Given that we currently lack a description what is on-topic (Where is the definition of an on-topic question?) I think maybe we should make a list of things we think should be considered on-topic.

It would help for each example you provide if there can be a counter example of when a similar question strays into off-topic territory. I recommend a separate answer for each on-topic/off-topic pair so people can vote for the categories they agree with.

The examples needn't be questions that have actually been posted, or even theoretical sample question, a broad outline of a type of question will do to get things started.

I'm not suggesting this list be any sort of definitive statement of what questions are right for this site, but having a pool of ideas about what are good and bad question would go a long way to help in figuring out what should be on the help page: https://puzzling.stackexchange.com/help/on-topic

And as Emrakul pointed out, you might just want to define the sort of thing that is off-topic and should be avoided.

  • $\begingroup$ I think that only pairs of questions can help, a single question doesn't mean anything, everything older than a week and not closed is 99% on topic. $\endgroup$
    – leoll2
    Jun 26, 2015 at 14:04
  • $\begingroup$ I should have been more clear in my question. I'll update it. $\endgroup$
    – Bob
    Jun 26, 2015 at 14:05
  • $\begingroup$ Excellent! It's great that someone's started off this discussion. May it be long and productive :-) $\endgroup$ Jun 26, 2015 at 15:42
  • $\begingroup$ It's also worth noting that a lot of sites put things they don't want you to ask in this section, too - i.e. we might put "no math problems, please" there. $\endgroup$
    – user20
    Jun 26, 2015 at 20:38
  • 1
    $\begingroup$ I'm curious about the downvotes on this question. Eventually it will be important to put more topicality guidelines on the help center page. Is the thought that listing examples of on-topic and off-topic questions isn't a good way to start? If so, it might be nice to have a different meta post with the same goal of eventually coming up with something for the help center page. $\endgroup$ Jun 27, 2015 at 17:35
  • 2
    $\begingroup$ @JulianRosen Yes, I think the intention is good but that listing examples isn't effective. It makes sense for a Q&A site (hence the help center boilerplate), but there's too many types of puzzle, and a puzzle is judged on its quality as content rather than its relevance to a site topic. Centralizing our existing criteria onto a visible page is a good idea though. $\endgroup$
    – xnor
    Jun 28, 2015 at 9:03

4 Answers 4



is our top tag, so let's start there. The tag wiki description already has some good guidelines, which I've copied here (emphasis mine):

What makes a good riddle?

  • Riddles must be based on calculation and/or interpretation and not rely on assumption. The answer must be unique and clearly fit all the given clues. Riddles that have a large number of possible answers often get closed as "too broad".
  • Riddles must contain enough information and have an unambiguous, clearly correct answer. Hints should be optional: A riddle should be fully specified regardless of whether any user looks at the hint or not.
  • Riddles should be clever and creative. Riddles that are too easy or unimaginative are not fun to solve, although it is hard to draw a line here since difficulty is very subjective. In addition, riddles with crucial details omitted can frustrate solvers.

Examples of good riddles include this, this, and this.

Examples of bad riddles include this (too broad), this (too many hints needed), and this (many clues so obscure that it became almost a "guess what I'm thinking" puzzle).


Here's an obvious one to get things started....

Questions about how to create a puzzle

on-topic: How can I make this word search harder?

too-broad: How can I make a word search?

....if we can get a dozen or so of these then we will have something to boil down into description of an on-topic question for the help page.

  • 2
    $\begingroup$ I don't agree with the second one here. "How can I make a word search?" could be a very useful question and inspire a great answer about how to distribute the orientations of words, what sort of letters to put as fillers, whether to include red-herring almost-words, etc. In fact, how to make a good word search is something I'd be interested to learn - maybe I'll ask that question! :-) $\endgroup$ Jun 26, 2015 at 15:22
  • 1
    $\begingroup$ @randal'thor Maybe your right. If you can think of a too broad puzzle creation question go ahead and edit it in. $\endgroup$
    – Bob
    Jun 26, 2015 at 15:59
  • $\begingroup$ @randal'thor and if you ask that question I'll probably answer it. I've been working on word search building software and all those thoughts you had are on my to do list of features. $\endgroup$
    – Bob
    Jun 26, 2015 at 16:01
  • $\begingroup$ Here you go! I'll try to think of a too-broad puzzle creation question as well... $\endgroup$ Jun 26, 2015 at 16:05


is our third most popular tag, and the situation with this tag has been well summed up by xnor here. Quoting from his post ("s"s mine):

Math[s] puzzles are on topic, math[s] problems are not

So, what makes something a math[s] puzzle rather than math[s] problem? I think there's a few features.

  • Clever or elegant solution, often an "aha" moment
  • Unexpected problem statement.
  • Unexpected or counterintuitive result.

Multiple examples can also be found in xnor's answer.

Some (probably not all) maths problems would be on-topic at Maths.SE, but I can't answer for that community!


Codes and Ciphers

This post provides plenty of dos and don'ts:

Code Puzzles: What (Not) To Do?

Not sure that I agree with all of them, but in general pretty good advice.

Here's what I would recommend:

Good questions provide suitable clues to the encryption method and key. The question should provide guides to the correct direction in the process of solving. Any guessing required should be from a limited set of possibilities.

Bad questions expect the encryption method and/or key to be guessed from the many hundreds (or sometimes millions) or possibilities that exists. Or at the other end of the spectrum are simply encodings (Base64, Morse code, Binary ASCII, etc) and only test the ability to recognise the encoding.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .