What's our (maybe updated) take on questions about (quantum) chemical software?

Question

From What topics can I ask about here?

Some kinds of questions aren't allowed here:

• Computational questions: If your question is purely about numerical methods you are using in a simulation/etc, it is probably more appropriate at Computational Science.

Recently (or maybe not) I noticed that we are getting a few more questions about the use of software packages likes Gaussian, gamess, etc. While I would like to avoid turning chem.se into a user help forum, I sometimes do appreciate these questions and have to admit that an answer is not always readily available. Obviously not all of them are good or even worth keeping.

I also know from personal experience that these questions will not get a lot of attention on CompSci.se. They will probably be just ignored there. I don't remember when and where I had the conversation with one of their mods, but the essentially they said that they would refer askers back to support pages for the respective software.

I spare you links to examples to avoid meta-effect, but if you need them, find me in chat and we'll talk about it.

From the above quote I would interpret that questions, which are only about the use of software are off-topic on chem.se

So in principle the discussion points I would like to raise are:

• Do we still consider questions about the use of chemical software off-topic?
• If yes, should we migrate questions of that case immediately to CompSci.se?
• If no, should we establish a few guidelines for such questions to prevent them from cluttering our site?

Obviously I already have an idea about this and if you read between the lines you might even guess it. However, I would like to leave everybody as unbiased as possible. Please weigh in with your opinion/proposal. If nobody has ideas similar to mine I'll provide an answer later, too.

Answer 1

I think that question about how use comp chem software should be on topic. For example, the question about how to set up the geometry in Gaussian can be compared on how to set up all the glass ware and whatever else reaction stuff in an actual lab. Those questions then would probably be on topic and receive an answer.

My PC is my lab and up to a certain point I am doing chemistry there. So, in my eyes those are chemistry problems that will not fit into any other SE better, as they are too specific and we have the expertise for many of those programs around here.

Answer 2

Prompted by Martin, and as a purely synthetic chemist which is potentially against the overall trend here. But my $0.01 (or £0.01 as may be more appropriate now).

I'd argue that there should be a distinction between conceptual questions and how-to questions. To give two examples on what I would consider to be opposite sides of the spectrum of on-topic-ness.

Q: I just installed Avogadro/Maestro/Gaussian (delete as appropriate) and can't work out how to import a set of .xyz coordinates

These type of questions do seem to get asked on occasion, and I think its quite right that they should be off topic. If you ignore the fact its using a piece of chemistry software, its basically just an IT support issue which should be addressed by the relevant company (or even just in chat). I have a feeling that these kind of questions are a self fulfilling prophecy- someone answers one of them which means google directs people to here when they have similar issues. I'd also make the argument that we wouldn't give people advice on how to increase bond length in ChemDraw or how best to import structures into Keynote.

Q: I'm working with Maestro and can't decide what basis set would be most appropriate for calculating the minimum energy conformer

This type of question also pops up on occasion, and I think is quite rightly on topic.

The question is conceptual and general, with answers that would be useful to a wide range of people, not just people using a particular version of software X since the question is, at the heart of it, asking about the choice of basis sets for calculations, which would be transferrable.

---

Ignoring all of that, and raising a much larger topic, I'd bring up the fact that Mathematics.SE and Mathematica have different sites.

As computational chemistry has gotten more advanced, it seems to me that the type of people doing it are no longer just chemists banished out of a lab, but actually a wide range of people encompassing people not only with chemistry degrees but also those with computing/mathematics/engineering degrees. Perhaps not now, but certainly in the near future I think its completely right that computational chemistry deserves to have its own stack exchange site.

Answer 3

This answer was converted from a comment on pH13's answer per Martin's request.

Good questions ask:

How do I go about setting up this computational experiment I'm interested in?

Bad questions ask:

I tried to make this input, why is it broken?

Or:

My computation had an error; what's wrong?

Questions of the first kind are okay for the site, I think, because they prompt exposition of how the software achieves the computational goal.

Those of the second and third types are typically mere proofreading/debugging that should be carried out either in chat, or off of Chem.SE entirely

Answer 4

It's been about a month since I posed the question and before going into action I'd like to summarise a bit and add my own thoughts.

---

First and foremost, questions relating to computational methods and the understanding thereof have been on-topic and they will always be. That in general is true for all conceptual questions pertaining to chemistry.
Since this is the case, this post has never been about these questions, but rather about the application of software pertaining to chemistry.

---

In general there seems to be more or less agreement that questions regarding the use of certain (esp. computational) chemistry software should be somewhat on topic.
I think Philipp said it best:

[...] the question about how to set up the geometry in Gaussian can be compared on how to set up all the glass ware and whatever else reaction stuff in an actual lab. Those questions then would probably be on topic and receive an answer.

I agree with this.
While not all questions about the experimental set-up of equipment are good, voting is pretty much taking care of putting the good into the pot, the bad into the crop.
As such, we need to detach what constitutes a good question from what constitutes a question that is on topic on this site. Recent advances on that matter are still open to discuss: A new policy of closure: November 2016.
However, the a policy should act as a safeguard to prevent certain types of questions to become too common. It should act as an instrument of quality control and should weed out unsuitable questions from the start. It is unlikely that a new user is reading our policies before posting, rather than orienting {her..him}self on existing questions. The homework policy did a reasonably well job for a long time, but there has been a long process of changing the whole thing. It can, however, serve as an example of how policy works, and why we need it.

---

There are a couple of thoughts that I find important, which need to be considered. Some aspects have been raised in the other answers and I already addressed them in the comments there. In case you didn't follow the discussion, here is a (hopefully) brief summary.

• The questions will come, whether or not you like it; we need to address how to deal with them.

  [...] I have a feeling that these kind of questions are a self fulfilling prophecy- someone answers one of them which means google directs people to here when they have similar issues. [...] (NotWoodward)

  That has already happened. That will always happen. No policy can protect you against this.
  Answering a couple of these questions is probably not the only way they get indexed by search engines.¹ As long as some of us keep answering questions with a few calculations, visual representations of molecular orbitals, structures, etc. this will lead new users to believe that these questions are okay here.

• Users of (computational) chemistry software probably have already an extensive background in chemistry.

  This type of question additionally is far more likely than an average chem.se question to come from active practioners of chemistry, which is something that, at the margins, we probably want to encourage. (Curt F.)

  This site is growing, it will be further growing. With a few ups and downs the site continues to grow. As a result it will hopefully attract more users that answer questions. In order to maintain the current quality of the questions and answers, it would be preferable to have some users with a firm background in chemistry. Anyone who starts to use software to solve problems probably already has a good understanding of chemistry or is on {her..his} way to obtain it. I agree with Curt that it would be a good idea to gain the interest of this user group.

• Computational chemistry as its on stack exchange site.

  [...] certainly in the near future I think its completely right that computational chemistry deserves to have its own stack exchange site. (NotWoodward)

  This is most likely not going to happen. Computational Science is older than chemistry.se, but it continues to have a quite low influx of questions. Proposing a site with overlapping interests to two existing stack exchange sites is probably not gaining the critical mass to become successful.
  Additionally, just because a question is on-topic on another site doesn't make it off-topic for another.

• There already are (help/ support) forums for software.

  However, program specific questions probably have better outlets. There are typically active forums, mailing lists, for most (all?) software; if it is commercial software the company typically supports it in some fashion. (brose)

  Indeed, they do exist. Are they better outlets? Not always.
  Most of those typically active forums are not really that active, not really that helpful, and almost always not canonical. They also have many garbage suggestions (that don't get weeded out), terrible formatting, long lines of discussion before you get to the answer, non-existing search features, and other smaller problems.
  The availability of these forums makes information very scattered, hard to find and often you have to concern yourself with many of these to get your solution. Even quite simple problems don't find an answer for weeks or even forever. Most of the times they then only offer answers for a specific problem without the necessary contextual, conceptual underlying issue. One thing that chemistry.se successfully accomplishes is going beyond the actual question, answering it in a much more concise way. Why cut out a large portion of users that could do the same for a more theoretical setting?

• We don't want want to become a debug/ proofreading service.

  Those of the second [why is my input broken] and third [why do I get this error] types are typically mere proofreading/debugging that should be carried out either in chat, or off of Chem.SE entirely. (hBy2Py)

  This certainly could be regarded as an issue, however it probably isn't. If there is an underlying conceptual problem with the input, it goes beyond proofreading or debugging, and might be a more common issue then you would think at first. These are very similar to amirite (Am I right?) questions, and can be treated analogously.
  Additionally, such questions will probably attract no audience, and regular voting will take care of them, or they will be covered by our new closing policy.

• It is not all about computational chemistry.
  While certainly the majority of software related questions, the use of programs to do certain tasks has increased dramatically in recent years. These questions have so far been positively received; at least generally better than organic chemistry questions.²
  The questions can be found in all branches of chemistry and apply to a broad scope of users. The critical mass is already here, and, so far, it sets good examples. By writing up some basic guidelines we can keep this track record and prevent the homework catastrophe.

• How to {compile, install, .., fix, debug} a chemistry program?
  This should be off-topic. There is a multitude of websites (including some on our own network) that already deal with that. There is no link to chemistry. Those questions should be closed and eventually removed or migrated.

---

Footnotes

1. We have already answered questions that are mainly use of software. The crack is already in the door. We would have to remove these questions, and it would still be uncertain if that would have any effect.
2. There are currently 63 tagged computational chemistry and software, the latter tag has 162 questions total. Only 24 of them are with a score < 1, that's 15%. For comparison organic chemistry has 769 of 3875 score < 1 questions, 20%, homework has 33% "bad" questions (858/2569 score < 1).

Answer 5

Do we still consider questions about the use of chemical software off-topic?

As a blanket statement, I think questions about the use of chemical software should be on-topic, so no.

If no, should we establish a few guidelines for such questions to prevent them from cluttering our site?

For this to work, we definitely need guidelines.

---

I wanted to take a small survey of the questions currently under the tag, so here is some sort of arbitrary categorization of the first 50 questions on the "Newest" page. I placed each question into more than one bin where appropriate, but this is completely my biased judgement. P and C are for "program" or "concept", with "?" where it could be disputed or I couldn't decide. Other things I would have done differently: lab software (lots of HPLC stuff) went under "math"/other, and I didn't create the visualization bin soon enough so that's an underestimation.

$$ \small \begin{array}{clc} \hline \textbf{Categorization} & \textbf{Topic} & \textbf{Count} \\ \hline \text{P} & \text{Amber} & 1 \\ \text{P} & \text{Gaussian} & 10 \\ \text{P} & \text{Open Babel} & 2 \\ \text{P?} & \text{PDB} & 2 \\ \text{P} & \text{Avogadro} & 6 \\ \text{C} & \text{Charges} & 1 \\ \text{C} & \text{Transition states} & 3 \\ \text{C} & \text{"Math"/other} & 9 \\ \text{?} & \text{Technical} & 14 \\ \text{P} & \text{TINKER} & 1 \\ \text{C} & \text{Method comparison} & 1 \\ \text{P} & \text{Other MD package} & 2 \\ \text{C} & \text{Comp. chem. concepts} & 5 \\ \text{P} & \text{What package(s)? (survey)} & 10 \\ \text{P} & \text{GAMESS} & 1 \\ \text{?} & \text{Data source(s)} & 3 \\ \text{C?} & \text{Programming} & 4 \\ \text{P?} & \text{Cheminformatics} & 3 \\ \text{P} & \text{Visualization} & 5 \\ \text{C} & \text{Reference request} & 1 \\ \hline \end{array} $$

The "technical" topic is the one we're most interested in here. To draw an analogy, it's the homework of software. Maybe we should create software-homework? I jest, but it's the practicioner's equivalent of homework. It bothers me that so many of the questions revolve around basic usage of a certain program named after a kind of mathematical function, and I wonder if these get a free pass due to it being that program, so any question surrounding it must be complex. There should be more scientific merit to some of these questions. The same goes for Avogadro, most of which are not about visualization techniques but are more user support.

I'd like to bring up Philipp's answer (by no means to pick on it):

I think that question about how use comp chem software should be on topic. For example, the question about how to set up the geometry in Gaussian can be compared on how to set up all the glass ware and whatever else reaction stuff in an actual lab. Those questions then would probably be on topic and receive an answer.

I disagree here, for a few reasons. At least, I think this deserves elaboration.

• I was under the impression that "best practices" questions were off-topic. At least, it isn't clear to me from our tour page. "Experimental techniques and technology" are ok, but would it be acceptable for me to ask about tips on running a chromatography column (test solvent system with a pipette column, TLC every x minutes, etc.)? In my opinion, those are things every synthetic chemist should know. They're also opinion-based to a certain degree. Why does this bar seem to be lower for computational chemistry? People probably wouldn't like it if I went around saying "people doing calculations shouldn't use Windows" as a best practice. (Yes, this is pretty extreme.) Do we want to accept questions on best practices?

• "Is my input correct?" is dangerous. If it's something like "help me construct my z-matrix", maybe there should be a canonical question with appropriate internet and literature references, but definitely nothing beyond that. Where does something like "What's the difference between calling b3lyp and td-b3lyp in Gaussian" fall? One might view it as practicioner's homework, another might view it as a chance to make an important conceptual point.

My PC is my lab and up to a certain point I am doing chemistry there. So, in my eyes those are chemistry problems that will not fit into any other SE better, as they are too specific and we have the expertise for many of those programs around here.

I am in general agreement with this point, but there needs to be some minimum barrier to entry, otherwise the tag will become even more of a mess than it is.

Perhaps something to help us decide what's on-topic is...

Some initial tagging guidelines?

...tags, tags, tags. These questions need more tags.

• Only software is bad. Only one tag in general is bad.
• Only the software tag plus one of computational-chemistry, quantum-chemistry, theoretical-chemistry, or cheminformatics is an indicator that the question might not have enough scientific focus and is too broad, or it's conceptual in nature. These need to be policed.

My thought is that if you cannot envision a question have more than just the software tag, it isn't a good fit for the site. I say this having done so twice out of my four questions asked! One question could have more tags, and the other's quality is debatable.

Summary

Yes, these should be on-topic, but the bar needs to be raised.

I am more focused on questions like "How do I set up a geometry optimisation in Gaussian?" in this post. Strictly speaking they are not chemistry questions and have often been viewed as off-topic here. The question is, whether this should be changed and under what guidelines.

That shouldn't be allowed. I have had good success in luring these people into chat, though.

I realize I'm kind of mixing topics here (best practices vs. "homework questions") and deviating from the question but it's easier for me to air everything at once.

Side notes for later discussion

Other things I discovered in doing research for this answer which aren't pertinent to my answer but don't deserve their own questions quite yet:

• Transition state questions seem to be common, both the conceptual and technical aspects. Should there be transition-state for both experiment and computation? This is not the same thing as what's currently under transition-state-theory, which might be subsumed under the new tag.

• I think all instances of "Gaussian 09" or "G09" should be changed to "Gaussian". The version being used is inconsequential in all the cases we have, and no one is using other versions.

• Someone created nwchem and orca. Do we want to create more program-specific tags, like gaussian or avogadro? In this specific case, these tags have 1 Q and don't make sense to exist since they aren't even NWChem- or ORCA-specific.

Answer 6

Since I am unable to comment...

General questions relating to methods and understanding seem great. However, program specific questions probably have better outlets. There are typically active forums, mailing lists, for most (all?) software; if it is commercial software the company typically supports it in some fashion. A list could fairly easily be put together of "orphan" software, programs with no support outlet, if any exists. If the program specific questions are here, they remove some of the questions that others users can easily search on the 'official' places where they are more likely to go.