TLDR: 1. You DO need to teach the interface and basic operations, and concepts. But keep it brief.
2. Experimentation (or as I prefer to think of it, "play") is better done in the game itself.
3. Usability testing is rad. It lets you know your audience and gives perspective. Your audience is rarely who you think it is.
Even assuming that your numbers are right, which they can't reasonably be as people are not generally morons, especially not people who tend to pick up games of that style, and assuming that the system of playing the game is not trivially simple to master anyway (seriously, you have click to move, 2 moves per turn, cover, how hard can it be) that would still leave your argument empty and hollow due to the simple fact that any system which prioritizes 1/3 of your customers at the expense of frustrating the other 2/3 is not a good one.
I mean, the thirds thing is almost by necessity true assuming average task and average group. Its similar to the idea that 50% of people are below average at any given thing.
Your use of "moron" suggests a potentially toxic attitude. As a person with an extremely uneven skill profile I have been in all three groups simultaneously and decades ago in school this would cause a certain amount of expectation whiplash. Teachers who encountered me first in a high ability group wrote me up as having behavioural problems relating to some tasks when in fact I was simply struggling. In the reverse situation they would initially be suspicious and then would ignore me as being inexplicable. I'm saying this because I think being in the moron group a few times would give you a broader perspective.
One problem here is you think noone could be that stupid, but it is a misconception that ability to game is (only) a function of intelligence. Games and interface usage are largely based on knowledge and experience and there are many good reasons relating to ability, cultural familiarity and convention that they may not be highly skilled with them. Its not unlike learning a foreign language. Heck, lets get super basic: Young children have to be taught to take turns and to roll the dice. They are not morons for lacking this knowledge, and a person is not a moron for lacking the knowledge of what a movement point is.
Another problem here is that you think you know the customer base, the "people who tend to pick up games of that style", which I'm sure you feel are people just like you and who you'd describe in words you'd like to apply to yourself. Unfortunately I don't have Firaxis dataset - what I have is the Global Steam achievements. 85% of players complete the tutorial, 57% of players get to the facility, 34% of players complete the game at any difficulty level, 6% of players have completed the game on difficulty 3/4. (Aside: These player retention numbers are actually very good! Dropout would be higher from an overlong tutorial.) Firaxis brings gamers of all experience levels in house and just watches them play their game, a form of usability testing. Firaxis knows their userbase and that is why the tutorial is the way it is.
In a way it is charitable of you to assume that everyone is as smart and knowledgeable as you are, but there are hints that you would very quickly lose patience with those who can't learn as fast as you could teach.
Either way it is irrelevant because that method of teaching won't benefit anyone. Least of all the ones that need extra care. It simply is one thing to accommodate your teaching method to the worst students you have and quite another to give up on the concept of teaching altogether in favor of mindless dictation of test results.
The later method literally transfers no knowledge, no applicable skills or information beyond how to obey orders. And thus it is worse than useless. It is actively detrimental.
I'm seeing a bunch of words and descriptors with emotional and value associated baggage here but not a lot to engage with.
Now, you seem to think I am against all forms of handholding. That is not correct. What I am saying is that there is a massive difference between careful handholding and hamfisted handholding. The former can work just fine and is seen in games like UFO Aftermath or Portal 2. There you still have very specific things you need to do (move to objective, look up etc.) but the player is given room to experiment with the controls in his own time and the objective is there as a test of that understanding as opposed to a click to proceed button.
I don't think there is a huge gap in what Firaxis and Valve did. Portal 2 is in a genre that is able to do away with UI while nearly no strategy game is. The Portal 2 intro is an extremely tightly controlled environment with an additional layer of using Wheatley to distract you from bumping into the walls of your tiny fishtank. Its very effective but I cant imagine the resources that went into the creation and iteration of it, though I have no doubt they used similar observed playthroughs. I bet the commentary mode has some interesting insights.
An aspect that you don't seem to have considered here is that in Portal 2 it is (usually) impossible to make a wrong choice, so the tutorial cannot be failed and experimentation is harmless. How would you allow this in your hypothetical strategy game tutorial. Its a tricky question. Maybe some kind of rewind button? (Into the Breach shoutout).
The hallmark of a good tutorial is that it contains all the elements of a good educational project. You have your demonstration of applicable skills (your controls are explained), room for students to experiment with those skills so as to become comfortable with using them and finally a test of their mastery of the subject (the objective).
Strong disagree. This is not the tutorial - this is the tutorial + the first few hours of the game.
Dark Souls is sort of the exception that proves the rule in terms of accessibility, but is a useful example in this case. Possibly because (nearly) everyone approached the game as novices. The Asylum was essentially a long corridor where messages primed you with a response, then you encountered a situation where you could trust that that response would help you. Its a mini-rehearsal for the Undead Burg which is the chance to employ your learned skills and yet, importantly, is most definitely within the game and outwith the tutorial. DS1 even accommodates experimental overachievers with the lockpick and sequence breaking, but only after making the "correct" path of the aqueduct clear.
I thought DS1 was much more effective in this structure than DS3. The graveyard was both too wide and too small compared to the asylum, and Iudex Gundyr was too hard for a first boss.
I'd consider XCOM2 as fitting this structure in the form of Tutorial Mission / Early Missions / Early Plot Mission that gates the Midgame. Tutorials should be brief and teach the essentials. Experimentation is play and should be in the game, not the tutorial.
The XCOM 2 tutorial offers very little of that. It arguably does have a demonstration of what you can do with your units. But that's it. It does not offer you the time or space to experiment and get familiar with the skills you are supposedly being taught. And in the end it has no test of those skills unless you consider a multiple choice test with only one option testing. Thus the player leaves it with a very vague idea of what the game is about, no practice at playing it (the most vital part of any tutorial) and no idea why he had to do the things he did other than plot. He gains no understanding of the subject matter and his first real test is the first real mission which he will inevitably struggle through.
In a more complex strategy game that sort of tutorial would literally have been disastrous to new players.
You are damning the virtues of brevity, clarity and simplicity here. Lets consider your very first example that started this all - that you were being grumpy about being forced to make a wrong choice. Ever hear of show, don't tell? When is the appropriate time to teach a player what an HP bar is if they went through the tutorial without getting hit?