[Aldehyde]

They mentioned in the original flash chromatography paper (J. Org. Chem., 1978, 43 (14)) that it is ideal to get your target compound with an Rf of ~0.35, I don't understand why though. What is wrong with getting a target compound with an Rf of 0.7 or something and flush it out first?

[TheUnassuming]

It all depends on how complex the mixture is you are trying to separate.  If your compound is the least polar by a long shot (one compound with a salt contaminant for example), you can just flush it as you say.  In the literature this is usually referred to as filtering through a silica plug (or something else similar).   
The Rf of 0.2/0.3 gives the "optimal" amount of interaction with the silica gel to separate out your compound from any surrounding compounds.  Its really just a balancing act because if it moves to quickly then it will not separate from any nearby compounds, but if it moves to slowly the band will get too broad and will blend with nearby impurities.