2
$\begingroup$

My question was closed for lack of enough detail and no private feedback was provided as to what detail is exactly missing, which is not how I read it is supposed to be. Even though I cannot think of a single part that requires detail, I added information to the one part I could think would benefit and it got rejected for opening, again without further explanation. How can I request in such scenarios for more information regarding the closing reason?* I believe if someone has enough time to read my question twice and vote to close it, they have enough time to write one line explaining the reason.

*: in case it's not obvious, this is the question I want answer to here

$\endgroup$
3
  • 3
    $\begingroup$ Re your last sentence: I disagree here. We do get questions that are so unclear that it is easy to determine that they are unclear, but hard and time-consuming to write an explanation of what exactly is unclear. So people do vote to close as unclear, without leaving an explanation. ... $\endgroup$ Nov 12, 2023 at 10:01
  • 3
    $\begingroup$ ... Yes, this is frustrating to the OP. Please look at things from our perspective: we spend a lot of time here to help people, with no remuneration. And often when we close a question as unclear, the OP never comes back to make the requested clarifications - whether or not we explain what is unclear. Or conversely, especially for "unclear-unclear" questions, a long discussion ensues in the comments, which often enough leads nowhere. Again, I understand that you are frustrated, I am just trying to help you understand the other side. (I have not seen your question.) $\endgroup$ Nov 12, 2023 at 10:03
  • $\begingroup$ You didn't receive a message in your inbox stating that your question was closed (along with the reason like: 'needs details or clarity')? $\endgroup$ Nov 13, 2023 at 22:08

1 Answer 1

9
$\begingroup$

Writing a question in Meta asking for feedback is a way to solicit feedback.


The blue box at the top of your post reads "Closed. This question needs details or clarity. It is not currently accepting answers. Add details and clarify the problem being solved. This will help others answer the question. You can edit the question." This is the feedback.


I didn't vote to close the question, so I can't speak for the people who did. However, there are several parts of the question that could be clarified.

  1. The question is completely general. You want to compute an expectation of an arbitrary function where neither $P$ nor $f$ are described. Often, solving a very general problem is harder than solving a more specific problem; providing detail about what you want to do and how the proposed procedure fits into it could make the answer more tractable.

  2. In general, estimating posterior distributions is challenging. It becomes more challenging when estimating very many parameters and highly multimodal and non-identified likelihoods, such as neural networks. Bayesian inference tools exist to estimate posterior distributions, but it's not clear from the question whether you are using them at all, or if you are conflating them with SGD.

  3. You describe SGD as "sampling parameters," which is not what SGD is. SGD does not generate random deviates from a desired target distribution. SGD is an optimization algorithm. The purpose of SGD is to adjust parameters $\theta$ so that some loss achieves a local minimum. When the SGD loss is a negative likelihood, then it is analogous to a maximum likelihood estimator; both are point estimates of parameters given a dataset and a model specification.

  4. Bayesian neural networks are a specialized topic in their own right. It's possible that this is what you're after, since you're using a neural network and desire estimates of distributions over $\theta$. If you are asking about Bayesian neural networks, then this question is very broad and would benefit from an explanation of what you know and what you would like to find out.

It's possible that you have some specialized procedure in mind that resolves these inconsistencies and ambiguities. If so, you should edit your question to describe them.

$\endgroup$
4
  • $\begingroup$ Thank you! Since the feedback is characterized as "private" in the link, I expected something for detailed in my inbox or sth like this. Also, I do have response to your objections to my question but since this is supposed to be meta, I am not going to write them here. Clearly you misunderstood the question which does in fact show I wasn't able to communicate it well. But even if all your points hold, that would still not translate to a question with lack of detail. There's a misunderstood but clear question there that can be answered. Also, the main question in this post is still unanswered. $\endgroup$
    – Feri
    Nov 11, 2023 at 18:32
  • $\begingroup$ Some parts of that message are only visible to you. I don't have a way to test which ones. I'm sure some Q&A in the main SE meta explains how the closure message works. $\endgroup$
    – Sycorax Mod
    Nov 11, 2023 at 18:34
  • $\begingroup$ I see what you described. Nothing more. $\endgroup$
    – Feri
    Nov 11, 2023 at 18:37
  • 7
    $\begingroup$ The closure message says "This question needs details or clarity... Add details and clarify the problem being solved." It's disingenuous to suggest that the question was closed solely for a lack of detail when the closure message directly states it was closed for a "lack of detail or clarity." As I said, I wasn't involved in closing the question. Three other users, none of them moderators, voted to close it. So the problem is not that I specifically don't understand the question, but that at least 4 site users, myself included, don't understand the question. $\endgroup$
    – Sycorax Mod
    Nov 11, 2023 at 19:41

You must log in to answer this question.

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