The rule against multi-load games has been in the rules of the contest since the first edition in 2013. Its intention is to set an equal base for all contenders, as well as to fix sizes of games so that producing them on a single cassette is viable. Also, as the contest is annual, memory limitation also puts all contenders in a space where best quality results are achievable from one contest to the next. This also prevents (not completely, but mostly) an uneven competition between games developed in 2-4 years time with games developed in 6-months time. A 6-months time game for a 64K target can achieve similar quality levels than a 2-4 years production. Therefore, people wanting to compete are not to be discouraged by superproductions, as it is always possible to find enjoyable and original ideas that can be done in N-months time and will be ready to compete.
In this sense, the best possible rule would be "fit everything in 64K". However, loading screens are traditional and presenting games in cassette without loading screens doesn't feel good. So we decided that loading screens should be permitted. Then, it is difficult to draw a line between types of loading screens without bothering authors too much. For instance, loading screens such as Dragon Attack's are really nice and valuable, but could be easily prevented with something like a "one loading screen maximum" rule. We have discussed using rules like this, and they are periodically considered for the same reasons as the rest of the 64K rule. However, at the moment, we feel it ok as it is by now.
However, the 64K rule is clear: once the game starts, no more loading is permitted. Any loading from that moment onwards is a violation of the rule, and leads directly to being disqualified. And, of course, we have many times being asked for permitting loaded ending screens. This represents a source of troubles. First, an ending screen outside your game (loaded) gives you more space for your game. That is contrary to the idea of limiting all contestats and requiring less development time to have a competitive game. But one could argue that same is applicable to extended loading screens. We agree, and that's the main reason why extended loading screens are constantly being discussed. However, the greatest source of troube is making exceptions to the "64K rule". Once you start with an exception, you give a new valid argument for making new exceptions. We don't want to go down that path. That path goes agains exposed principles of the contest in its present form. So, it is much more probable that we end up limiting extended loading screens, than we let contestants add a loaded ending screen.
Hope that this explanation helps understand our design decisions better.
Maybe, in the future, we could think of other categories and/or branches of the contest. Or, even better, may be other organizers will launch to create different contests with different conditions to better suit other developer interests. We are not against different ways of creating games for Amstrad: we are just fitting community, developer and educational requirements into CPCRetroDev the best we can, up to the limit of our abilities. As Amstrad users and fans, we would always love to see as many different developments and new games as possible.