Please refrain from closing proper but imperfect questions so quickly

Question

I have been thinking about this for a while. This recently on-hold question really touches my nerve. The question is put on hold due to being too "broad". I will use it as an example to explain my concerns.

First, "broad" (as well as "off-topic" etc) is subjective. In this particular case, I think the question is on the border line at best. I certainly would not vote to close it, even though I am actually quite picky. To most new NGS users, they would think this is a fine question, too.

Second, even if "broad", this question is interesting and useful. It has attracted 4 upvotes in 13 hours, which is relatively fast in this new SE. The answer tells me something new and equally interesting. It has got 5 upvotes, too.

Third, broad questions may inspire epic answers. For this question, someone familiar with QC might take the chance to give a succinct but comprehensive review of existing QC tools and measurements, something you occasionally see at SO or other SE sites. This would be invaluable to everyone. You can't get such an answer from a specific question on a particular technical problem with a QC tool.

Fourth, at this particular time we are struggling to pass the private beta phase, I speculate that closed questions are not counted. @RobertCartaino said we had less than 75 questions a couple of days ago. However, as @gringer pointed out, we had 77 questions almost at the same time. In addition, on the question page, we have 102 questions right now, but on the Area51 page, we only have (15+1)*4.9=78.4 questions. Why the difference?

Finally, even if not just for private beta, we need both experts AND new users to keep this site alive. To me, the questions here are a little too "narrow". They focus too much on specific technical issues. Questions slightly deviating from this focus quickly got downvoted or closed. Note that we are competing with BioStar. Although criticized for all the problems, its broadness and friendliness to newcomers are undeniable. If we keep closing border-line questions so aggressively, we would push new users to biostar and finally kill this SE, again.

So, please refrain from closing proper but imperfect questions so quickly.

EDIT: here is a list of questions that are currently closed or put on hold.

• Influential papers. I wouldn't vote to close it, but I can understand why others want.

• Bioinformatics vs biostatistics vs computational biology. Hmm... same comment as above.

• Stable download URL. A little general, but I think it touches a real problem. I have voted to reopen it.

• QC for NGS. This is the question I was mentioning at the beginning. It is a proper question.

• Smalt. This is a good bioinformatics question and should not get closed.

• Public multi-omics dataset. This is a good and relevant question. Many newcomers will have the same question. The answer has got 6 upvotes.

• Constructing gene network. I am not familiar with the topic enough to tell.

• NGS metadata schema. OP was essentially asking what relationships are common to general schemas. This is a proper question and should not be closed. Even if there were no general solutions, this would not make the question itself non-specific.

• Scripting parallel processing. Closed due to off-topic. I don't think it is bad enough to merit a closure. @DevonRyan's answer is informative. Note that OP is a highly experienced SO user. He/she did not show up after this question.

• On bootstrapping. Closed due to off-topic. OP has really put a lot of efforts into the question. The answers are good, too. I don't think it is that off-topic, either. We use bootstrapping often. Again, we pushed a potential new user away.

• Genome vs transcriptome. Closed due to off-topic. New bioinformatics students with a CS background are likely to have the same question. I don't see we should close it. Again, OP was asking only one question and then gone.

• Difference between common formats. Closed due to too general. I don't see the point: a basic question may look general to experts, but it is often a good one to many people. The answers are very helpful.

Answer 1

My only caution is to urge you not to trade in good moderation practices for the sake of holding onto more questions. That will reflect poorly on a site more so than having a few less questions — but otherwise a good balance to consider.

Answer 2

This is an excellent point. The goal of a private beta is to seed the site with some questions, answers, and members. Everyone here is interested in having this site succeed. Let's not ruin those chances. StackExchange wants to see activity. If you really hate the question, downvote it, so at least everyone else has a chance to see it. Don't close it unless it's clearly not relevant. Once the site is approved and there is a healthy community, we can be more selective. Treat this baby gently for now.

Keep in mind, one of the top SO questions is: "How to exit the Vim editor?". This could have been closed because it's off-topic (not programming).

Answer 3

Please don't confuse "well defined scope" with "unfriendly". We do not compete with BioStars, they allow discussion and have a far more lenient scope. The success of the SE model is precisely that we only deal with specific questions that have an actual correct answer. Closing things that don't fit the SE model is essential to developing a healthy site.

If you want such broad questions as "What is the best foo for bar?" then you should absolutely go ask them on SeqAnswers or BioStars where they would be welcome. We are not in direct competition with those sites; we aim to build an SE-style site here, so no chit-chat, strict scope and exclusively questions and answers.

So no lists ("what is your favorite foo?"), no polls ("Who else thinks that bar is better than foo?") etc. While I agree that we might want to be a little bit more lenient during the beta phase, we can't do so at the expense of quality. And the notion that closing questions is somehow " unfriendly" is one we should root out from the very beginning. It isn't unfriendly, just like downvotes aren't mean. These are the only tools we have to keep the site focused, on topic and with a high signal to noise ratio.

Now, of course broad and off-topic are subjective. But that's what this very meta is for: here is where we will hash out the details of our scope. And that's also why it takes 5 people to close a post: it's subjective, so we need consensus. Not closing is not the answer.

In addition, please note that "closing" isn't actually closing. It is putting a question "on hold". As soon as the question poster (the OP) or anyone else edits a recently closed question, that question is put into the "reopen" queue where users can vote to reopen it. If three do, the question is reopened and no harm done.

So, putting questions on hold is a very useful way of telling the OP to please fix the issues with their question to make it fit the site better. It is absolutely not a one way street, and can be reversed relatively easily. Especially in a small community like this one.

---

That said, here's my personal take on the examples you listed:

• Influential Papers: this is the very definition of a question that is a bad fit for SE. it is asking for a long list, there is no "right" answer, it will be awful to try and maintain it. This sort of open-ended list request is explicitly off topic on any site of the SE network.

• Bioinformatics vs biostatistics vs computational biology: the distinction between these terms (or rather between the first and last, the middle is quite different) is very much a question of personal opinion. There aren't really very strict definitions and everyone sort of uses them as they like. There are no sources we can cite to support our "definition". I don't see how this could be anything other than "my opinion vs yours".

• Stable download URL: This isn't "a little general", it is incredibly broad. Even if such stable URLs existed, what would an answer be? Should it have a list of the stable URLs for all NCBI databases, EnsEMbl, UniProt, Gene Ontology, STRING, TGI, UCSC . . . Database N? There are dozens if not hundreds of such databases around. If the question were edited to make it about one specifically, that would be fine but as it is, it is basically looking for a discussion of the (very real) problem of changing URLs. That would be a great topic for the Bioinformatics Chat, but it isn't a good question for the main site.

• QC for NGS.: I am afraid I strongly disagree with you on this one. That is a very, very broad question. If it were asking for the differences between specific QC tools, that would be answerable. But "what else is there" isn't really answerable. How can there be one good answer? The problem can already be seen in the existing answers. There are currently three answers, each of which is suggesting another tool. The answers are all variations of "I like this one".

  Ideally, questions on SE should have one correct answer. How can the OP accept one of the ones they've received? How is one more correct than any others? This would be a great question for a site designed to allow discussion. This isn't such a site though, this is strictly Q&A.

• SMALT: The question has been edited so that it no longer asks why SMALT is "so popular" for bacterial data and is simply asking for what features make it different or particularly well suited to bacterial genomes. That's great! This is indeed answerable with a single, correct answer. The original was asking why it was "the most popular" and that isn't really answerable without opinion.

• Public multi-omics dataset.: I voted to close this one as too broad asking the OP to provide more information. The question was indeed edited and made more specific and it was reopened. Yay!

• Constructing gene network.: this is the clearest case of the lot. The OP wanted to construct a network of 100000 different proteins and their interactions. This is obviously impossible since there is no species with so many proteins (or anywhere near so many). A discussion in the comments clarified that the OP is working on something that has nothing to do with biology and just mistakenly thought that biology could give a good dataset for what they want to do. So the question isn't even about bioinformatics, really.

• NGS metadata schema: This one I would agree is fine and wouldn't have closed it either.

• Scripting parallel processing.: agreed. I think we should allow such general computing questions when they are the sort of thing that a bioinformatician can encounter in their daily work.

• Bootstrapping: I can understand this closure. The term isn't from bioinformatics nor specific to it. It comes from statistics and is in fact a common expression in English (although obviously in a different context). Still, it is something that is relevant to bioinformatics so personally I would consider it on topic.

• Genome vs transcriptome : In this case, one could argue that the question shows a lack of research (although it is hard for us "experts" to gauge how hard it would be for a layman to find this). Lack of research, however, is a good reason to downvote, not to close. The question itself seems perfectly on topic.

• Difference between common formats: this one seems like it should be on topic, agreed. I don't think it's a particularly useful question seeing as the three formats are used for very different things but, again, that might be because I am too used to them. No reason to close, anyway.