Hello Codeforces!
I've been a member of this community for about 1 year and i'm very glad of how things go there, with a single exception: Educational Rounds.
Ever since they became rated for div2 users, despite of their rather interesting format, they've proved to be a trap for many users, because of hacking period and contest format(extended ICPC). Basically, as a result, in 99% of situations, problem A, which is supposed to be the easiest one, becomes the most important problem, and this is not exactly good, because in many situations, people who have, let's say ABCD and a rather good penalty, end up getting only BCD but with a bad penalty for 3 problems, just because of a single mere error, while in any other contest they would only have gotten a small drop in standings.
Thus, my humble suggestion is to rethink a bit how Educational Rounds(and any ICPC style contest in CF with sorted difficulty problems, like div3 rounds) are working.
One of my ideas is to add weight to each problem, so instead of each problem having the same value (another point on number of problem solved), we can have something like A = 1, B = 2, C = 3 and so on, thus while still advantaging people who did ABCD, it will not hurt people who did something like BCD so much anymore.
Another idea i have thought at is while keeping the current system(people ranked by number of problem solved and penalty), we can add another criteria of separating people's ratings.
For example, a person who solves ABCDE should be better than a person who solves even DEFG, but a person who solves ABCD shouldn't be better than a person who solves, let's say, ABCE
What do you think of those ideas? Can they actually be an useful change for educational rounds and how does it work? Suggest any other ideas in comments and feel free to share your feedback.
I think the rules of the educational round are fine, just like they are fine in ACM ICPC.
In my opinion the problem is that Codeforces mixes two different contest formats for rating. The Educationals should have stayed "Educational" instead of making them also a rated contest. Them being rated (probably) puts way more pressure on people, even discouraging some people from participating. This pretty much kills the point of them being "Educational.
It doesn't affect me personally, but if I were div 2, I'd prefer them being unrated.
Being rated likely discourages less people than it encourages. At least I have noticed that the number of my div2 friends that do educational rounds has increased since they became rated. Often I also hear that the number of rated rounds is too low. Educational rounds being rated greatly increases the density of rated div2 rounds.
Besides, doing an unrated round is the same thing as doing an unrated round as a virtual contest, so if you want to have unrated educational rounds you can just do them as virtual contests.
I think the fix should rather be that problem A should be the kind of problem where solutions that pass pretests are rather guaranteed to work.
I didn't say less people participate, I only said that some people are discouraged, which I still think. I think it makes it less Educational and more like a normal round, but then it could have the rules of a normal round also.
Also rated rounds are pretty often for div 2 people without educationals too. I'd even say, in average it's more than one per a week, which compared to other judges IS frequent.
The number of div 1 contests is low in my opinion too, but having educationals doesn't help that.
As I noticed (maybe it is wrong, but I noticed so) that Educationals and div. 3, but mostly Educationals are contests which have the largest amount sorted in wrong way problems. So, it may be wrong idea to give such values to problems. BTW, I think if authors sort problems in right way, your idea would be great.
no please
stef you may become orange and not to complain about rating on edu. complaining about rating sucks.
I agree with you,I got fst on problem C in Educational Codeforces Round 47 (рейтинговый для Див. 2) and drop in the rankings.
The original intention of the Codeforces format is to avoid such a situation.
And in Educational Rounds,if you got fst on problem A,you will lose more score than fst on probelm F because you solve A early than F.
Interesting. I have an even better idea. Maybe we can make A have score 500, B have score 1000, C have score 1500, and so on. Then problemsetters can put weight based on difficulty. Maybe even decay the score with time in order to break ties!
So this is how the current contest system came into being :O
Hello!
I think that the ICPC style contests work fine:
Supose that someone sloved problem ABF and someone else ABCD.
the guy who sloved ABCD is stable, because he could solve more types of problems (which I think is also
the ideea) when the one who solved ABF probably was just lucky.
In my opinion it is just like participating in contests : It is better to be in top 10 in many contests
in a row than just winning one and the rest let's say place 20 or something.
Secondly, let's say that in one contest you missed the corner case for n = 1, in the second one, you
will have more atention on corner cases I would say. I do not say that I would like my solution to
have a bug on a corner case on problem A but any error that you have will (probably) prevent you from
making similar errors in the future. ( whichm is more important than a single round)
Firstly, in my both ideas, the guy with ABCD will be better than the guy with ABF (1+2+3+4>1+2+6), 4 problems > 3 problems
Secondly, not all corner cases are so obvious, i've seen lots of hacks for reasons like double precision, overflows or hash hacks, which are not that obvious, so it may not be fair for those people to lose a lot more than they should.
Also, ICPC format works very well with full feedback, but here we don't have it, and many unfair things can happen, as i mentioned in blog post