I suggest that we should have a list of scoring functions. When there is only one variable in the score for example:
- Length of submission (code-golf)
- Portion of test cases passed (test-battery)
- Number of languages (rosetta-stone)
This is easy.
When there is multiple such as any two of the above it becomes hard.
Some time the challenge wants to use two but restricts it to one; as the common "Number of correct test cases, in less than 128 bytes" This is less fun since it outright knocks out languages that need a lot of boilerplate.
The main way (and what this question is about) is scoring systems that combine multiple scores into one.
There are better and worse ways to deal with this. It is a heuristic, and there is a whole field of study relating to multi-objective optimization heuristics.
The best scoring function depends on the challenge, but for similar challenges it should be similar.
Consider, as pointed out by user202729 and Mego,
I would argue that the later scoring function is better than the former, because you can see in the former that the best scoring function is for solutions that only use 2 languages, where as to me the more interesting answers and the ones that are more challenging to write are the ones that do many languages.
Now even looking at just this problem there is no obvious objectively correct answer. So I suggest in this post posting one scoring function per answer, with reference to what it is good for, and links to questions that use it, and/or questions that could use it.
Bonus points for showing plots.