There's 2 similar but subtly different philosophies of scoring system design that, while often coexisting, exist in a tense relationship because they have some core incompatibilities. It's important to recognize which type is being discussed or designed cause it will help avoid confusion. It'll help devs avoid bashing their head against a wall trying to solve unsolvable problems.
I'll call them CAG (character action game, think DMC and Bayo) scoring and arcade scoring, though they don't only apply to CAGs or arcade games. If someone has better names I'm all ears!
CAG SCORING
CAG scoring is either like nudging, or like an extra "artificial" difficulty mode laid on top of an existing one. When it's a nudge, the goal is to get you into some kinda desired "endgame" playstyle like freestyle comboing or what have you. When it's an extra difficulty, it's usually just trying to filter out the most degen strategies and encourage you to play loosely, but more thoughtfully than just survival.
In a lot of ways, CAG scoring benefits from low caps because low caps allow for a variety of playstyles - your goal is general (achieve S rank equivalents) and how you get there is up to you. The games tend to not be very interested in evaluating your performance past that point (Bayo even has a soft cap on its encounter scoring) because they want you to feel like it's "good enough" in some sense. CAG players tend to complain when rankings are too strict and funnel them towards specific playstyles.
Similarly, CAG fans tend to complain when the scoring doesn't build on the games' natural gameplay - if a game has a million moves and a robust air combo game, then scoring abandoning all of that in favor of some arbitrary BS will feel like the game being deceptive.
Because of the "stepping stone" nature of the scoring, or its emphasis on clear, capped goals like ranks, CAG players tend to not mind things like exploits and infinite milking as much.
ARCADE SCORING
Arcade scoring is the endgame. It's trying to create as much moment to moment depth/challenge as possible, ideally so you can replay the game forever. It's not a stepping stone towards some kinda other unquantifiable playstyle, it's trying to create create a fun playstyle via quantification.
Low caps are poison to arcade scoring. At best, they are annoying since players have to find ways to work around them. At worst, they will kill a game outright. Scoring is at its best when it evaluates your performance past "good enough". It benefits from measuring stuff in a granular & robust way, creating a clear hierarchy of good vs bad options. Players can have some freedom & wiggle room if the scoring is complex enough and has enough internal conflicts to where best choices are insanely ambiguous, but that's not really the point.
Some people do prefer arcade scoring to build on a game's natural mechanics, but it's also just as common to see score players who don't care about that shit and will just appreciate a scoring system on its own terms. Detaching scoring from natural game mechanics even frees up the design space and lets devs go wild. The goal is to make it engaging & interesting to optimize long term.
Why Is This Important?
The two types of scoring have contradictory goals - you cannot create a particularly good arcade scoring system that has a low cap. Arcade scoring isn't about freedom or expression, it's about creating an interesting, often quite self-contained optimization challenge.
CAG scoring's tolerance of low cap jives better with their emphasis on complicated mechanics. It's easy to balance the what, 5 or so mechanics shmups have to create compelling score play. Balancing 80 moves though? Good luck with that. And when you have games with a lot of mechanics but only a tiny handful of useful ones, it makes them feel a lot more dishonest than simpler games - you expect complexity and get simplicity, instead of expecting simplicity and getting....a bit more simplicity.
So if you're, say, making a scoring system, you need to think about which style you're going for and focus on that. Trying to marry the 2, while not IMPOSSIBLE, is probably going to end up in undercooked scoring across the board where it'll be too hard for varied playstyles to be viable, but the scoring will be too hard capped to create cool long term optimization.
I will say though, I think it's easier for arcade scoring to have a CAG component than vice versa. In shmups all you have to do is throw in some easy-ish to get ranks. In CAGs, I think people will be disappointed if you don't manage to somehow work in natural mechanics into scoring, even if the scoring system is cool on its own terms.
Also shotouts to this Stinger article covering scoring systems from the CAG player perspective