From a purely abstract concept, I agree with you. If an 'ugly' theory is more effective than either the nature of what is theorized is too complex to be covered in an elegant manner, or that we have not yet discovered how to reduce the 'ugly' to something elegant.
In terms of game designers... I think the aesthetics of their rules should be something personal and be a part of the concept of the rules that are provided. In other words, a game designer should decide for themselves how elegant or ugly their rules are to be based on their own desires.
I completely agree that it is good for designers to make informed decisions about what design aesthetics they wish to include in their games. Again,
ignoring these things is a bad plan. My problem (and it's mostly directed at
fans, and in particular
playtesters) is when those decisions are NOT informed, but rather operate pretty much purely off of the expectation that a certain design aesthetic simply IS good design, not even needing to be defended because it is
obviously good design.
A D&D game with a keyword-based rule system as labyrinthine as M:TG would be a great example of a design aesthetic that has been pursued to the detriment of play-experience. Likewise, a TTRPG system that legitimately DOES try to have a distinct rule for every possible situation a DM or player might encounter has taken the
comprehensiveness aesthetic and turned it into an overweening obsession, rather than one consideration among many. This is why I said, for example, that "gamist"/"simulationist"/etc. doesn't really apply here, because these aesthetics can appear in literally any of those and can bend design out of shape
regarless of which thing is being pursued.
I will try to discuss it in more concrete terms because I'm more wired that way. Take the aesthetics of two wildly different game systems...
Dungeon World:
The primary aesthetics of Dungeon World is the concept of making moves. Maybe this is an elegant theory (I don't know if elegant or ugly are absolutes -- like do you intend for ALL elegant methods to be considered ineffective?). Make a move, roll the dice, have the GM respond with a soft move, hard move, or success. This is elegant. Whether it is considered ineffective is in the eye of the beholder. But this elegance is expressed in the universality of the resolution mechanic. The mechanic is the same, it is elegant, and its effectiveness is up to the beholder.
Certainly not all. Like I said, even bolded, design aesthetics--seeking theory elegance, in the physics analogy--will always be an important consideration, and designers(/physicists) ignore it at their peril. But treating elegance as
the most important thing seems to come up a LOT, particularly in playtest discussions, and this bothers me. Playtesting is, at least in part, about helping suss out all the questions, like design effectiveness, the game's message or what it communicates to the players,
and aesthetics (both "direct" aesthetics like art and descriptive text, and "meta" aesthetics like comprehensiveness, conciseness, and centralization). I see these meta-aesthetics being elevated to the most important concern, bar none.
But yes, I quite like Dungeon World, it's one of the reason I run it despite generally favoring much crunchier games.
D&D Basic / Expert
There is no real primary aesthetic. Perhaps it may be considered 'ugly' by your metric. Attacks are d20 consulting a 'to-hit' matrix. Performing a thief skill is a percentile roll. Listening at doors or lifting gates is an 'X in 6. chance. Turn Undead is a 2d6 roll. Ability rolls are 1d20 roll under your score. Reactions, Morale, so on, and so on. All different approaches in mechanics. I am a fan of B/X and I would consider this to be 'ugly' under your definition. <snip>
For example, if a Cleric decides to Turn Undead... the physical act of picking up 2 d6's (instead of a d20) creates an aesthetic that the character is doing something different and unique. The different mechanics kind of lead to an idea that there is a uniqueness in the action taken. Likewise, when a thief tries to disarm a trap, it is a unique experience because the d100 is used instead. The existence of the different mechanics creates a 'meta-aesthetic' that the actions taken are different. The game promotes a 'meta-aesthetic' by enforcing different mechanical resolutions based on different actions.
Now, In my personal experience, I don't see this as much in universal resolution mechanics. It feels more same to me because when I Turn Undead, in more modern games, I reach for the same die and execute the mechanic in the same way that I would an attack or a skill check or a saving throw or so on.
I don't know if this is even close to what you are wanting to discuss. I hope I'm close and you take this as a sincere effort

.
It's certainly in the same ballpark.
My response would be that you have identified two different, mostly mutually-exclusive meta-aesthetics. I'd call them "unifiedness" and "distinctness." Fate is a game that is almost maximally unified, where there's very close to only
one mechanic that really is used for nearly everything. And the thing is? I would completely agree that pursuing unified mechanics
purely because one considers them elegant, is a bad idea! Even though I really like unified mechanics,
presuming that unified mechanics are better than distinct mechanics is an error. And I'll give you an example of an argument I, myself, have made where I believe the concern I'm highlighting (putting the aesthetic before any question of function or message)
doesn't apply.
Even though in some ways 3e moved to "totally" unified mechanics, by having almost all random numbers come from d20s, even that isn't totally unified. Damage dice are still other types, for example...but even the use of the d20 is not totally unified. Attack rolls, skill rolls, initiative rolls...all of these things are unified, they work the same way and benefit from the same kinds of things symmetrically. But one type of d20 roll--a very common one!--sticks out as different.
Saving throws. Saving throws make a lot of sense for some people, but they also create some design wrinkles. 4e looked at that and said, "Well...mathematically, there's no difference between the
target rolling a save, and the
caster rolling to hit. Let's ditch spell saving throws and have
all offensive abilities use hit rolls." (It's worth noting, 4e still had things
called "saving throws," but they served an entirely different function.)
One might say, "Wait a minute there, Ezekiel. You just said doing things for meta-aesthetic reasons is a problem, yet now you support it?" But that
wasn't the reason they chose to slay that particular sacred cow. The reason was that making spells use attack rolls the same way regular attacks do
opened new design space. The 4e Warlord or Cleric did not need distinct abilities in order to benefit
both their spell-using allies AND their weapon-using allies. A Warlord handing out an attack bonus, or extra damage on successful attacks, is just as beneficial to a Druid/Sorcerer/Warlock/Swordmage party as she is to a Hunter/Rogue/Avenger/Fighter party, because every time someone uses an offensive ability that isn't guaranteed to land, they make an attack roll and do attack damage when they succeed.
Mechanical distinctness can absolutely be valuable. Having different classes feature distinct mechanics, for example, is a pretty uncontroversial form of distinctness. But really high distinctness--where every mechanic works pretty differently--has some serious costs. TTRPGs have to compete with a lot of other leisure activities, which means that many players will make a "time invested vs fun gained" evaluation regularly. That doesn't mean distinctness is now
worthless, but very very high distinctiveness may be too much for many modern players. Having thief skills be so different from any other skillful activity, for example, can be a real headache for some players. (I have known multiple players who legitimately couldn't retain their knowledge of how to do basic Dungeon World moves from one week to the next, even after a full year of playing the same character. Having extremely high distinctness, where nearly every name-able action has a unique mechanic, will drive those players away very quickly.)
And all of the above--both your post and my own words--are what I
really want to see. Earnest, positive discussions about what "meta-aesthetics" one would like to see pursued
and why they should be. What concerns me is people asserting an aesthetic without defense or discussion, and stating or implying that it
needs no defense or discussion because it's just obviously Better Design. Whether it's mechanical distinctness, unified mechanics, keywording, centralization,
whatever, I far too often see these meta-aesthetics simply assumed as gospel and projected onto
all possible rules structures, no matter what consequences may result. I guess you could call it the "other side" of the sacred cow--that is, where a sacred cow is a structure preserved outside its original context, the thing I'm opposing is
inserting structures regardless of context. Golden calves, perhaps?