The potential unevenness you identified is why it takes 300 points to be excluded from a level.
If you normally attend comps which are small, you will accumulate fewer points by dancing in them. If there are usually 4 or fewer couples in your event, the most you can earn in a single event is 3 points. The same level couple dancing where the field is larger will most likely have a result with similar point levels. Couples who regularly win in small comps may not do so in larger comps. Couples who regularly win at either will point out in similar time frames. Couples who are better dancers will regularly place higher, earn more points, and be pointed out more quickly than couples who are just beginning, or who have really not mastered the proficiency level yet. How does this 'hurt' newer dancers? It gives them plenty of time to work at that level before being required to move up. Remember, there is no prohibition from moving up at anytime you feel ready to do so.
I agree that dancers are always able to dance up, but do dancers interpret a proficiency system as "lowest level" to dance, or as a "qualifying" system? From the collegiate scene, it seems more of the latter - that couples will not(or with hesitation) move up until they have "placed out" of the previous level. However, if the former, the new system would certainly increase the quality of dancing at each level - including bronze(or which ever for beginners). With higher quality dancing at the beginning level, it raises the barrier to entry for newcomers, and this is what I mean by "hurts" them. It seems that a better system would ease dancers into more competition, rather than a "linear" system like this. That is, make it easy for dancers to start in bronze, set low standards, and move people out of there quickly. As the levels progress, it becomes harder to place out(or "required out"). Those who stick with it to the highest level are already the 0.1% of those who stuck with it, so already demand more rigor and higher quality of dancing and competition.
As for the small/large region issue, although I don't have the empirical data to show this, the intuition is that regions of the countries have certain size comps(with variations between the comps), and that a region maybe the limits of how far competitors are willing to travel(on average). The boundary situation is where the region(Region A) has an average of semifinal, vs. a region(Region B) with an average of say a 1st round. The winner of the first region will get 12 points, while the winner of the larger region(say 4 rounds, 48 couples) will get 14 points, ceteris paribus. If we assume all dancers are equally capable of dancing well(quality is normally distributed amongst all competitors, and average quality for a level is the same), then beating 48 couples suggests the 1st placer will more likely beat the 1st placer who beat 12 couples.
It seems arbitrary to have a formula with a hard cut-off of semifinals especially when the number of competitors grow exponentially! Why not: POINT = total # of competitors in 1st round - placement? In the example above, a competitor in Region B can also get 12 points out of 48 couples by beating 12 couples, or placing #36 at the comp. Of course, competitors in Region A will most likely compete against the same people more frequently to get the same total points.
Of course, a good dancer will do well regardless of the competition, but on average at nationals, it may be that the competitors from larger regions(who do not dance up) will more likely do better than competitors from smaller regions. As a system, it seems to bias larger regions unless there is an implicit understanding that competitors from larger regions should also dance up.
