During the last few months most of the contests where Div.2-only, and, logically, there must a bigger transition from Div.2 to Div.1, rather than from Div.1 to Div.2. Now we will try to figure out if this is true

Generally, we need to have data about ratings before pleiad of Div.2 contests, after them, and after Div.1 + Div.2 contest. But I became interested in such a comparison too late, that's why I have only the second and third data, which I'm going to compare

On the first 2 plots I've compared ratings of active users before (img. 1) and after (img. 2) contest pair Div.1 + Div.2 took place. You can see a significant leap around 1900 point (margin between Div.1 and Div.2), which is slightly smoothed on the second plot

Making graphics for all users, and not only for active ones, helps us to simply find the old margin between Div.1 and Div.2 (it was 1700). Img.3 is before, and img.4 is after div.1 + div.2 contest

We can suggest, that there is a lot of people, who drops competing after they become Div.1, considering that this is a good achievement, but 2х leap can't be explained only by this factor. Removing users who competed only 1-2 times doesn't change the leap significantly. Does moving of the strong coders from Div.2 to Div.1 makes Div.2 competition weaker and Div.1 one stronger? You judge.

Rating subject is a quite popular one on the CodeForces, and the last few discussions I saw where about — if you can't go up to Div.1 after a year of competing here, you need to drop this occupation.

In a year you can spend just about 10 hours on trainings, or even more than 800 hours, so measuring your results in this way is not a deal. That's why I want to suggest another shitty method of measuring the results — statistical expectation of your rating basing on the number of rated contests you competed. Calculations where made for Active users (last 6 months) and all the users ever competed here.

Table number one (blue line — contests, numbers — expected rating):

You can mention, that active users have a higher value of expected rating. This can be caused by rating inflation, or by the fact, that active users have better results, and those who had worse ones just dropped this activity

Table number two (rating and number of contests, necessary to achieve it):

Somewhere near 2000 rating points there is a tendency change, where number of contests is not playing a significant role anymore. I can confirm this from my own experience — it's about rating you can gain, mostly knowing nothing about algorithms, without regular solving problems archives — just competing and competing

Table number three (rating by number of contests):