10
$\begingroup$

#tldr; Apologies in advance for the length of this.

Periodically someone will either request/require an edit to a question or attempt to close it because it does not adhere to Stack Exchange standards. In particular, questions are nuked for seeking opinions or for having multiple parts. This has led to occasional testy exchanges.

My personal opinion (no pun intended) is that some questions that ask for opinions are perfectly fine. This would include things like "best practices" for cleaning data, whether constraint programming or integer programming is better for scheduling problems, etc. Granted there may be no definitive correct answers, but for example an OR noob looking to solve a scheduling problem might benefit from some experience-based guidance. As far as multiple parts go, trying to tie multiple single-part questions together using links would just create more work for people willing to answer them.

My guess is that these standards tie to SE starting out as a place to ask coding questions. Questions about how to do something in code may be by nature easier to break down into simple "one bullet" queries. Also, my recollection of the early days of online forums includes "flame wars", and the no-opinion-questions policy might very well be SE's version of fire retardant.

So my question is: do we want to be rigid about opinion questions and multi-part questions, or whether we want to be more relaxed than SE seems to be in general? Can we reach some sort of consensus?

$\endgroup$
  • $\begingroup$ Possibly relevant to OP's question: this comment and this Q/A. $\endgroup$ – SecretAgentMan Nov 11 '20 at 16:54
  • $\begingroup$ Two issues I see with the question as worded: (1) the use of the word "anal" may be seen as a loaded term by some and that it inherently places a judgement on one side; (2) your question rightly addresses "best practices" and approach selection, but does not address questions seeking predictive options (example). $\endgroup$ – SecretAgentMan Nov 11 '20 at 16:59
  • $\begingroup$ +10 internet points for "My personal opinion (no pun intended)..." $\endgroup$ – SecretAgentMan Nov 11 '20 at 17:00
  • 1
    $\begingroup$ @MarkL.Stone your comment may be seen out of context based on the latest revision. Just a friendly heads up. $\endgroup$ – SecretAgentMan Nov 11 '20 at 17:01
  • 2
    $\begingroup$ I vote no for riigid and overmoderating, yes for liberal interpretation of what is on-topic. If I wanted to have strict rules for everything, with only one correct way of doing things, with little to no discretion, I would have become an actuary. $\endgroup$ – Mark L. Stone Nov 11 '20 at 17:02
  • 2
    $\begingroup$ @SecretAgentMan While I do come down on one side of this (the more relaxed approach), I changed "anal" to rigid so as not to offend anyone (and get side-tracked). As far as questions seeking predictions, I personally have no problem with them. I'm more concerned about students asking us to do their homework than about people asking for predictions. $\endgroup$ – prubin Nov 11 '20 at 17:03
  • 1
    $\begingroup$ Man do I agree with this by @prubin "As far as questions seeking predictions, I personally have no problem with them. I'm more concerned about students asking us to do their homework than about people asking for predictions. " $\endgroup$ – Mark L. Stone Nov 11 '20 at 17:06
  • $\begingroup$ To be clear, I also share @prubin's and @Mark L. Stone's concerns with low quality and/or homework style questions that do not follow OR.SE's help center guidelines (though that is a different discussion). Also, possibly related. $\endgroup$ – SecretAgentMan Nov 11 '20 at 17:17
  • 4
    $\begingroup$ I think it would be a shame if we cut out these questions around e.g. best practices, tips&tricks etc. The practical aspect is an essential part of operations research in my opinion. We can maybe get some inspiration from: workplace.stackexchange.com and academia.stackexchange.com where there also are similar kind of questions. $\endgroup$ – Michael Lindahl Nov 12 '20 at 8:59
5
$\begingroup$

I would like to leave it at:

Let's not be too strict and have guidelines keep us from having useful Q&A.

But surely that's not convincing (and for reference if something like this comes up again), so:

Subjective content

There are more or less uninformed opinions, guess-work, and wishful thinking. And then there are evidence-based extrapolations, (possibly opposing, which makes great answers) expert expectations/opinions, industry consensus...

As an example, because it has been mentioned, this question is perfectly fine. It asks for extrapolations from experts. The first 3 question are closely related and ask if a naive extrapolation of the past is reasonable, and if not which direction is more likely. The fourth question asks through which technologies it will be achieved. In my book, that's an answerable, well-defined question. Whether it can be considered a single question is somewhat irrelevant. Let's be tolerant with these kinds of "opinion" questions.

I don't think drawing inspiration from workplace or academia is too helpful, as those are far more subjective and situation-dependent than the opinionated questions on here. Other opinionated sites: parenting, project management, politics... The closest to our case that I can think of is statistics ("Which statistical test to use on this data/to test for this hypothesis?").

Opinionated questions

OR is pretty rigorous anyways (compared to software development, meaning SO). There is already a lot of non-opinionated treatment of OR topics online in the form of papers. Additional, distilled and condensed collections of interesting topics here on SE is a great supplement.

Opinionated treatment of OR topics is pretty rare, but immensely helpful. It can be found on many of the great OR blogs. Sometimes there are opposing opinions/interpretations/insights on the same topic on different blogs. Even more helpful.

In moderation, let's allow for that to have a bigger platform than just blogs. The same goes for questions asking for evidence-based extrapolations, expert opinions, and industry consensus. I don't say let's actively encourage it and get flooded with those, but think twice before voting to close.

Multi-part questions

The same goes for multi-part questions. Let's think twice and maybe let's put some of our editorial effort - if necessary, I don't even think it is necessary here - into suggesting to the authors to frame their question as a single overarching question and specify which topics they would like to focus on.

I do think the question that sparked this can be changed to better meet SE guidelines. Although, I'm not convinced that it improves the question.

To be specific, from

Title: How much can we expect to increase the speed of mixed integer programming in the next 10 years?

  • What speed-up is it most likely we will see in the next 10 years?
  • What are the reasons for a lower rate in the future than what we have seen in the last 30 years?
  • What are the reasons for a higher rate in the future than what we have seen in the last 30 years?
  • Which trends and technologies will have an impact on this?

to

Title: How much can we expect to increase the speed of mixed integer programming in the next 10 years?

Question: What speed-up we will most likely see in the next 10 years? To narrow it down, please focus on

  • the reasons for a lower rate in the future than what we have seen in the last 30 years,
  • the reasons for a higher rate in the future than what we have seen in the last 30 years,
  • the trends and technologies that will have an impact on this.

What good does this do? Straight up editing questions just in order to not have more than one question mark in the question does not seem like the way to go.

Opinions welcome.

$\endgroup$
  • 1
    $\begingroup$ Thanks for your comment. I think it is a fair comment to reformulate the question into one questions with sub-questions. That was also the original intention with the sub-questions. To ensure that answers gave additional context and reasoning behind the answer. $\endgroup$ – Michael Lindahl Nov 16 '20 at 7:48
4
$\begingroup$

I don't think there is a single answer to the question "how closely should we stick to StackExchange norms?" Each norm might have a different answer.

I can think of four norms that we have grappled with so far:

Opinions: It seems clear to me that the consensus of the OR.SE community is that questions seeking opinions are acceptable. @prubin and @ktnr make the point that we should allow questions seeking expert opinions ("is MIP or CP better for scheduling problems?") and discourage those asking for strictly subjective opinions ("is EJOR a better journal than NRL?"). I agree with this point.

Multi-part questions: I think there is a distinction here, too, based on how closely the sub-questions are related. I'd probably be OK with

  1. If the primal is unbounded, is the dual infeasible?
  2. If the primal is infeasible, is the dual unbounded?

But these two would be better posed as separate questions:

  1. If the primal is unbounded, is the dual infeasible?
  2. How is the Lagrangian dual similar to the LP dual?

Pleasantries: SE norms discourage "thank you" and other pleasantries, and encourage edits to remove such pleasantries. There doesn't seem to be a consensus yet on OR.SE about how we feel about this norm. You can weigh in on this one here.

Big lists: SE norms discourage "big list" questions (like, "what are some good books on [XYZ topic]?"). OR.SE has mostly welcomed them, e.g., this post and this post.


To me, it is essential that we disagree about these issues in a respectful and constructive way. Our community has a diverse set of opinions about these norms, and a given individual's opinions might change over time. (My own opinion has moved from "rather strict" to "much less so" over the past year.)

It's fine for a user to post a comment noting that a given post violates SE norms. It's also fine for a user to reply, arguing that we should allow that norm to be violated. Local issues can be resolved in comments and by voting. More global issues can be discussed here on Meta (as the OP has done). But no one should be faulted for voicing their opinions or for voting them, as long as we remain respectful and constructive.

$\endgroup$
3
$\begingroup$

I don't have much of an opinion of how and where you should stick to the norms of Stack Exchange (or specific sites on SE), mostly because it has been while since I've been active here.

But if you (as a community) have decided what you want questions that go somewhat against the main norms and expectations of the SE platform (from what I remember, I think a reasonable number of users in this community do want that), I have the following suggestion:

Use a special tag to indicate which questions go against the norms of the platform

For an example of how this could work, take the "soft-question" tag on Mathoverflow, Math.SE or even cstheory.SE. Observe that the precise rules around this tag differ. Do note that it is important that the tag is has broad support in your community, if the existence of the tag is only supported by a tiny cabal, it will do no good.

I think there are several advantages to this. First of all, a tag filled with highly voted questions of a certain type sends a clear signal that such questions are accepted here. Second, you can tell all the haters users who do not like such questions that they can improve their experience on this site by simply avoiding questions in that particular tag. Third, if you have any special rules or recommendations for handling those questions, users can now learn them by reading the tag description.

I also think there are a few possible issues. You may think that all this is a lot of pointless meaningless work and don't want to bother with this. From what I remember, there were quite a few users that already were doing extensive tag maintenance. If they're still around, I don't think they'd mind adding yet another tag. Another disadvantage is that there is now one less spot for tags describing the topic of the question, but I don't think this is a big loss.

A more serious problem is that now people have the opportunity to argue which questions should get this special tag. I still think this is better and will happen less often than the alternative where people argue the question should not be posted at all (or should be closed, etc.), but I'm not super sure. You'll have to see what happens in practice to be sure, I'm afraid.

But wait, isn't this a "meta-tag"?

First, read this answer, it also gives a few more examples of "special tags".

Second, well... If I may paraphrase an infamous man of the sea: most conventions and "rules", especially those made a long ago time (long before OR.SE was a thing. Long before even the thing you used to create OR.SE was a thing), are more of what you'd call "guidelines" than actual rules. Almost every time, a meta-tag is a bad idea. Which is why the "rule" arises. This is one of the reasons I have written this rather long post. It is an attempt to convince you this is not one of those times.

$\endgroup$
  • $\begingroup$ Thanks Discrete lizard. I was unaware of meta-tags, and I find this intriguing. There's a discussion of some of these tags at math.meta.stackexchange.com/questions/143/…. Despite the objection posted there, I think I prefer "subjective" as a tag rather than "soft-question" (which sounds to me a bit pejorative). What think others? $\endgroup$ – prubin Nov 18 '20 at 0:37
  • $\begingroup$ The "subjective" tag sounds good. $\endgroup$ – ktnr Nov 20 '20 at 11:56
  • $\begingroup$ @prubin I think the "soft-question" is supposed to be a bit pejorative on MO, because those questions are regarded to be "second-class". For example, votes on those questions should not(?) give users any reputation on MO. I don't think you want to treat the questions as second-class, so a different name seems a good idea. "subjective" seems good, maybe "experience-based" might be better if you want the answers to be based on the experience of experts, rather than arbitrary opinion. In general, I think it is best to make these tags as specific as you can afford to. $\endgroup$ – Discrete lizard Nov 20 '20 at 13:10
  • $\begingroup$ @Discretelizard I think this is an important point and maybe deserves its own Meta post asking whether we should create such a tag. Feel like posting it? $\endgroup$ – LarrySnyder610 Nov 20 '20 at 17:02
  • 1
    $\begingroup$ @LarrySnyder610 To be honest, no, not really. Not anytime soon, at the least. However, if you or anyone else wants to do that, feel free to copy as much from my answer as you wish. $\endgroup$ – Discrete lizard Nov 20 '20 at 17:17
  • $\begingroup$ @LarrySnyder610 I just ginned up a question about it: or.meta.stackexchange.com/questions/316/…. $\endgroup$ – prubin Nov 21 '20 at 18:55

You must log in to answer this question.

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