How to ask a good question?

Question

How do I ask a good question?

Imagine that I walked up to you and asked you this question without any context. I believe that the above question is not a really good question, because the scope of the question is too broad. One could for example answer by saying that a good question is asked by speaking clearly, while I'm actually interested what the content elements of a good question are.

So here I can already improve the question myself to be more specific by rephrasing:

Which elements does a good question contain?

I would like to know, what are your rules of asking a good question? How do you formulate a question properly and lastly, which elements does a good question contian?

Tim

Answer 1

Every question rests on the simpler question: "What can we agree on?" Whenever we ask a question, we want to know what's normal, conventional, or accepted within some domain; the language game 'asking a question' implies that we want to know the rules, standards, or structures that pertain in that domain.

People often get tangled between questions and arguments, and that can cause some difficulties. For instance, consider the following two questions:

1. Why do we do this?
2. Why don't we do that instead?

The first is a 'pure' question: it wants to discover the norms or principles that lead us to do a particular thing. The second, however, is the lead-in to an argument. It implies that that might be superior to this; it doesn't ask for an explanation as much as it begs for debate. The first wants to know what is conventionally agreed on, while the second begins to display disagreement with conventions. It's easy to slide down that slope and write ostensible questions that have no interest in an exploration of convention — i.e., are not actually questions —  but instead seek to change conventional agreement through argument.

To ask a good question, we should step back from the factual concern — the specific answer we look for in a particular case — and consider that what we really want to know is the principle by which the answer is derived. We don't merely want to know that 5*12 = 60, we want to know how to multiply. We don't merely want to know that Russia is at war with Ukraine, we want to know where that war came from, and what it signifies. A question should contain a context, and it should point out a confusion or conflict — in other words, it should describe the place where we don't know what rules to apply or how to apply them — but it should not stray into argumentation. If you're ready to make arguments, you should give up on questions and start presenting answers. Don't ask people what the conventions are if you're ready to try and convince them what the conventions should be.

Answer 2

A good question need not contain elements. 'Why?' can be an excellent question, the Greatest Question. What is important is context, and outcome. The goodness of a question is not determined within the question, but by it's role in discourse with others or oneself, by how it helps provoke into advancing understanding.

We can look at the classic case of what happens when a child responds to every provisional answer with 'Why?', discussed here: "Why ask why" and its scions In this context, we understand a good question is part of situating ourselves, so as to be capable and effective in the circumstances we find ourselves in. They help elaborate the strange-loop we are constituted of, our extended consciousnesses, and so by choosing what to ask, we are involved in choosing how to be. That is, not by asserting who we are, by giving answers only, but by engagement in dialogue with the world, which chooses to make space for new ways to be, with what we question.

In Zen it is said the Great Koan is 'Who am I?'. In the Zen way this cannot be answered ever for all time, we answer each moment the question fills us, with all that we are. Only then, by looking, by being, is there ever any answer.

Hold an empty space, offer to the world an empty cup, and be in dialogue with how to be.

Answer 3

The inherent cause for questioning is to fill gaps or holes in our mental web or model. The existence of missing pieces, especially ones in areas important, brings anxiety and cognitive dissonance.

Two hurdles when asking a good question are:

• Finding a link between our mental gap and the recipient's mental model. If our respective models, or starting contexts, are too different, we may need to paint the recipient a new model before pointing out the hole in said model.

• Leading the recipient to the hole in question. Once we share enough internal context, we need to point out the right hole for them to know our concern. In the case they already filled this hole in the past, they can lend us the missing information. In the case we both have a gap of understanding in this area, we can work together in bridging the gap.

Put another way, step one is to find or create compatible context, or model overlap, in both minds; step two is to bring both our attentions to the same deficit, or same hole. Usually if the other already has the answer, this much will be clear to them once context is found. If both minds are incomplete, then the key is focusing on the same gap. For elusive problems, a back and forth may be needed to build the same peripheral context in both minds -- to reify and illuminate the cavity.

A question need not be an interrogative. Any context-building or focus-directing behaviour may suffice, as long as both minds see the same hole at the same time. For example, literally drawing or writing a partial model or thought process can shine upon the gap, sometimes enough that no sharing is necessary.

The overarching theme is context and focus. Once these are aligned, two minds become one.

On a related tangent, language transformers, or machine learning models capable of intelligent text prediction, operate on similar principles to the recipient described above. These carry forth user-provided context using probabilistic inference to fetch and create new textual output. Like walking up to a stranger, an untuned language transformer knows only the context you provide. To get the answer you want, you should provide the key background, along with hints of the focus you seek, or the gaps you want filled. The art of asking questions is analogous to the art of prompt-crafting, as they call it, which is the process of giving a language transformer the right context to get the desired output. If interested, I recommend playing with some. One option worth mentioning is GPT-NeoX, which can be found in the Model dropdown at the TextSynth Playground. A powerful paid option is GPT-3. Or you could search YouTube for videos of others playing with these. This task should help to give you a picture of working from nearly zero context in building a "question" that leads to a desirable answer.

Answer 4

Good questions open up interesting or important problematic fields where rigorous interventions can be performed.

---

"Imagine that I walked up to you and asked you this question without any context" -- it is interesting that a lot of Derrida's work is structured this way. Imagine someone says such-and-such, what do you imagine they would mean by that? How could we be sure? Even if they agreed with our interpretation, is that authoritative?

A good question takes into account the transcendental syntax of the real: in particular it acknowledges the reality prior to every question, which is to say, there is an audience, someone to hear and listen and understand. Without others, responding to and being responsible for other people, no questions are possible.

The question of the question is an old one: it is ultimately the question of existence, What is being? But the grammar of this question already seems suspect. For instance, can we imagine a being of time itself? Any question you can construct presupposes the existence of time, since time and the other come before every question.

What is a good question? Isn't that a matter of taste? Better than asking what something is, it can be more illuminating to inquire how it works; after all, what name is given and how it works may be very different. And assuming we are after strongly-adequate functional models, then we want to ask questions that open up problematic fields where rigorous interventions can be made so that we don't overextend conceptual resources.