The story: Five days since the last round and waiting for the next one makes me want to do something, and seeing posts a few days ago on "no upcoming contests", I somehow wanted to see if we really are having fewer contests. Below are the findings❗
The data counts for all contests with a div. [#] included in their names and all global rounds.
In specific, I counted their divisions, date, and participant number, and then visualized the data into two types of graphs:
Frequency count of rated contests by divisions, based on their month and year.
Average number of participants for a division in each month of year.
Overall distribution
There seems to be no clear trends of frequencies within a year (e.g., I was expecting there will be more during summer holidays), however, there is a faint decline in contests over the past two years 💔:
Division 2 Distribution
Here are the distribution and average # of participants for div. 2.
A clear pattern is that there are more participants during the beginning & end of the year and the summer holiday, as shown by W-shaped curve for each year. On a greater scale, the average number of participants is also increasing steadily over year, with a huge boost in 2020 (curious why it happened, COVID perhaps?)❗
However, despite the stable amount of division 2 contests from 2020-2023, we have fewer in 2024. There has not being a single month with less than 6 div. 2 rounds from 2021 to 2023, but this year we already have four!
Division 1 Distribution
Here're also the distribution and average # of participants for div. 1.
I haven't participated in any div. 1 contests yet, and therefore can't make clear explanations for the weird trend of participants # which took a boost in 2015 and 2020 but then declined for a while (would love to hear about some possible explanations).
Again, however, the number of div. 1 contests is going down this year. Only 1 rated contest per month for five months this year is too hard to bear. This worries us potential div. 1 participants as well!
Are we having fewer rated contests?
From these graphs, the answer is sadly yes. However, the purpose of this post is not to blame anyone for the trend. I find it enjoyable to explore all the trends and patterns (there are definitely more) and also feel motivated seeing an increasing number of participants (as I'm one of them). However, with these steady increases in the # of participants, I think that we should have more contests, definitely to an extent that is no less than the previous years.
Do you want more rated contests?
Which one would you want to have more the most?
Would love to see comments on this or cool patterns in the graphics
you count them one by one? where can you get contest info
I copied the contest info from the contest page directly, which means 19 copy pastes, and then performed some data processing. (Extremely brute forced)
show the code
What's the word?
please...
Here, although I just uploaded it, haven't written a good readme yet
I think div.1 participant spike and decline in 2020 is pretty obvious. People who had a job and were doing CF but quit due to work, came back and enjoyed great WFH period, but then quit CF again because of their job.
and I personally believe this year is peak (in terms of #of participant) of CF due to obvious reason.
can you tell me what is the obvius reason?
The rating system was changed in 2015. The old system had a problem with rating inflation and "too many" people got to Div. 1. The massive drop is when the new system rolled out.
Thanks for the explanation, that makes a lot more sense to the plot
Considering that div1 participants have the widest range [2100-4000+], it would be more preferrable if we could have more contests.... so that we can update rating just like other participants do.
I agree, it's actually wild to imagine that you can only have one rated contest per month. That means if you ruin it up, you have to wait for an entire new month for the next attempt. Also, watching the LGMs play against each other is fun xd
I honestly think having a system similar to atcoder is preferrable, where ARC are rated upto 2799 or around 2800 in terms of cf ratings.
Widest range is indeed true, however not largest population. It is rather rare to be over 2100. You represent a population of not even 3k. It seems better to work to div2 and to make div1 centered around it (div2 + div1).
Also it is absurd to use the wise and smart tourist's rating to assist your point. He represents only a single person, who is an outlier.
Just because they represent a few amount of people doesn't mean they should get less contests, using this logic IOI participants should get less training than normal National OI participants because they are only 4 people.
The case is different slightly. IOI Participants will represent the entire nation, so their success is more cruicial and essential.
Of course I do not wish that one group of people gets less contests, but of course the Div 2 must be a priority.
I would agree, if Div2s were rated until 2799 similar to AtCoder's ARCs.
It's probably just that div1 contests are very hard to prepare. I can't even imagine how people come up with div1 E-F level problems.
yeah, it's probably out of their hand.
I think it is better when the contests have different starting times like recently... really good news for programmers in other countries.
This case, I don't mind whether these contests are rated or not.
Some variations are good, as they are more friendly to more time zones, but at least half of the contests should still be in the same original time, as a lot of people have already adapted to that time frame.
That's because PinkieRabbit cheated in a recent Luogu contest by using unpublished pre-written code. Problemsetters are scared of him.
no divs 3/4 this period also.
I'd like to be a little optimistic bcz I think the problem setters are creating processes to make new questions be more resilient to AI prompting and Anti cheater
(hopefully we get better queuing system and servers for high demand div 3 & 4s)
atleast I am hoping so