One effect that we see from using DRS in qualifying is that the relative performance of the cars change from qualifying to the race. For example, Mercedes qualified relatively well in the start of this season, but performed worse in the races. (Although this was mainly due to tyre wear and poor reliability, it could also partly be a result of an efficient DRS.) This is important, as DRS will only aid overtaking in situations where a slower car is in front of a faster car. The straight-line speed advantage of DRS does not fully compensate for the loss of downforce when one car is trailing another, it only reduces the overal disadvantage. Even if overtaking was easier than ever, we would rarely see one if the cars started in the same order as they would perform in the race. The fastest guy would start in front and take off in the lead, the slowest guy would start in the back and stay there. So allowing DRS in qualifying is in conformance with the other measures that are made to improved the show by altering the performance between qualifying and race; tyres that wear out faster, low fuel in qualifying and no refuling during the race.
If we compare this to the situation before 2005, it has improved the show a lot. Back then, the cars qualified with the ammount of fuel that was used at the start of the race, and the tyres were much more consistent and refuling was allowed during the race. This meant that the relative performance of the cars was almost constant and we hardly saw overtaking at all. Some positions were altered du to the fact that some qualified with less fuel than others, but the result was not an overtaking on track, but alterations of positions during the pitstops.
Another important aspect, of course, is to avoid what happened with KERS in 2009 where it turned out to be an advantage to optimize the car without using the system at all.