How to capture all levels of difficulty

Question

I previously asked

Should CS.SE authorize any level of difficulty?

The answer is yes, but it feels hard to ask an easy question: we expect that all questions should have been the object of thorough research before being answered.

• That is good because the question become more precise and people can sometimes answer it themselves so they learn more.

• That is bad because of exactly the same reasons:

  • Very precise questions are harder to find with a search engine
  • Less easy questions are asked and less people will solve their problem

  (I like to think it will be like SO: if you got an honest problem, the answer will likely be here)

The case of homework is a little different: this may be a good idea to wait before giving a complete answer but I have the impression that having a complete answer is better in the long run. The answer of a question should be clear both for OP and any future visitor so only hints may not be enough. Also it is not our responsibility to prevent students from the temptation of cheating. The internet will grow without us anyway.

To sum up, I think that having a lot of easy questions is a good thing for the health of the site. (Feeling that it is easy to ask a question is welcoming. Feeling that it is easy to answer some questions helps keeping users. However easy questions might repel (busy) experts, but those were never here)

1. Do you agree?
2. How can we encourage easier questions?

I am seeking both your opinion and ideas to help promotion.

Answer 1

Yes, we do cater to all difficulties, but we won't allow bad questions.

There have been plenty of easy questions that were posed (relatively) well and have been answered, as far as I know without any snarkyness towards the asker. For example

• What is postorder traversal on this simple tree?
• Is there an undecidable finite language of finite words?
• Low-degree nodes in sparse graphs

There are, however, bad easy questions, too, for instance

• Construct an ε-NFA
• Construction of a regular expression

Those and similar questions have been closed with extreme prejudice, and rightly so. Why? They

• Copy assignments (apparently),
• Give no indication of effort on part of the asker and
• Are too specific to be of future use to others.

In order to deal with such questions indirectly, we have reference questions, such as

• How to prove that a language is not regular?
• How to prove that a language is not context-free?

We should have more of those if more easy and bad questions regarding specific things come up. The idea is that askers go there, try what is presented there and then come back to update their question with specific problems in their attempts (rather than "solve my assignment"), if there are any left.

There is some middle ground, that is "dump" style questions that are interesting or general enough to be of use, for instance

• Prove that regular languages are closed under the cycle operator
• Transforming an NFA into an NFA of similar size but without ϵ-transitions

I think these are not harmful for the site, and can remain and should be answered without reprisal. Commenting in an inquisitive way regarding attempts of the asker are fine and should probably be obligatory, if only to shape awareness of our preferences.