We're probably going to have to agree to disagree. As a GM from personal experience I know which one gives me more control over the narrative, and as I speaking from personal anecdote and experience no amount of argument on your part is going to convince me my own experience didn't happen.
But in an effort to not leave it there, I'll try to make a couple of arguments as to why you are wrong.
1) Rulings are public and "fudging the fiction" isn't. This is a very powerful advantage that altering the fiction has over rulings if my intention is to maintain control over the narrative. Whenever you publicly usurp player agency you are risking player revolt and at the very least player dissatisfaction. You really can't afford to offer up many unfair rulings without completely losing player trust. But you can definitely fudge the fiction as much as you want and if you aren't stupid about it the players will be none the wiser. This is huge advantage in power level compared to rulings because I can wield the power and not be challenged on it.
Reading that paragraph it seems like you are talking about railroading, which yes, is both abhorrent and robs player agency so that the DM can control the narrative. Railroading includes illusionism that's virtually undetectable. That's not what I think of when I speak about DM fiat, though I suppose it is an abuse of it.
When I talk about DM fiat and fictional changes, I'm talking about 1 of 2 things.
First, the DM changing the lore or fudging something. An example would be a DM stating that the Spellplague didn't happen in his Forgotten Realms. I did that. Another example would be the DM who placed an ancient dragon for the party to encounter, but because he knows that the party is down more resources than expected due to poor rolling, changes it to an adult dragon to make it a challenging encounter, rather than one that will probably kill a PC or two.
Second, the rules change that affects the fiction. This would be the DM only realizing that the ancient dragon was beyond the player resources after the encounter has begun, and lowering the AC and some hit points to make the fight easier. Or making it so that his game uses the Gritty Realism rules. These are rules changes that also change the fiction.
The DM fiat you describe above is already covered under the term Railroading, so think it just confuses things to include it here as well by a different name.
2) Most rulings by a good GM don't actually assert much or any sort of force over the fiction. A lot of my rulings come down to, if you boil it down to the essentials, "The rules are vague here, flip a coin." Leaving the result up to a fortune check is not nearly as powerful of control over the narrative as altering the fiction. At most a ruling says, "No." to a player proposition, but again, it's a public "No" and you have to justify it in a way you don't have to justify 4 new orcs appearing from a previously unseen room to join the fight. The only way to "check" the integrity of that is to look at my notes, and that's generally something considered out of bounds.
I don't understand the assertion that leaving it up to fortune is not nearly as powerful as altering the fiction. Almost every change to a rule via a ruling is going to have profound effects on the fiction, regardless of the method of introduction(random chance vs. choosing).
Let's take the infinite wish simulacrum hack. If you rule that it doesn't work, it drastically changes the capability of wizards to interact with the world in a major, major, visible way. They no longer can. If you rule that it does work, then wizards can easily reshape reality on a daily basis. Unless you rule that the first wizard to figure out the trick used 100000 or so wishes to ensure that anyone who tries it later fails and dies. It doesn't matter whether you rule for or against it by chance or intentionally, the fiction is going to be affected in a large way.
3) Umbran said there was no way to measure fiat and strictly speaking he's right, but I think we could as D&D players measure fiat by analogy to PC spell use. If the GM's usurpation of the narrative were a spell being used by a player to assert force over the fiction, what level spell would it be? Like what level would you assign to a spell that was trying to do the same thing like add hit points to a target or increase the DC of a roll or summon 8 new orcs or create an entirely new previously not there passageway complete with an autolocking door that closes behind the villain as he runs away. I think if you look at it like that with a humorous but reasoned eye you'll see on the whole just how much more narrative force you are asserting as a GM over the game when you create fiction than when you make a ruling. Most rulings effectively only slightly alter the odds of success of a proposition, and as I said don't necessarily involve any attempt to gain narrative control at all. All alterations of the fiction are attempts to gain narrative control.
Well, yeah. The DM has the ability to drop mountains on a city just because he says, "A group of cultists uncovered a rare tome that taught them how to do it." DM fiat is tremendously powerful, which is why it is both the best tool in the DM's toolbox and the most dangerous if abused. It's not really measurable, but we can estimate that rulings that change both rules and the fiction are at least as powerful as those that only change the fiction.
It's not necessarily against the rules to give a creature more hit points. If I adjust the griffins hit points from 40 to 60 on the fly, that's still probably going to be within the range of normal hit points for a griffin.
However, adjusting the hit points of a monster whose hit points had be previously secretly established is not a ruling at all. It it is however taking control of the fiction.
An alteration of the rules by the DM falls under the "rulings over rules" mantra. "Rulings" in the context that 5e uses the term isn't just for adjudicating rules issues. It's also rules changes. Fudging is in the DMG as a DM tool in 5e by the way, so a DM altering hit point after being previously established would not be a ruling, it would be RAW.
