SBMM only really work in theory. Skill distribution is largely a bell curve, as soon as you're a moderately above average player it just starts to fuck you over instead; in order for a game to find a match for you it either needs longer MM waits or just matching you against people outside your skill bracket instead.
I think newbie divisions vs everyone else in non-ranked is the best solution. Keep the very newest away from the most experienced for a while, then let randomness do it's job the rest of the time. Given the distribution of skill, anyone in the middle 50% of players will see far more people in their skill than they will others, and those outside of the curve shouldn't play against people way above them too often.
SBMM is what ruined Apex for me, as soon as I was in the uncomfortable position of being quite far above average, but nowhere near a top player, the game became impossible, because it was just matching me against people several times better than me in every match; it's a system that actively makes the game less fair for certain players.
Why don't you want to get paired with people of your skill level? Seems to be fine to me.
Because 99% of the systems don't work; if you're above average the systems break completely. I've never played an SBMM game that has actually sucessfully made me play against people my level, as a decent but not excellent player. I can smash the average player, get destroyed by high rank players, but I never seem to get put against people I can just have a good game against once I've been playing enough to get decent.
Sadly, it's impossible to discuss because you always get a bunch of people driveby posting "lol u just wanna stomp people" when the reality is, we're sick of getting fucked over repeatedly and having to face off against players far better than us. It's game ruining.
The other issue with SBMM is you're asking an algorithm to correctly and accurately judge, with a single number, how "good" a player is. You could have a fucking army of data scientists trying to figure out a good way to accurately quantify that, but it's just not possible. Especially as a game gets more and more complex. It's like a BR game - who is "better", the player who wins 50% of their matches but loses most their gunfights, or the player who takes risks and gets a ton of kills, but doesn't win as often? There's so many variables, it's a purely subjective assessment baked into an algorithm. It's a fundamentally imperfect and biased system.