Hopefully, actually the trouble is that you spend more time on selecting of problems to solve and knowledge to follow nowadays more than actually on problemsolving and things are getting worse almost on every active platform..

In private contests or finals of different worldwide contests mostly, because there is no purpose for anyone to set problems just for lulz..

There was dozens techniques known by people before Chinese NOI, typo:

• Sort on subsegment an array A[L..R] and find a sum; Seems that was a problem with implicit pre-requisites to Segment Tree Beats

• IOI (Aliens trick is promoting so often);

• Moscow Pre-Finals by MSU Tapirs (where tourist solved not so many problems, you can find out that there is a problem about Bertrand's postulate (!) that in combination with Ramanujan's and Eratosphene's heritage may provide you using basis of prime numbers you can notice that between [2^i and 2^(i+1)] there is prime numbers so maybe you would try to prove Goldbach's conjecture if use p-adic properties of mod 2^j (maybe other bases not only 2) or some inversion tricks, etc);

• rng_58's, jqdai0815's, antontrygubO_o's or Errichto's problems;