I don't disagree that there is an objective component to most questions. But there are a couple of issues with requests for information in which objectivity is not clear, some of which relate to the SE format:
1. Questions that are open-ended (the famous "what should I do?"-style questions) are subjective, but not so much because of the content of the questions themselves. The problem is more about connecting answers to the question actually asked.
It's more a matter of individual IPS users having to determine for themselves what context the asker is in, what things the asker wants, how the asker would value any tradeoffs they might have to make, and on and on. The result is potentially infinite answers, all slightly different from one another, none of which are more clearly valid or related to the question or the desired result than others. Such questions are essentially undefined, and so addressing them requires answerers to invent details until they are defined enough to be answerable. "The feeling of the users" may diverge wildly, and without a lot of context from each user defining their views (and the assumptions that led to them) voting doesn't really address this.
The more subjectivity there is in a question, the broader the scope becomes for potentially valid answers, and the chances for answers which don't actually apply to the asker's situation or goal become greater. Questions become less clear, and answers less useful.
2. Ironing out the subjectivity from (1), such that everyone can agree they are discussing the same thing, doesn't work with the SE format.
As is often pointed out, SE is not a discussion forum. It is not designed for, nor intended to facilitate, ongoing conversations or discussions. It's about questions and answers, which is why we require that questions include things like clear, specific goals relevant to the stack. The amount of discussion and consensus-seeking to resolve (1), or to make such floods of answers usable in the intended SE mode, requires more communication than the short, one-off essays that comprise answers here. Voting simply doesn't convey enough information, and even if it did votes can only be cast after the answerer has made every assumption and baked it into the answer.
This is especially important in the context of guessing at the minds and motivations of third parties. Taking your example situation, we don't know the person with whom you would like to be persistent but not stalk. An interpersonal approach to that issue is going to involve a lot of observing her reactions and responding. An answer like "9 out of 10 people surveyed would not consider behavior X to be stalking" is not useful here-- there is only one relevant opinion, which we can't know or even guess at very well. Collecting a nonrandom sample of opinions from the internet isn't even a terribly reliable guide to what "society" would think-- the IPS.SE user base is not likely to reflect broad groups of people around the world very precisely on issues like social appropriateness. This is decidedly not a "wisdom of crowds" scenario, and IPS.SE is no more a polling site than a discussion forum.
3. Formulating a fundamentally subjective question into an objective mold doesn't address these issues.
It is objectively true that I have an opinion about the plans expressed in your question which prompted this meta thread. That my opinion exists is an objective fact, but this does not in any way dilute the subjectivity which informs that opinion.
Is it socially acceptable to do X?
is not a question that has a firm, objective answer. X might be OK in some places, but not others, etc. Reframing this as a question eliciting an objective statement only means that a different question has an objective answer.
Does Upper_Case think it is socially acceptable to do X?
is a question with an objective answer, where X is defined and I have an opinion about it. But now the question is about what I think, and not about what is objectively true. By changing the question to an easier one it becomes possible to answer objectively, but at the cost of all relevance to the original question of whether or not X is OK to do. And, as above, it's not a more objective question for anyone other than myself-- nearly 100% of the subjectivity problems remain.
This is always a problem on IPS, where questions skirt the line of having a clearly, objectively correct answer at all. That's why the "back it up" policy is so important here-- why you think an answer will work is more important here than on, say, StackOverflow, where solutions can be tested easily, safely, and quickly to verify that desired results will follow.
Whether or not posters would like to discuss more subjective topics, the SE format is meant to elicit clear questions with informed answers. Being interested in what people think about a topic is fine, but the SE model can't support it well. Case-by-case review on subjectivity is probably necessary, per Jesse's comment. But I feel that we should be more skeptical of subjectivity here, not less. I just don't see what good things a more relaxed approach adds, while I do see downsides.