21
$\begingroup$

I have spent a lot of time on StackOverflow and am pretty familiar with the types of questions that are and are not allowed on the site. I am a little unsure as to how well the strict question rules transfer over to this site, however. Typically, questions that do not have concrete answers or are opinion-based are discouraged, as are requests for listing of resources. I don't think that this level of stringency is appropriate for this site. An example:

What is the best tool for computing gene-level expression?

This might immediately be flagged for deletion because it is both a resource request and opinion based, but keep in mind:

  • It can be difficult to find scientific tools, especially if the analysis or function they perform is not very common or new.
  • While people might have opinions on what tool is the best, there are quantifiable ways of demonstrating that some scientific tools are better or worse, which can be important to capture an upvote for future reference.
  • There are a lot of unknowns in science, and therefore a lot of opinions, so one could argue a lot of things as opinion-based. Therefore, some leniency should be expected in question asking.

Thoughts?

| |
$\endgroup$
  • 2
    $\begingroup$ Another point could be that some communities might have some general form of agreement on the most appropriate tool for a specific analysis (i.e. blast for simple homology search), despite having many other alternatives. $\endgroup$ – mgalardini May 16 '17 at 20:15
  • 4
    $\begingroup$ While I agree in principle with your bullets, I think that the appropriateness of the example question you posted would depend a lot on the body of the question. If it was just that heading, it's too vague, but if the questioner lays out specific features, there might actually be an objective answer. $\endgroup$ – kevbonham May 16 '17 at 20:37
  • 2
    $\begingroup$ And there's another thing to consider - answers to questions like this will fluctuate wildly over time. We'll need to have pretty strong norms around modifying previous answers or marking old questions as out of date. $\endgroup$ – kevbonham May 16 '17 at 20:39
  • 1
    $\begingroup$ @kevbonham: this is something that occurred to me as well. Just because an answer was the best possible solution at a given time doesn't mean it will even be close to correct 5 years from now. $\endgroup$ – woemler May 16 '17 at 20:43
  • 1
    $\begingroup$ @woemler right. This would be an issue on SO as well, but most of the questions I see there are referring to specific languages/versions. Are there questions like "what's the best framework for designing a website?" on SO? How do they handle that? $\endgroup$ – kevbonham May 16 '17 at 20:46
  • 1
    $\begingroup$ @kevbonham that answer simply wouldn't be accepted on SO, I don't think. On the other hand, if you said "I'm interested in creating a website with these features that runs like this..." then perhaps it would be okay. $\endgroup$ – Scott Gigante May 17 '17 at 6:56
  • 3
    $\begingroup$ I suggest that bioinformatics 'experts' should ask what they consider to be good questions and provide their own good answers to get this site moving in the right direction! This practice is encouraged on other sites like SO stackoverflow.com/help/self-answer $\endgroup$ – Chris_Rands May 17 '17 at 8:43
10
$\begingroup$

I also don't think that "What is the best foo" can ever be a good question. I would instead suggest something like:

Is there a comprehensive review and benchmark of the various tools for FOO? What tools had the best results?

Something like that can be answered by citing a specific review paper and the measurements there. In any case, the "best" usually depends on what you're doing. For example, while toolA might be the best for computing expression in speciesA or tissueA, toolB might fare better in another species/tissue.

So no, let's not allow "what is the best foo" questions, but instead try and make them as specific as possible while still being general enough to be worth asking.

| |
$\endgroup$
  • 1
    $\begingroup$ Great answer. This would be a very good way to frame this sort of question. $\endgroup$ – woemler May 20 '17 at 0:35
7
$\begingroup$

Such open-ended question should be closed in my opinion. A better question for the site would be

What is better tool for computing gene-level expression A or B?

This question is still open to opinionated personal experience using the tools but can be answered with articles comparing both tools.

This way we can have a tag for and group there questions comparing . The reader after reading the questions and articles might end up with a good knowledge of the tools available and used by the community.

As suggested in the comments:

What are the pros/cons of computing gene level expression with A or B?

This might be a better alternative

| |
$\endgroup$
  • 2
    $\begingroup$ Perhaps even better: What are the pros/cons of computing gene level expression with A or B? This moves it slightly further away from opinion and towards objective fact. And is probably a better way to think about tools anyway. $\endgroup$ – kevbonham May 17 '17 at 13:15
  • 1
    $\begingroup$ But what about cases where there is an objectively better tool, C, that is not specifically part of the question? $\endgroup$ – woemler May 17 '17 at 14:06
  • $\begingroup$ @woemier If a better tool C exists, you can suggest so in comments either in the question or in any answer you want, and then the next question would be "pros/cons of C vs A?" and "pros/cons of C vs B?". $\endgroup$ – llrs May 17 '17 at 14:09
  • 1
    $\begingroup$ @Llopis agreed. I've expanded my comment into an answer below. $\endgroup$ – kevbonham May 17 '17 at 14:35
3
$\begingroup$

I have a proposal - tool questions/answers should be limited to discussions of pros/cons for a particular tool or tools. In other words, the question you posed (and bouncing off of Llopis' answer):

What are the pros/cons of computing gene level expression with A vs B?

This approach has a number of benefits:

  1. Answers to this question can be more objective. People may still quibble about whether a particular approach is better or worse, but the upsides/downsides of any approach shouldn't generate as much disagreement
  2. This is probably a better way to think about such questions anyway, and answers to such questions are more generalizable to different experiments/approaches. In other words, even if a questioner asks about a particular experimental setup, good answers about the pros/cons of a particular tool will likely apply to other experimental setups.
  3. These answers are less likely to go out of date, since new software will not change the pros/cons of particular approaches, but might change what the best approach is. And even old questions about out of date software will retain value for learners (since they will contain expert thought about how to assess the value of an approach).

I'm not sure if this approach is applicable to all questions of this type, but it could certainly be applied to most. It will take more effort from questioners, since they will have to do some research to actually bring up a tool or tools, but this will probably make for better questions anyway, and there's no reason an answer couldn't bring up some new tool that wasn't asked.

| |
$\endgroup$

You must log in to answer this question.

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