I like the premise of Blitz's idea. Not all maps had to go through a long hard process to get approval (especially not the rigorous system that exists now). Crosswords and other early maps were posted, touched up a couple of times and then loaded to the site. They would never be approved today in their current condition, and therefore it does make sense to bring them up to the standards of the rest of the foundry. In fact, that is the entire reason that so many maps have already been revamped.
However, I think the proposed system is too strict. As many have said, gameplay affects how much a map is played, not just graphics. Revamps only take graphics into account, unless the original artist wants to drastically change the map. That can't be done without the original artist's help, so simply revamping the bottom 3 maps won't always be possible or necessary. Keep in mind that there will ALWAYS be a least popular map, even if it has great graphics.
The useful part of this suggestion, IMO is simply ranking the maps. This used to be done in various threads, I remember cairns kept a lot of detailed stats. You need to look at how many times the map is played each day, each week, each month, in total, etc. The more data the better for analyzing how popular a map is. Once this list is compiled, a judgment can be made as to how to change the bottom ranking maps, if necessary. For some, a graphical update is all that is needed. For others, it might be better to simply pull them from the site and really change the gameplay to make them more interesting. For others, they might be perfect the way they are because they satisfy a small niche... a map might not be widely played, but that doesn't mean it is not well liked by some members. If the map is unique, and it satisfies a niche, then there is no reason to change it even if it is not played very often.