16
$\begingroup$

On Chemistry.SE we have gotten a question, which is about to be closed as off-topic: Band structure of cell with doping in CP2K (or via the Internet Archive).

I understand that the topics involved there are certainly on topic (as they are on Chemistry). I am more concerned by the type of question, which is basically asking about how to operate the program.

Does this community want these questions to be on topic, why or why not?

$\endgroup$
16
$\begingroup$

It was questions like that one that I had in mind when I first made the proposal.

My answer to your question during Area51 alluded to the fact that VASP, MOLCAS, CFOUR, GAUSSIAN, LAMMPS, CP2K, DIRAC, Turbomole, MOLPRO, ORCA, MRCC, ADF, PySCF, PSI4, DALTON, QuantumEspresso, QChem, CASTEP, CPMD, ABINT, DFTB+, ABAQUS, OpenBabel, may all have their own individual forums or Q/A sites.

Rather than giving my email address and making a password for each and every forum, it would be nice to have one unified place where people can ask a question with a "tag" that reaches the desired audience, and have it potentially seen by 30x the number of people.

Area51 was not an easy process to get through, and people wondered why myself and Rashid, and Cody, and others worked so hard at it: it's because we believe in Stack Exchange. Questions and answers are available for anyone online to benefit from, so that they don't have to bang their head against the wall for the same problem that someone else already solved. The search function is excellent. The automated search for similar questions, helps askers to avoid asking a question that was already asked. The upvoting helps reward users for providing high-quality answers, and motivates (some) people to put real effort into their questions.

To register for the VASP forum, you need to be given a registration code, which is only given to people that have paid the $6000 fee to keep their license valid. So where do I go if I (not yet a VASP user, but interested in whether or not VASP might be able to solve my problems) want to ask a question about VASP? That aforementioned Area51 answer shows that there's 1000+ VASP questions on ResearchGate and 2 questions/week on the VASP facebook group (back then, much more now). However, primarily VASP users follow those groups. The answer you are really looking for, might only come from a PySCF user, because VASP users will be very unlikely to know about the brand new cutting edge features of PySCF such as coupled-cluster for solid-state materials with periodic boundary conditions using plane-wave basis sets (a rather new area, since until recently coupled-cluster was way too expensive to do on solid state materials). But how does the user know that the PySCF forum is the right place to ask? It would be impossible to expect them to know which of the 200 software communities to reach out to.

However we have built a community here with major representatives that are lead developers or major developers from many of the most important software packages, committed during Area51, here's their Area51 profiles:

Many of them were unfamiliar with Area51, and a little bit put off by the UFOs and aliens and overall "look and feel" of Area51 (which I admit has been that way for about a decade), so they sent me emails asking to clarify about the confusion. I told them to hang tight, and the only reason we don't yet see some of them here is because I don't have confirmation yet about our chances of being approved for Public Beta. When we do, we will have these lead software developers come roaring back and in many cases a big community of followers will come with them.

$\endgroup$
12
  • 2
    $\begingroup$ All that text means yes, right? $\endgroup$ May 1 '20 at 17:08
  • 2
    $\begingroup$ Yes :) I was also about to write you another message in the chat, apologizing for writing such a long answer. I didn't have time to make it short, yet. I saw the question and wanted to get these thoughts conveyed, and I do plan to come back to this when I have more time (some co-authors want a Table in our paper changed, so I'm working hard to fetching the data for that right now). $\endgroup$ May 1 '20 at 17:29
  • $\begingroup$ I think the site is a good idea, but I'm less sure that basic questions like "how do I run program X" are a good fit for Stack Exchange sites. I think they're better handled in a research group, docs for that program, and/or support for that program. Otherwise, there may be some issue/bug that I (as a developer) only hear third-hand because someone eventually sends me the link to the Q here. $\endgroup$ May 2 '20 at 0:48
  • 3
    $\begingroup$ @GeoffHutchison thanks for the feedback! I see your point about developers not seeing the questions. We got commitments in A51 from developers of quite a large number of programs (for example the lead developer of MOLPRO), so maybe they will see the question, especially when they get the email if they are following the "molpro" tag. On the other hand, there's a question I posted on the Psi4 forum 2 years ago about compiling with a bigger basis sets option, that I know very well Lori Burns can answer but she's too busy. Here I'd put a 500 point bounty and I'm sure I'll get the answer! $\endgroup$ May 2 '20 at 0:55
  • 1
    $\begingroup$ Also, it just occurred to me that as a developer, I don't want to be re-explaining the same solution dozens of times. If someone asks about my software here and I answer it, anyone else that has the same question does not have to send me another email if they see the solution here :) $\endgroup$ May 2 '20 at 0:58
  • $\begingroup$ Some software-specific questions are going to be fine - but I think the community needs to come up with some guidelines like the "homework" one on Chem.SE. I'd be careful about basic 'how do i run program x' questions. $\endgroup$ May 2 '20 at 0:59
  • 2
    $\begingroup$ I also appreciate your feedback! I alone have been personally responsible for a lot of really stupid questions to devs (including @GeoffHutchison haha) Therefore such a place for software questions could help weed out more basic questions. $\endgroup$
    – Cody Aldaz
    May 2 '20 at 1:00
  • $\begingroup$ As for e-mail, I strongly suggest a boilerplate "please post in the forum so other users can see my answer" message. Happy to forward one. $\endgroup$ May 2 '20 at 1:00
  • 3
    $\begingroup$ @GeoffHutchison Thank you for the advice: It will be good to remember to make a guideline like the homework one, if we start getting too many software questions that we don't like. For now everything has been manageable, at least for me :) About the boilerplate message on the Forum: that is true, but a lot of people might not necessarily want to join the forum, whereas they quite likely may already have an SE account due to questions about TeX or programming (SO) or Linux, etc. $\endgroup$ May 2 '20 at 1:03
  • 1
    $\begingroup$ I have had to deal with (basic) software support (like) questions for quite a while now, and they wear you out. It is often the most basic things that the asker is unwilling to do, e.g. adding some input to reproduce the problem. I've also answered a few of the better ones, where the OP was engaging. Many of these questions get closed (and/or down-voted) on Chemistry, because they simply are too specific for a problem and won't help many other people. Also, these questions involve a lot of back.-and-forth, something the SE platform is not really equipped to deal with. $\endgroup$ May 2 '20 at 10:27
  • $\begingroup$ If you could make your post more concise, so that (especially newer) users can see a clear message of whether their question is considered on topic here, I'll accept it. I personally don't find it at all beneficial to publish the names of celebrities, who might or might not contribute. $\endgroup$ May 4 '20 at 12:16
  • $\begingroup$ I wanted to reply today, but ended up spending a lot longer on my answer to the tags question, than I thought I would. So perhaps I will address this one tomorrow :) $\endgroup$ May 7 '20 at 7:00
8
$\begingroup$

Here is a more concise answer: I personally would like to welcome such software questions. They would still have to conform. If we get questions where the answer is easily found in the software's manual online, the question could still be closed for lack of effort/research, but if the question does not have an answer online that's found with reasonable effort, I think people should be able to ask here.

Where else will they be able to ask those questions? In some cases they can send an email to the developers of the software (or the support team if it's a larger scale software that has a support team), but what if the developers are retired, no longer alive, too busy, or have moved on to other areas of research?

This is a great place for people to help each other for questions about molecular and materials modeling software.

$\endgroup$
4
  • 2
    $\begingroup$ I understand what you are trying to say, but quantifiers such as 'easily' and 'reasonably' are problematic, because most of us have a completely different understanding of these than any of the new(er) users. On chemistry, about 9/10 questions dealing with software are of low quality type and we have no real handle on them. You should develop rigorous and easy to follow guidelines (for your core user base) on which questions to accept. From my experience with the homework policy on chemistry, I can also say that effort is a really bad metric, possibly the worst. $\endgroup$ May 11 '20 at 15:53
  • $\begingroup$ @Martin-マーチン Thank you again Martin for the very valuable advice. I will think of some more "rigorous" criteria to help guide people as to when and when not to close a software question. $\endgroup$ May 11 '20 at 16:13
  • 2
    $\begingroup$ I would suggest anything along the line that the problem should be completely reproducible: e.g. include the program and the used version, input code example, complete error description (if applicable), enough context to understand the goal (see XY problem). I'm sure you can think of others, but that may get the ball rolling. I'd also suggest discussing in a separate meta post to develop these guidelines, then writing these guidelines so that they can be used as a close reason. $\endgroup$ May 11 '20 at 17:21
  • $\begingroup$ These are great ideas @Martin-マーチン !! I agree with all of them! $\endgroup$ May 11 '20 at 17:22
-5
$\begingroup$

I personally don't think basic operation of a program, i.e., how do I generate input for program X and run it are great topics for Stack Exchange sites.

I think there are several better ways of answering those questions: - In a particular research group (i.e., expertise at running frequently-used programs) - In the documentation for that program (i.e., tutorials) - In the user forum for that program

Basic questions on simple operations of a program are often from students just starting out - in which case they may need help on a range of common tasks. Even when they're not it's better in my opinion to point to the documentation or a program-specific forum.

1) Good documentation should be encouraged, and every program should have some obvious tutorials.

2) Program developers need to know if there are problems for starting users (e.g., bugs on Windows or with a specific Linux distribution, etc.)

I think there's great merit to the forum overall. The question Martin linked from Chem.SE seems like a useful one here, but this one here seems like it's better suited for documentation / tutorials for that program.

$\endgroup$
9
  • 5
    $\begingroup$ Point 1) There is good documentation for a lot of software but sites like Stack Overflow exist. How is the question there different than when I go onto stack overflow looking how to make a folder with python? Point 2) yes and such questions could be exposed here with less hassle to the devs. Through comments and up/downvoting it could be determined whether more specific advice is needed. $\endgroup$
    – Cody Aldaz
    May 2 '20 at 1:11
  • $\begingroup$ If you go to Stack Overflow and ask how to run Python or how to do basic programming tasks, your questions will also be closed. (Trust me, I've had lots of SO questions closed.) Your specific example is notable because it's not 'how do I do something basic' but more 'how does X work' $\endgroup$ May 2 '20 at 1:17
  • 3
    $\begingroup$ "Good documentation should be encouraged, and every program should have some obvious tutorials." I agree with this, but how do we enforce it? The relativistic quantum chemistry software BAGEL had no documentation at all for almost 2 years, but I got most of my problems sorted out by friends who had more experience with it than me. $\endgroup$ May 2 '20 at 1:17
  • 2
    $\begingroup$ I used this question today stackoverflow.com/questions/123198/… it's not closed as far as I can tell. $\endgroup$
    – Cody Aldaz
    May 2 '20 at 1:20
  • $\begingroup$ @CodyAldaz - I think my point stands. Maybe you see 'copying a file' as a basic programming task. I don't, because it's not part of the basic language. Stack Overflow very much separates what questions to ask - basic tutorials are not part of that site. $\endgroup$ May 2 '20 at 1:34
  • 1
    $\begingroup$ More on-topic, I'm simply advocating that this site needs to decide what 'basic tutorial' questions might be, and to define what is and is not useful. I made my opinion here - others are welcome to weigh in. That's the point of meta. $\endgroup$ May 2 '20 at 1:35
  • 1
    $\begingroup$ I agree with you. Questions, even the one I linked, like these tend to attract a lot of one-on-one conversations, that means lots of comments. The possibly relevant information is buried deep down somewhere, not obvious to find (so it's as bad as a forum, or worse), or the posts themselves require a lot of maintenance. Users seeking a quick answer will hardly ever come back to make the post helpful for future generations. Defining what to keep or close is something this SE must do, because everything you keep is basically an encouragement of what you want. $\endgroup$ May 2 '20 at 10:12
  • 1
    $\begingroup$ @CodyAldaz The 'open' status of a question is not necessarily a criterion on SO. As far as I know, they have not been able to clear the close review queue for years now, which is clogged up with >1k questions. The question you cite is different though, basic enough, but at the time (>10 years ago) possibly without many external resources. $\endgroup$ May 2 '20 at 10:16
  • $\begingroup$ Our software has (IMHO) good documentation and it has lots of tutorials, and we still get lots of simplistic questions tagged on SO that are already answered elsewhere. I may be frustrated by many of them, but I still feel obliged to answer them. I agree that, for troubleshooting, SE is a terrible format, but many users don't understand that and seem reluctant to migrate to better fora like our mailing list or github issues. What can you do? $\endgroup$
    – jeguyer
    May 20 '20 at 21:10

You must log in to answer this question.

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