I posted an unclear question that I later tidied up, but now commenters are doubtful that the accepted answer (which was edited when I clarified my post) is correct.

I posted a screenshot to prove it, but is that normally needed? Is my question unclear?

  • I don't see the reason to bring this up to Meta. I subscribe all of what @fixer1234 said and your latest edit it's just a image "proving" that the accepted question worked.
    – CaldeiraG
    Jul 29 '19 at 10:15
  • I'm not sure what you're saying, @CaldeiraG. Fixer1234 says the accepted answer "doesn't do what you've now clarified as the problem", whereas my image proves it does. I do not see how you can say you subscribe to both...? Apologies if I've misunderstood you. Jul 29 '19 at 10:24
  • You don't need to prove anything: you received a solution that does what you need by accepting it but that doesn't reflect the original question (aka is unclear). I think you should change this meta question to something like: "How can I improve this question?" if you're looking for advice on how it's unclear.
    – CaldeiraG
    Jul 29 '19 at 12:15
  • @CaldeiraG I don't think my original question is unclear though. I've described my problem, and I've given example input and output. Jul 29 '19 at 12:32
  • @Pureferret - Why do you care that another user does not believe the accepted answer is correct? If you found the answer the most helpful, and it answers the question you had, you should accept it. As the author of a question you can accept any answer you want (which isn't deleted).
    – Ramhound
    Jul 30 '19 at 12:15
  • @Ramhound because it then lessens its value to the community if they are casting doubt on it's veracity. What use is a useful question no one reads because it's downvoted? Jul 30 '19 at 12:19
  • 2
    The accepted answer did not receive any downvotes. The question received downvotes after you accepted the answer. Since you don't know who downvoted your question, or the reason they did so, there isn't much you can do about it. As for the comments that lessen the value of the answer to the community, if they are inappropriate it, report the comments to be deleted.
    – Ramhound
    Jul 30 '19 at 12:25

I'll focus first on your question here and then on the question on the main site that prompted it.

Meta Question

Your question here asks if you need to prove an accepted answer works, and then paraphrases it to ask whether your question is unclear. Those actually are different questions.

No, you don't need to prove an answer works. You are free to select any answer, even gibberish, if you think it helps or solves your problem. Acceptance of an answer is totally at your discretion, and you don't need to justify your decision.

An answer doesn't necessarily have a relationship to whether a question is clear. It's common for a question to be ambiguous, various people will answer based on different interpretations, and one of the answers will address what the OP intended. That answer gets accepted. The question is still unclear, but the answer helps shed light on what the OP meant. Based on the answer and acceptance, someone can often improve the question to clarify it.

But that can also work in the other direction. If an accepted answer doesn't appear to others to be a solution, it makes the question unclear, because it implies that the question means something other than what it says.

Those are your two questions, but there's a third aspect. The question doesn't exist in a vacuum. The purpose of the site is to be a knowledge base of solutions for other people. It's great that you got a solution for your own problem, but what about other people looking for a solution to their problems. They will land on your question through a search (that's where the great majority of readers come from). If your question isn't clear, it isn't fulfilling the site's purpose.

Main Site Question

The question is one that got both of us frustrated. The screenshot you added was actually helpful, although the appropriate place for it would have been as part of JvdV's answer, not the question.

For my own part, I was focused on the ambiguity, and the endless possible variations that aren't clarified in the question. It's unfortunately too common for questioners to not realize that their question can be interpreted in ways other than what's in their own head.

People volunteering their time to help, waste time developing an answer that's a legitimate response to the wording in the question, only to be told that it wasn't what the OP meant, or doesn't account for some condition not described in the question. In fact, that happened on this question. Rajesh S developed a solution that was legitimate for what was in the question at the time. Then you added other examples and disclosed that the data was dynamic, which invalidated his efforts (plus he got a downvote for his trouble).

To be honest, I missed that JvdV's revised answer did address the additional examples you added (specifically "Two Tokens"), at least as long as there isn't a trailing space. There are still ambiguities in the question, though. For example, can you have a word ending in a number, like "This is example1"? JvdV's answer assumes that if the last word ends in a number, it's entirely a number. It is adequate for your personal need, but it isn't a robust solution that handles other variations that would be covered by the description in the question. Other people trying to use the solution for their similar problem could get wrong results.

So my last comment was inaccurate. The accepted answer does, indeed, address the specific examples listed in the question.

I'm happy that you got a solution for yourself, but I still think the question kinda sucks for other people with a similar problem, and even other people who might want to offer another solution. The issue is that it doesn't actually define the pattern you're looking for. It gives some examples, and they may cover all the relevant cases in your data. But defining a problem only by a few examples leaves it very ambiguous in terms of the data other readers may have. It also still leaves open the potential for another answerer to waste time on a solution that doesn't meet a requirement you haven't thought to mention. And those are problems given the purpose of the site.

The question would be much clearer if you start with some context. As-is, all we have is some invented data that you think covers the range of conditions in your data. To readers, it's just a random collection of nonsense terms that could be placeholders for anything. It's like a guessing game where people have to keep asking for clues until the problem has been fully defined (and there's no basis to know when it is). If you describe what your data actually represent and what it comes from, that puts some boundaries on what it could contain, and what a solution needs to handle.

You must log in to answer this question.

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