And that's where you wrong, they would fit perfectly if you made an effort to accommodate them instead of being dogmatic about them.
Just because they can legally ignore user feedback doesn't mean that they should.
In our field, again, there are a great many questions which seem to only be matters of opinion--and that's okay! That's fine! That's how people work, and to pretend otherwise is foolish.
Many architectural decisions in computing are heuristic, right? Many solutions are the result of opinion, because nearly everybody outside of hard-core mathematicians and computer scientists lacks the language to even describe their problems in such a way as to avoid opinion. Even the folks that do have that ability are likely working on a problem where the assumptions are incomplete and ill-defined anyway.
This is a faulty binning of questions into "is question about opinion" and "is question that is not matter of opinion". I posit that the former bin is quite useful and shouldn't be worked against.
In the "but, questions and answers!" vein, it's the "answer" part that's important. I believe they've come out and said they only want to handle questions that have one single answer, in full knowledge there are many questions that don't fit in that mold.
Architectural decisions are a red herring: they don't want to answer heuristic-driven architecture questions. That's not in their scope.
No one's saying those questions aren't useful. They're only saying that SO is not the place for them. This is not complicated. As a programmer, avoiding feature creep is something you almost certainly already understand. It's basically the same.
Disclosure: they have closed a few of my answers that I actually went to some length to make good - but apparently "which library" type of questions are not appropriate for a programming Q&A site?!? Go figure.
Exactly.
Also, answers (by themselves) cannot be closed, only questions can be closed.
There's nothing stopping you from setting up your own opinion-only SE clone.
Your observation is a perfect example of a fact which is not an answer.
So why keep asking?
You'll get better results creating it yourself, or persuading someone else to create it.
We're in a young-enough industry and practice that opinions are really the coin of the realm--there is not the same standard accept methodology you'd see in, say, mechanical engineering.
To pretend that opinions aren't somehow a useful component of learning here is absurd--all the more so because beginners need opinions to start. Once they learn more, once they get exposed to other ideas, then they can form their own opinions. But to pretend that this happens in a vacuum is quite wrong.
Your example "Which is better, Ruby or PHP?" is exactly where opinions, properly backed-up, are useful: a good answer will say "Well, Ruby has these great metaprogramming features, but PHP has a much larger developer pool, and so on". Bad answers will of course just be "ruby is teh 1337 n00b". If only there was some kind of way that Stack Overflow let users filter good answers from bad answers...
At the end of the day, opinions and their debate are what are most useful to a beginner, especially when they don't know what questions to ask or issues to consider. The big failing right now is that you aren't trusting your community enough to filter out the garbage.