that CF should start with merging tasks from different authors and allow submiting only one task (support paying by task not by contest). Many coders have $$$1-2$$$ cool ideas and they are not able to set whole round. Normally if someone has all tasks for $$$D1$$$ round there is no reason for merging. if someone has only $$$D2$$$ or $$$D3$$$ problemset and he won't separate tasks, it is ok too, but such setters should have lower priority for their rounds.
I believe all this stuffs will increase quality of CodeForces rounds and make better rating distribution.
I am interested in your opinion and maybe see reasons why this system is not supported by CF currently (maybe I am missing some obvious reason).
How did you miss the fact that div1 problems are more difficult? You won't get a div1 round by merging even ten div2 rounds.
div1-e problems are a bottleneck and indeed it would help to allow proposing a single hard problem.
The main point of my blog is the second part with merging tasks from different authors and allwoing partial proposals.
You probably think about first part and my suggestion for this $$$4$$$ consecutive rounds. D1 — E is bottleneck I agree, but I would go with count of "harder" tasks. Look only at last tasks for previous 3 $$$D2$$$ only contests (round 649, educational round 89 and round 648). In case you put that tasks for the last three in combined round wtih $$$7$$$ tasks and $$$2$$$ hours, I am not sure about amount of full scores at the end of round ( I am pretty sure it would ne lower than $$$10$$$).
From one side it looks more like "speedforces", but from another side I do not see anything wrong to have scoring like $$$500 - 1000 - \dots - 3000 - 3000 - 3000$$$. It is enough challenging for top guys in two hours, masters and lower reds would solve at most $$$2$$$ of last $$$3$$$ tasks. I think the accepted rate per problem will be more balanced than current one with usally big gap in the middle of problemset.
I propose to merge 10 Div3 rounds into one Div1 round, with 20 problems for 2 hours. I am not sure about amount of full scores at the end of round.
Well, in last combined round you had $$$8$$$ tasks, and I am talking about $$$7$$$ (so I do not see anything wierd in amount of problems). Second thing, what is the purpose of $$$H2$$$ when there is $$$0$$$ AC, even $$$0$$$ real tries?
I didn't read $$$H$$$ and I have any opinion about it. But from my point of view, in case you expect $$$0$$$ AC rate at some task, you should not give it at the contest — you could write blog with title "Nice task".
Anyway, the main thing is the second part of blog.
H2 was actually just a natural extension of H1, as it turns out that the dp in H1 can be nicely merged using segment tree, so I don't see a problem in introducing it as a separate subtask to the contest.
If my calculations are correct, and based on the fact that Um_nik is worth around 1000 cyans, merging 1000 div3 rounds should get you a Div1.
I don't know why are you lowering my cost... I have never said anything about "around" 1000 cyans, I just said more.
so cringe..
Um_nik's comment:
You can't read again?
And I'm not saying that I'm important like Mike. I'm more important than 1000 cyans.
ik it's cringe idk what I was thinking when I wrote this
Um_nik said that he is MORE IMPORTANT to CF than 1000 cyans. He didn't compare people, saying he is better than someone. There is no known common comparator for people.
Everyone is special in their own way and you can't judge that by how helpful you are on a Programming Website.
Judge that? Um_nik only judged one's importance to CF.
I propose that we merge the last 15 tasks into one single task. It will have as many solves as a legitimate Div1F because no self-respecting Div1 participant wants to code 2000 lines.
Dont worry , i get the idea ....
As far as errichto , he's just pissed that he only got 1 submission in his last div 1 round , thats why he is trying too hard to make div 1 look too difficult ....
xD I would have upvoted it if you had guts to use your real account. So instead I downvoted it. Kid :)
While I mostly agree with you, I'm confused on your point about proposing hard problems.
If you really could set a Div1E difficulty problem, is it likely that you will have problem thinking of ideas for easier problems? I find it more likely that a problem proposal system will find itself swamped by (probably poorly thought out) Div2A ideas.
I would guess that some people can come up with Div1E problems but are too lazy to prepare the whole problem set, so allowing the person to propose a single problem would be helpful.
Regarding your concern about people suggesting mostly bad Div2A problems, simply allowing only high rated user to propose single problems would cover that I think.
It is indeed allowed to propose a single hard problem. If you really had a good Div.1 E problem but don't have easier problems, I'm sure you can find any coordinator and make a Div 1 round with some other authors. I have done it once when anton asked around for Div 1 F problems for Ozon Tech Challenge and I'm sure hard problems are in very high demand right now (to the extent where I think if you have one you can create an entire round from it, possibly with problems from others if you don't want to do it alone).
You should be able to propose a single div1-d/e problem, not something easier.
Oh, I see now that your proposal involves limiting the range of problems that can be proposed. Thanks for clarifying!
I risk another 30 dislikes :)
It is possible that someone underestimate problemset and one harder task should be changed with easier (such situations are rare, but each task is valuable).
If you proposing problem you should choose expected level of task too. Again someone can choose E div 1 for task of level B div 2, but I think CP community is pretty fair.
"CF should start with merging tasks from different authors and allow submiting only one task"
Strictly agree!! Some cyans and blues can came up with beatiful D2C-D problems, but they can't organize full problemmset
In my opinion, quality of Div1 ABC in Div1 rounds should be better than in Div2. For example, in Div2 we can use something educational-ish, but it is not good to use such tasks in Div1. That's why it is easier to set up Div2 contests, and even with last hard problems sometimes it is impossible to create Div1 from Div2.
While I agree with others, that merging problemsets that are supposed to be div2 is not a good idea for div1, I think possibility to propose a single problem would really help. Also I don't know if it's still true, but I remember that some time ago some coordinator told that they don't have any problem with lack of hard tasks. They'd rather need more easy tasks which are hard to prepare to make them be nontrivial and suitable for low-rated users. There is also question about contests that can be in queue even for months. Some time ago there was discussion where a lot of people confirmed that they proposed div1 contest and after a long time it's still in the queue. In that case I don't understand why we have so few such contests nowadays, especially seeing that there is some div2 contest conducted every ~3 days. Maybe it would be better to decrease frequency of such contests and give coordinators more time to focus on div1? I can understand that conducting more contests for low-rated users can attract more people to do cp but it really shouldn't affect div1 contests frequency.
I'm against the whole idea of someone constructing the contest from separate problems. I understand that sometimes we have to do this because of time pressure (we have set the date for the contest before we even started working on problems) but to do this for regular rounds sounds stupid. Take your time and come up with whole round, what's the problem? Find friends and collaborate with them so that you can make a whole contest as you want. Discuss tasks with each other, estimate relative difficulty, think how well they look together in one contest.