The optimal amount of water to change was examined by FAMA I think back in the 80's. Setting practical vs optimal aside for a second, it turned out this was a regression problem involving dilution vs. replacement; that is some stuff you want diluted, some stuff needs replenishment. Water changes do both ....Anyway if you think about it; changing 3% daily is not the same as a single 21% per week ...
I can't recall all the math and arguments, just know there is a difference.... anyway to sum it up, if I recall it was that single larger H2O changes are more effectively dilutive than smaller/more frequent, but obviously less stable...
I am going to try the DOS but maybe do 5-10% weekly