Is it time to increase the blocksize cap?

Given the impossibility to quantify an optimal blockspace demand, are you aware of proposals suggesting a dynamic blocksize that would work similar to the block difficulty adjustment?