In this question, we softly addressed the acceptability of the subject. There's no contribution other than my own answer, and it's not exactly a very attended post, but build questions haven't really come up all that often since the start of the site.
Assuming for a moment that we're still okay with accepting build questions (as, for the reasons I explained in my earlier answer, builds always have some "core" aspect that is very much concrete for us to use as the subject matter of questions, otherwise they wouldn't be "builds"), my question is, how do we sculpt them into proper authoritative and objective Q&A?
This comes up because this question, as it is phrased, is very much a subjective and open-ended question about the "best" build. Looks pretty close-worthy to me for that alone. But then I looked at our highest voted build question, which pretty much asks the same thing, sub "late-game" for "early-game" and minus the explicit usage of the word "recommend". Neither of those attributes actually change the nature of the question, yet we rolled with it.
We can attribute the lifespan of that early question on account of both it being long before we starting to watch out for quality issues and mayhem, and also the fact that it got an incredible answer that isn't a mere recommendation, but an analysis of different builds and how they are constructed. There's a reason it's climbing up to be our first Great Answer candidate.
We can't rely on age and answers alone, though. As the question stands, it's bait for polling, and I'd like to avoid that. How can we revise questions like the DotA so that they can attract high quality answers instead of just random recommendations? How do we make our examples not look like the "What is the Best X" that everyone loves to jump on everywhere else? Is it possible? Or is it the kind of thing we have to keep eyes on, and just trust our answerers to salvage?