I think how-to questions are an important part of GD.SE. Often there are many ways to a solution and reading through different approaches one can learn a thing or two. because for the QA it might be a specific solution, for everyone else it's all about the process and workflow.
However, and this is where I agree with Brendan, currently they often tend to be quick fixes which does not provide a challenge, which I feel is a big motivation to answer questions. But it's not all black and white, I have answered questions that can be summed up with RTFM and very simple ones that only need one mouse-click, but were still somewhat challenging or at least the information wasn't readily available. And I have often seen it that other people than the QA thanked someone for an answer simply because it was useful or interesting to them, even when the original question was already answered satisfyingly. And this is important (to me) comunity wise. Exchange of ideas, workflow and know-how. Often more than answering specific questions.
So now we have some "bad ones", but what makes them bad and how would we want them instead?
I personally feel most frustrated when the question is so specific that I feel there is a overlying problem/task which is not part of the question, which makes it so hard to answer. I am of course capable of answering the question, but I often wonder if the solution fits the problem at all. Like it is all on a need-to-know basis for the designer. Which discredits the people you are asking for help. Let me try to outline this with an example. Recently I answered "where to put the menue on a website". While I tried to give insight on how to tackle the problem, I felt the only answers suitable would be "why?" and "it depends". Because there wasn't enough information to give a more detailed answer on the actual problem. Now the QA might have expected an answer like "left" or "top". But that could be highly unsatisfying to the person giving the answer and most likely unsuitable for QAs specific problem. Why? Because it's a design-o-mat question. Just like "What's a nice font?" or "good colors for websites?". Ka-ching, design-o-mat has produced (1) answer. That's just not how creative problemsolving works. Or how community works.
How to encourage people to ask questions with a bit more insight on the actual problem I unfortunately do not know. We can vote, we can encourage them in comments to add infos (I have noticed they get closed a lot faster recently). But yes, like Brendan said, if they don't return, there is no point.
Yes we need them, and like always, we need to focus on the "good ones". With voting, ecouraging and with answering.