6
$\begingroup$

Let me just start by saying that I don't have any strong opinion on this, and won't be voting yes or no to any proposal in this direction. I am just interested to see what the community thinks, and I'm sure the other members of the community will be able to come to consensus despite me remaining indifferent.

This site was very different when it first went live in 2018, and in recent times, tags that barely existed back then have quickly caught up and even overtaken the most popular tags across the whole site:

There have actually been more questions asked than the number of total question on some entire SE sites that have been around for several years:

That tag is also on pace to soon overtake several other sites in total questions, such as:

  • Beer, Wine and Spirits SE,
  • Language Learning SE,
  • Coffee SE,
  • Stellar SE,
  • Ebooks SE,
  • Tezos SE,
  • Korean Language SE,
  • Arts & Crafts SE,
  • Mythology & Folklore SE,
  • Sustainable Living SE,
  • Freelancing SE,
  • Internet of Things SE,
  • Poker SE,
  • Martial Arts SE,
  • Ukranian Language SE,
  • Portuguese Language SE,
  • EOS.IO SE
  • Stack Apps SE,
  • Lifehacks SE,
  • Mathematics Educators SE,
  • Windows Phone SE, and
  • probably plenty of others

All of those above sites are averaging far fewer than 400 questions/year which is significantly less than the tag has been getting lately, and the latter is growing.

The questions about the theory/mathematics of quantum information processing are significantly different from the questions we're getting on "quantum programming", the latter tending to often look like:

I have heard opinions about this dichotomy from some other users, and I've also seen people closing (many) questions due to being too much about programming, so I wanted to see what the wider community's feeling is about this.

There exists a StackOverflow for programming questions, which is very different from Computer Science SE and even more so, Theoretical Computer Science SE.

Would it make sense for there to be a "Quantum Stack Overflow" for quantum programming questions, versus a "Quantum Information Science" which is more like Computer Science SE and Theoretical Computer Science SE?

$\endgroup$
1
  • 2
    $\begingroup$ I'm also interested to see what the community thinks. Right now, my gut says splitting the site does not make sense, but I'll have to put some thought into it. Thanks for bringing it up! $\endgroup$
    – Auden Young Mod
    Aug 22 at 1:32
6
$\begingroup$

I agree that there is a weird dichotomy on this site, but I don't see a good solution for it.

On the one hand, if I try to imagine how a proposal for a new, say, qiskit StackExchange site would go, I don't see it going well. It would be hard to convince people that a specific python package/framework should warrant a dedicated site. If they ask you "why not just ask these questions on StackOverflow?", what's the answer? If it is that using quantum computing frameworks such as qiskit requires domain-specific knowledge about quantum physics and related areas, the obvious follow-up would be: "ok... then why not ask them on quantumcomputing.SE"? I wouldn't be personally against such a site, and I definitely see the rationale and usefulness for it, but it might face an uphill battle as an Area51 proposal. Then again, arguably this site did as well, so maybe some sponsorship kind of deal might make it happen, I've got no idea.

At the same time, the dichotomy on this site is very real. As you say, "quantum programming" questions are significantly different from more "sciency" ones. I think this is natural: what makes a programming question good is quite different than what makes a more "math/physics/science-based" question good. The former are often more centered around getting a recipe, a snippet of code that can be readily copy-pasted somewhere else. This is not the case at all for more "academic" types of questions, in which the focus is generally on understanding some underlying concept.

But there are reasons to ask some "quantum programming" questions on a site like this. In particular, these questions involve domain-specific knowledge about quantum mechanics and related areas, which are not the purview of most programmers. The problem, as I see it, is that it gets hard to reliably recognise the distinction between "pure programming questions", which should really be asked on StackOverflow, and programming questions for which some knowledge of quantum computing is actually useful, which might be better served here. And even if we know how to tell these two things apart, new users will inevitably get endlessly confused as to why some of their questions are fine here while others are not.

This makes things... weird. We effectively get a site that lives a double life. And these two aspects of the site not only accommodate different questions, but also follow different standards of moderation. Questions that would get closed when viewed from the "academic" point of view might be fine if understood as programming questions.

So what's the solution? I don't know. This is a problem ultimately caused by the inherently multidisciplinary nature of the subject at hand. I feel like the best thing to do from this perspective is to filter out more aggressively programming questions, making sure that "pure programming questions" get sent elsewhere. But this would need to come from the people that most closely follow those questions, and so whether it happens depends on whether they agree that questions should be filtered this way.

$\endgroup$
13
  • 3
    $\begingroup$ While I agree with the point that quanutm programming questions requiring quantum computing knowledge to be answered are on point on this site, most of the qiskit questions I've answered did not require such a knowledge. I'd even argue that most of these questions are purely programming ones, like being confused by Qiskit's little-endian convention or why doesn't the code work. Most of the time, the reason why the code works is a misunderstanding on how to use Qiskit rather than a misunderstanding of quantum computing. $\endgroup$ Aug 27 at 13:00
  • 2
    $\begingroup$ That said, I think it'd make sense to redirect questions that are focused on code to StackOverflow. The thing is I wonder whether it is too late to make such a change: most questions about qiskit on StackOverflow are unanswered (I only took a quick look, that may be incorrect) while the IBM Quantum Support team is particularly active on this site. Won't it be confusing to users to suddenly reject all qiskit-focused questions? $\endgroup$ Aug 27 at 13:04
  • 1
    $\begingroup$ @TristanNemoz that's pretty much my impression as well. It would probably be confusing for some people, yes, but I don't think it's too late to do it, if we want to. But again, that would need to be done by those most active in these questions. The fact that less of these questions get answered on stackoverflow might simply because those that know how to answer them hang out here more than there. That could easily change if questions are redirected there. $\endgroup$
    – glS Mod
    Aug 27 at 13:16
  • 3
    $\begingroup$ One thing that might help is for the qiskit, ibm etc websites to explicitly tell people to go on stackoverflow for purely programming questions, and here for questions about quantum computing, rather than redirecting everyone here as they do now $\endgroup$
    – glS Mod
    Aug 27 at 13:17
  • $\begingroup$ I agree that pure programming questions belong on SO. We would probably see more such questions rerouted there if a suitable close reason were added to the existing "belongs on QC meta". $\endgroup$ Aug 28 at 3:28
  • 1
    $\begingroup$ We could also update the programming tag with some form of a criterion for when a programming question belongs on SO and when on QCSE. A possible criterion: "answer requires knowledge of quantum theory". $\endgroup$ Aug 28 at 3:29
  • $\begingroup$ @AdamZalcman good points. I was actually convinced the excerpt of programming already said something to that effect, but clearly it doesn't. Adding the routing option to SO is also doable, and was brought up before (though I'd note that it is doable even withotu the explicit option, you can just flag it for mod attention; albeit this sidesteps the close-vote queue so it's not ideal) $\endgroup$
    – glS Mod
    Aug 28 at 10:03
  • 2
    $\begingroup$ My perspective as a long-time consumer of the StackExchange sites is that their value is in the decade old answer still being exactly what I happen to be looking for. Personally I find it a bit odd to see framework-specific questions on this site or even StackOverflow, since the answers to these questions (especially in qiskit and IBMQ) is subject to rapid change. I think the qiskit slack page is a better place to find help. It is admittedly less well archived, but I'm not sure archiving most of these sorts of questions is terribly useful. $\endgroup$
    – jecado
    Aug 31 at 1:27
  • 1
    $\begingroup$ @jecado I definitely agree with that: Slack, GitHub issues, or dedicated forums (like other frameworks such as pytorch and tensorflow have) are sometimes better venues. But then again, it's not my place to tell people not to ask questions on StackOverflow, and much of SO has the problem of questions getting obsolete when the underlying software changes anyway; it's not a new phenomenon, and the site has been doing fine regardless. $\endgroup$
    – glS Mod
    Aug 31 at 7:14
  • $\begingroup$ "doing fine" is not how I and many other would describe it. $\endgroup$ Aug 31 at 19:20
  • $\begingroup$ @user1271772 what are you referring to exactly? I'm saying "doing fine" in the sense that it is still the #1 place everyone goes to get answers to programming questions, and there is hardly any competitor in that space, that I know of. Or in other words, I don't know of any site employing a better paradigm of programming Q&A. Personally, I still easily find posts about old and new frameworks/software, so I don't see the problem of software Q&A getting obsolete as that bad... or at least not yet $\endgroup$
    – glS Mod
    Aug 31 at 20:05
  • $\begingroup$ I agree that SO is still the #1 place for Q/A but this very well may be for historical reasons, and they are not necessarily "doing fine". I'm not sure SO is the best place for Qiskit questions, and the data shows us that Qiskit questions don't get answered there much. Some programming frameworks/languages/software/etc. are better off with their own site, as many do already have! $\endgroup$ Aug 31 at 20:29
  • $\begingroup$ Most of your answer seems to be based on a premise which I believe is false: which is that it would be hard for a "Quantum Programming" site to get through Area51 (i.e. to convince SE to launch a site like that into Beta), in the event that we decide that those questions are off-topic here and that this site should be "Quantum Information Science" or something like that. $\endgroup$ Sep 1 at 15:43

You must log in to answer this question.

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