Rule 2 seems a little unfair, where the user has no control over the code created. If you’re coding directly you would obviously write as concise as possible, whereas generated code is not going to do that.
I agree, and I think there would be no objective way to compare the “amount of code” between systems. After all, Inform 7 code is measured in words; others are measured in lines. Furthermore, two different coders can write exactly the same code but have different line counts because they have different coding styles. Some coders use separate lines for all code block characters; others do not. Some coders leave more comments. Most coders leave blank lines between object definitions, but they almost certainly don’t have to.
Writing the shortest possible program that does something is maybe a fun exercise if that’s your poison, but generally speaking I don’t think it’s a particularly good metric of how “good” the code is. It’s a measure of how short it is. Just saying.
Anyway, If you do go ahead with this I agree rule 3 is silly. What you should do is pick a winner for each system, that way coders using a particular system can compete against each other, but it will probably be easier to win a prize if you pick an obscure system.