Issues with the linked question:
As I see it, the problem with the linked question is that it's not clear what would make a good or bad answer to the question, nor what would make one answer better or worse than another. The criteria for making those determinations aren't defined in the question. That responses are possible is irrelevant, and a standard like "internally consistent" isn't a high bar to clear.
It's not that different from asking a question about what Rogue build is best, but failing to state which game or edition the question relates to. Answerers can certainly offer an answer from some arbitrarily chosen game and edition, but it's not clear why that would be better than the question itself indicating that information and getting a more focused answer.
An optimized build for D&D 3.5e might be totally different than for 5e, and so each could be equally correct for its assumptions, without any way for one to be a better answer for the question than another. If the question is about which optimized Rogue build is best across all D&D editions, answerers will need some common set of criteria to compare to arrive at an answer. A question should make clear what it is asking.
SE is not great for discussions. Chat is an option, though.
What you seem to be suggesting here is that each answerer write an answer that is specifically not providing a direct answer to the question, but instead makes an argument for why their working definition of "stronger" is the correct one, and then that their strategy correctly optimizes for that definition.
That's a fine discussion to have! But that's just the thing-- it's a discussion, not an answer that clearly resolves a specific question. The SE format handles discussions poorly. There just isn't enough opportunity for responding back and forth to resolve the ambiguity which the question, at this point, is explicitly refusing to define. That can make questions less useful to future visitors. It can also attract lower-quality answers, which we prefer to avoid where possible.
Really big, expansive questions can be problematic.
If an adequately argued, complete answer to the question requires several pages of description, along with exceptions and conditional cases, it may be too broad to fit the SE format. If there are a dozen scenarios which might affect how "strongest" is defined, it might be better to have a dozen questions each covering one of those so that each answer can have a chance at being clear and concise.
Asking that answerers address every possible explanation of "strongest" or "optimized" fits into the "really big, expansive" category for me (and apparently for other users as well). And that is what this question demands, as each answerer would need to justify their asserted definition as the right one even before laying out their proposed strategies for optimizing for that definition. That's a lot of work, both for answerers to write and for future visitors to read, and fits poorly with the Q&A framework of the site.
The sum of two good, independent questions isn't necessarily a good question.
What is the build that offers the highest consistent DPR for a magus, one based on Str or one based on Dex? is a more focused question. What is the build that avoids or negates the highest average damage for a magus, one based on Str or Dex? is also a more focused question.
But if they are simply combined, some problems might arise. If there is a tradeoff to be made between higher DPR and greater damage evasion or negation, which of those should be preferred? If we're trying to optimize for both at once, that's not clear. If you know how you want that tradeoff to be evaluated, you can explain it in the question. If you're not sure, then yet another good question might be Which is more valuable to a magus: higher DPR or greater damage evasion/negation?
I don't see much benefit in forcing that sort of information to be undefined assumptions in an answer versus the focus of a specific question. Again, that might be a great starting point for a discussion, but this isn't a discussion forum.
Optimization questions in particular suffer when insufficiently defined.
There are a lot of factors which might affect what an optimized result is. Optimization itself tends to make use of very precise information, and so can be sensitive to changes in underlying assumptions. It may matter if a campaign takes place mostly underwater, or on another plane of existence.
Decisions you make about how your character develops might change whether or not a given strategy will work, which could be hard to appreciate without taking those developments into consideration. Are you intending to tank damage for your party? That sort of decision matters a lot, and there are many possibilities that an answerer would need to consider to answer in the general case.
Your character's access to certain items or equipment, or the types of enemies they encounter, might render entire strategies suboptimal or irrelevant even if they might be optimal in circumstances you aren't dealing with.
Even this meta question seems to define what you're looking for more clearly than the actual question: a framework for evaluating which approach is better, and which you can apply to any specific game situation. Such an answer does not seem, to me, to be an optimization.
If, as you state, the answer you're expecting is
"Roughly balanced in terms of strength, so it comes down to priorities, specifics, and preferences."
I don't understand why it would not be acceptable to include some of the priorities, specifics, and preferences you might have in mind in the question itself. That sort of information would define the question in a way that allows for more focused, directly comparable answers. If character tier is the standard you want to use, the question could be changed to be something like What is the highest tier that can be reached by a magus based on Str versus Dex?
Or, alternatively, if "it's based on priorities, specifics, and preferences" is the kind of answer you want a better question might be something along the lines of Under what conditions would a Str-based magus outperform a Dex-based magus for measures A, B, and C?. That might run into trouble due to being opinion-based (how can answerers directly compare preferences?), but at least the dimensions of the question would be better defined.
