If the flash is run slowly, the molecules can diffuse via Brownian motion to make the bands mix (blur/flatten/broaden, whichever word sounds best). If you go slow enough, that can undo the separation, but is not usually a problem unless you go really slowly. I think the main issue is that with smaller particle sizes, the flow is very slow without some pressure, plus you tend to get more voids in the column without some pressure which is bad for the separation as well. That is why Waters used to put radial pressure on their older columns in their HPLCs, but now stainless steel and higher pressures have made that unnecessary. But using an ISCO for example, you can see voids in the column forming if the flow is too low, at least with some solvents (CHCl3 shows it well).