11
$\begingroup$

I ask motivated in particular by this question, but also by the general pattern seen so far.

That question was originally titled Should generators be their own type?, but was now renamed to What are the pros and cons of having a unique type for generators?, in line with a number of others on the site. I think the former was a strictly better title for the question! Yes, it's technically a yes-or-no question, but a one-word answer that provides no rationale will be a bad answer regardless. A good answer to that question will propose a solution and present the supporting case. It's a fine question, and answers will be informed by evidence.

On the other hand, the memetic "What are the pros and cons of ..." format seems to have taken off merely to avoid the appearance of being opinion based. What is invited is a series of grab-bag lists of unfiltered ideas instead, and it's hard to see how that's better except for ease of responding.

There are some cases where this can be an appropriate framing, but it's far from universal, and in the concrete case here I think it's outright worse than the original. If questions have been being closed constantly because they're posed in the form of seeking a recommendation and this is why they all now need to be phrased that way, I think we should just stop doing that.

A question asking about the implications of a concrete design choice should be able to get genuine expert analysis, instead of being forced into a box of mixed-value top-of-head reckons to avoid the appearance of impropriety. Good quality answers, even if contrary, can be voted up, and unsubstantiated ranting downvoted; that's the whole point.

$\endgroup$
4

3 Answers 3

3
$\begingroup$

Both "what are pros and cons" and "should" give the wrong signal. The first one makes it sound as if PL design is a search for a compromise, and the second as if PL design involves moral judgements.

I wonder how many questions here are motivated by actual problems that people face when designing & implementing PL. So far everyone seems to be stuck on the syntax.

$\endgroup$
6
  • $\begingroup$ The word "should", in technical contexts, does not imply moral judgements ─ it makes a statement normative, or makes a question ask for a normative answer, but the relevant norms need not be (and here, aren't) moral norms. See e.g. the use of the word "should" in RFCs, which is defined as meaning "there may exist valid reasons in particular circumstances to ignore a particular item, but the full implications must be understood and carefully weighed before choosing a different course." So a question like "should I do X?" should be interpreted as ... $\endgroup$
    – kaya3
    May 19, 2023 at 7:23
  • $\begingroup$ ... asking what those reasons and implications are, and possibly also asking how to weigh them. $\endgroup$
    – kaya3
    May 19, 2023 at 7:24
  • 1
    $\begingroup$ Syntax is a common problem people get stuck on. Also, most questions are about syntax currently because they are the easiest to write. One we are more mature, there will be a nice base of dupe targets we can use. $\endgroup$
    – Seggan
    May 19, 2023 at 12:39
  • $\begingroup$ @kaya3: You have a point. So long as people are corteus and the "shoulds" don't clash with any "musts", we should be fine. $\endgroup$ May 19, 2023 at 12:45
  • $\begingroup$ "I wonder how many questions here are motivated by actual problems that people face when designing & implementing PL." The very question that sparked this meta question, for example :p $\endgroup$
    – lyxal Mod
    May 20, 2023 at 11:08
  • $\begingroup$ Well, another motivation for asking questions is the desire to learn, so I think we needn't worry too much about why people aks question. However, knowing where they're coming from helps shape the answers. $\endgroup$ May 20, 2023 at 20:57
1
$\begingroup$

They make up a significant portion of our current questions and seem to be answered with facts rather than opinions. They are high quality questions that are attracting high quality answers. The cons other answers have been over

$\endgroup$
3
  • $\begingroup$ Was wondering about downvote reasoning.... vs. up or abstain: Are uncommentexplained-votes equally welcome in either direction? $\endgroup$ Jun 21, 2023 at 2:05
  • $\begingroup$ PS: Hi, betadogfoodtestkitchenchefs! Nice to meet you [or will be once we hemispheres of my brain and body get around to fill8ng out a profilevand reading previous beta content questions, faq'-age, pre-1st-Q-post politeness-lurking, etc. anyway.... Just arrived :/〕 $\endgroup$ Jun 21, 2023 at 2:13
  • $\begingroup$ FWIW, I UPVOTED because I got info I didn't know that was important to me as someone verrrry happy to find a stack withba great ("and"-specific longer) title like this one, to complement mathoverflow and proofassistan[ce](sic.) where I've felt most most happy (as an unregistered reader) previously. $\endgroup$ Jun 21, 2023 at 2:23
-2
$\begingroup$

My opinion: "should" is a better wording for these titles since it's more concise; a question title using the word "should" should be understood to mean "what are the reasons for or against this", "what considerations are there when choosing this", or so on, i.e. inviting an objective discussion. Likewise, for questions asking only about advantages or only about disadvantages, I propose "why should ..." or "why shouldn't ..." as concise titles which in the context of this Stack Exchange site should be understood as asking for objective reasons.

Wordier forms like "what are the advantages and disadvantages of..." belong in the question bodies, where there is more space for it. That said, if a question fails to include those magic words, I don't think that should be enough for it to be closed as opinion-based or lacking focus.

$\endgroup$
6
  • 1
    $\begingroup$ I feel like interpreting it as "what are the reasons for or against" is still too limiting (eliminates articulating an actual solution) while also too broad (at least on the surface it's requesting just a couple of unanalytic lists of anything you can think of). Analysis of trade-offs still fits within a (notionally) closed question, you just have to actually do it. I'm not sure exactly where the line should be. (but, not my downvote) $\endgroup$
    – Michael Homer Mod
    May 19, 2023 at 6:50
  • $\begingroup$ @MichaelHomer I'm not sure what you mean about "articulating an actual solution" ─ the questions are about how to make a design decision of whether or not a language has a certain feature, or whether or not a feature should have certain semantics. Pros and cons are what we consider in order to make such decisions; "should a language do this?" is the question which motivates wanting to know the pros and cons for it. If anything, the wording "should a language do this?" allows more for a definitive answer than "what are the pros and cons?" does, if a definitive answer is possible. $\endgroup$
    – kaya3
    May 19, 2023 at 7:06
  • $\begingroup$ A "definitive answer" with an articulated rationale can synthesise the trade-offs involved—perhaps a contingent answer, "X is good if A, otherwise Y"—and can be good (or not: that's what votes are for). A list of pros and cons doesn't provide insight into the weighting or interplay of those facets, and carries a latent assumption that everything relevant must be either a pro or a con, which just isn't true. Sometimes that is appropriate, but I don't want to try to squeeze everything into that model and I feel like it's slanted too far in that direction currently (in part, for easy answering?). $\endgroup$
    – Michael Homer Mod
    May 19, 2023 at 7:16
  • $\begingroup$ @MichaelHomer OK, but again, in that case the wording "should ..." still allows for that kind of answer more than "what are the pros and cons?" does. $\endgroup$
    – kaya3
    May 19, 2023 at 7:17
  • $\begingroup$ Yes, I'm happy with "should" as the phrasing, but I'm not so keen on having it be a shorthand for the same write-some-lists question. Perhaps I buried that in the first comment. $\endgroup$
    – Michael Homer Mod
    May 19, 2023 at 7:21
  • 1
    $\begingroup$ @MichaelHomer Ah, I see ─ that's fair enough. Then my response is that "should" actually makes for a better question because it invites answers which go beyond the "listicle" format, weighing the reasons to reach a conclusion where possible. $\endgroup$
    – kaya3
    May 19, 2023 at 7:26

You must log in to answer this question.

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