Recently, a blogpost was posted suggesting to stop or decrease division 4 contests, as a way to increase the amount of contests of higher divison. I fully disagree with this idea: Div 4s are amongst the most popular contests on codeforces, and more traffic means more people get exposed to competitive programming, which is great. Also, Div 4s are very easy to set, so they do not affect the frequency of other contests.
However, I do believe that the lack of higher division contests, mainly Div 1s, is a problem. This scarcity has a very clear explanation — problems E and F. Both problems are:
Very hard to create — I've heard stories about authors who spent months thinking of a good E/F, or simply make it a Div 2.
Not relevant to the vast majority of Div 1 coders.
Think about it — in a regular Div 1 contest, low masters, which might've just been promoted from Div 2, have to face far more skilled coders, such as LGMS, within the same problemset.
My suggestion is simple: Div 1.5s, where the problemset is of levels Div 1. A, B, C, C, D, D, instead of the regular Div 1. A, B, C, D, E, F. ABCD is challenging enough for $$$<= 2600$$$, so adding two more problems would be even harder. We can see from the frequency of Div 2s that CD are not hard to come up with.
In conclusion, this could be a complete gamechanger, and can cause a massive rise in contests for higher rated participants.
I think a lot of div 2s are actually like the div 1.5 you described already
eg. if you look at all problems above 2600 half of them will be from a div 2
I am confused by the purpose of including those questions in div 2 rounds. It seems that only 1/1000 real div 2 participants can solve those, the rest of the solves are just alts or red coders.
also, most div 1 begin below 1500, and div 2 will have two 800 rated problems at the beginning even though div 3 and 4 also start at 800
Or maybe just increase the rated range of div2 to 2400/2300 and add one more div2 E/F in the problem set.
For many masters (and above), Div2 A-D are usually simple, and sometimes even E. One more difficult problem might lead to speed solving contests.
Alternative idea (it was downvoted)
Doesn't that increase div 1 scarcity?
It doesn't really ig? You don't use div1EF, and those are the only ones stopping it from become a full div1
I don't think this is a good idea. In my opinion, there is a thin line between the contest you described above and a div 1 round -- you need a slightly harder D than usual, and you got yourself a decent div 1 round.
Maybe it is worth increasing the current bound for div 2 rounds. I checked that there were almost no masters who solved all problems in the last few contests.
As a weak master, I support increasing div2 range since I never finished a div2 on this account.
so you got another account that can finish div2
That's a great idea,as I never gain rating in Div 1s.
But how will Div 1.5s be held?
With Div 2 and Relpace Div1(LGMs will speedcode)
or Div 1.5 only(That will kill specialists and below)
or Div 2+ Div 1.5 +Div 1(too little people at Div1/1.5)?
I think Codeforces Round #XXX(Div 3+Div 1.5)might be great,But many 1600s still need to have Prob A-B to have different ranks,and solving only A as 1600 will also be frustrating with rating drops.
I have another suggestion-Dynamic Ranks
last 20% of Div1's rating will be calculated with first 20% of Div2 (with an Average Performance of A,B added on Problem A and B,as penaltys for Div1 of ranking low) and they won't affect Div2 20%'s rating change
First 80% of Div1 will calc with last 20% of Div1(but won't affect last 20%'s rating) and also try to keep the rating change sum to 0
Div2 calculated as usual.
Some close system is in AtCoder. There are ARC, which are a little simpler, that Codeforces' Div 1, have upper rating bound, and are holded often, and there are AGC, which are like Div 1 without first problem, and are holded rarely.