Asking open ended questions while still complying to SE guidelines

Question

Proposal: Computer Graphics

A lot of potentially very interesting questions are closed as subjective or out of topic on Stack Exchange, including some popular "historical" ones. This trend has spawn quite some discussion. The guidelines state for example that:

Thus, questions that are not answerable — discussions, debates, opinions — should be closed as subjective.

Computer graphics are full of problems with no clear definitive best answer. I understand from the guidelines that questions about such problems don't belong here and should be discussed somewhere else. But if this means going to some other place every time a problem is open ended, it feels like a loss both for the user and this Q&A website.

How can we prevent some entire topics from excluded?

The question "How can glossy reflections be implemented efficiently when using Image Based Lighting?" (from the Example Questions) sounds to me both valid yet at risk of getting moderated away under such rules.

Answer 1

If they aren't too broad, IMO questions like this about open problems, or what technique to achieve a desired effect, etc. can be answered quite well within the guidelines. For instance, we can summarize the state of the art, compare high-level advantages and drawbacks of different techniques, and give pointers to references for further reading, etc.

If the questioner provides some more relevant context (e.g. target hardware/API, what kinds of scenes/materials they have, what sort of look they're going for, etc.) then we may be able to recommend a specific technique, or give ideas about how something can be specialized for their use, etc.

IMO, the guidelines about subjective questions don't mean that we can only have questions where there is a single definitively correct answer (which would be rare). Indeed, having multiple answers recommend different approaches is a sign of a good question (and a healthy community). The guidelines are meant to exclude questions that just result in a lot of yik-yak and don't come to any useful conclusions, like "is DirectX better than OpenGL" kind of things.

Answer 2

I believe that this site should hold itself to the same standards as any SE site. But don't forget that the example questions are merely titles - there is only so much specificity that can fit in a title. Whether the question you mentioned is ultimately a good question still depends a lot on the question.

I have left a comment along those lines on the example question How to write a GLSL bloom pixel shader?

I think, whether this is an appropriate question depends really on the question body. If the question is literally just "How do I write a bloom shader?" I think it's not a good fit (for SE in general). If it's "I'm trying to write a bloom shader, but can't get it to work. Here is what I've got." then it's a great question. I believe this site should set the same standards for programming questions as SO does, and one of the really big ones is that questions should be proofed against "What have you tried?".

I don't think computer graphics is inherently more open-ended than any other programming. That won't stop people from asking vague questions, but that just means we as a community need to control the quality of the site as much as any SE.

(This is also why I'm not very pleased with example questions like this ending up in the questions that reached 10 votes.)

Answer 3

I have nothing against broad questions as long as a broad answer is desired. I've seen many good questions on SO closed because of this (and in many cases with good reasons, because a specific answer was desired). I think the intent behind the rule is to stop a small, maybe uneducated, question requiring a whole book to answer. If it's clear what the OP is asking for but the question is open ended, a good answer would explicitly acknowledge this and maybe provide a brief overview of the topic, clearing up misconceptions and pointing to further information.

In short, broad answers for broad questions (provided it's still a "good" broad question).

For example in How to write a GLSL bloom pixel shader? if the body was minimal as suggested by @MartinBüttner. Rather than provide source code for an implementation, the answer might list a couple of general approaches. Just enough info to start searching for the right things.

What mathematics do I need to study in order to program graphics? is another good example, which could easily be dismissed. A long answer is possible, but makes it outside the scope of a Q/A site. Instead the answer could simply point out that its a broad question but list some initial mathematical problems a new graphics programmer is likely to encounter.

I think it'd be great to have community wiki posts for open ended questions where the answer keeps evolving as new work gets published, as in the question's example, How can glossy reflections be implemented efficiently when using Image Based Lighting?.

To answer the question, maybe the text in the close-question dialogue could be altered to address the concerns here and in other answers.