Then, I remembered (not sure why I didn't recall this earlier 😂) that I had already taken part in a live contest where we were asked almost the same question. It was problem C of a contest and I wasn't able to solve it that time, but with over 100 contests I've attempted, it wouldn’t be easy to find that problem again. I also tried the semantic search by TLE (look here if you're not aware), but it couldn't find the problem.

I remembered that a friend also participated in that live contest and solved that problem at the time. Since he rarely participates in contests, I checked his account to look for the latest contests he had been in. Amazingly, the first contest I picked on his graph was indeed the contest in question. (I picked that contest randomly on the graph.) In fact, the two problems were similar. Look at 1903C - Theofanis' Nightmare.

If you read both problems, you will see that they are almost the same; the difference is minimal. The older problem had a rating of 1900, while the latter only 1400. This indeed proves the fact that past problem ratings are not always accurate. (I guess it is because we have a huge number of competitive programming participants now, which helps ensure more accurate data, and also, it was placed as a D in the latter, which means some didn’t have time to reach it because they were stuck on C.)

Something similar happened to me also in the last Codeforces Round 952 (Div. 4) , where I couldn't solve 1985G - D-Function even though I realized during the contest that something similar appeared in Codeforces Round 913 (Div. 3), in 1907E - Good Triples. I haven't had much luck that time 😂